ci: trigger mergeability recalculation

chore(release): prepare v1.0.55-beta.4

.github/CODEOWNERS

@@ -1,4 +0,0 @@
-# Default code owners for all files
-# These users will be automatically requested for review on PRs.
-
-* @DingTalk-Real-AI/cli-maintainers

.github/PULL_REQUEST_TEMPLATE.md

@@ -3,22 +3,41 @@
 - What changed?
 - Why is this change needed?
 
+## Risk tier
+
+- [ ] Documentation-only: prose/assets only; no executable, generated, workflow,
+  packaging, or interface behavior changed
+- [ ] Standard: ordinary implementation change with a stable package graph
+- [ ] High-risk: workflow/policy, package graph, generated Schema/registry,
+  platform, auth/keychain, installer, packaging, release, transport, recovery,
+  or another fail-closed infrastructure change
+
 ## Verification
 
-For an exact in-place `CHANGELOG.md`-only pull request, the full-suite checks
-may be marked `N/A`, but the targeted CHANGELOG check is required. For every
-other pull request, mark the targeted check `N/A` and complete the applicable
-full-suite checks.
+Record the smallest targeted evidence that proves the changed behavior. Do not
+repeat the entire CI suite locally only to fill this checklist: CI expands the
+selected tier from documentation checks, through affected-package tests, to
+the complete high-risk suite.
 
-- [ ] Exact `CHANGELOG.md`-only check (otherwise `N/A`):
+- [ ] Exact in-place `CHANGELOG.md`-only check (otherwise `N/A`):
   `./scripts/policy/check-changelog-pr.sh --fast-path "$(git merge-base HEAD origin/main)" HEAD`
-- [ ] `make build`
-- [ ] `make lint`
-- [ ] `make test`
-- [ ] `make policy`
+- [ ] Targeted test/check commands and results:
+- [ ] Behavior evidence (test name, CLI output shape, or before/after result):
+- [ ] Documentation links/content/rendering checked (documentation-only, otherwise
+  `N/A`)
+- [ ] Full local suite run because the change is high-risk (optional for other
+  tiers; record command/result or `N/A`)
 - [ ] `./scripts/policy/check-generated-drift.sh`
+  (when generator inputs or generated artifacts may change)
 - [ ] `./scripts/policy/check-command-surface.sh --strict` (if command surface changed)
+- [ ] `./scripts/release/verify-package-managers.sh`
+  (after `make package`, if packaging or installer surfaces changed)
 
 ## Notes
 
 - Any risks, follow-up work, or intentional scope cuts
+
+The repository automatically requests one eligible peer reviewer, including
+after a new head push when another review is needed. Once the latest push has
+peer approval and all nine required checks are current and green, auto-merge
+completes the PR; authors do not need to coordinate a separate routine merge.

.github/workflows/mirror-to-gitee.yml

@@ -14,6 +14,11 @@ on:
   schedule:
     - cron: '0 18 * * *'
   workflow_dispatch:
+    inputs:
+      sync_release_version:
+        description: "Sync a specific release version's assets to Gitee (e.g. v1.0.55-beta.3)"
+        required: false
+        type: string
 
 concurrency:
   group: gitee-code-mirror
@@ -23,7 +28,6 @@ jobs:
   mirror:
     runs-on: ubuntu-latest
     if: ${{ github.ref_name == github.event.repository.default_branch && github.repository_owner == 'DingTalk-Real-AI' }}
-    # GitHub Actions 不允许在 job-level if 直接引用 secrets，故先用 env 暴露再在 step 守卫。
     env:
       GITEE_TOKEN: ${{ secrets.GITEE_TOKEN }}
       GITEE_USER: ${{ secrets.GITEE_USER }}
@@ -76,3 +80,42 @@ jobs:
           # main 镜像对齐；release tag 由 release.yml 单独校验后创建，禁止在这里 force。
           git push --force "$REMOTE" 'gitee-main:refs/heads/main'
           echo "✅ 已镜像 main（含 Gitee README 本地化）到 Gitee ${GITEE_REPO}"
+
+      - name: Download GitHub Release assets
+        if: ${{ inputs.sync_release_version != '' }}
+        env:
+          VERSION: ${{ inputs.sync_release_version }}
+          GH_TOKEN: ${{ github.token }}
+        run: |
+          set -eu
+          echo "📥 Downloading release assets for ${VERSION}"
+          mkdir -p dist
+          gh release download "$VERSION" \
+            --repo "$GITHUB_REPOSITORY" \
+            --dir dist \
+            --pattern 'dws-*' \
+            --pattern 'checksums.txt' \
+            --clobber
+          ls -la dist/
+
+      - name: Verify release artifacts
+        if: ${{ inputs.sync_release_version != '' }}
+        env:
+          VERSION: ${{ inputs.sync_release_version }}
+        run: |
+          set -eu
+          DWS_PACKAGE_DIST_DIR="$GITHUB_WORKSPACE/dist" \
+            ./scripts/release/verify-release-artifacts.sh "$VERSION"
+
+      - name: Sync release assets to Gitee
+        if: ${{ inputs.sync_release_version != '' }}
+        env:
+          VERSION: ${{ inputs.sync_release_version }}
+          GITEE_TOKEN: ${{ secrets.GITEE_TOKEN }}
+          GITEE_USER: ${{ secrets.GITEE_USER }}
+          GITEE_REPO: ${{ secrets.GITEE_REPO }}
+          DIST_DIR: ${{ github.workspace }}/dist
+        run: |
+          set -eu
+          echo "📦 Syncing release assets for ${VERSION} to Gitee ${GITEE_REPO}"
+          ./scripts/release/sync-to-gitee.sh

.github/workflows/reviewer-router.yml

@@ -0,0 +1,258 @@
+name: Reviewer routing
+
+on:
+  pull_request_target:
+    branches: [main]
+    types: [opened, synchronize, reopened, ready_for_review]
+
+# pull_request_target deliberately runs only this workflow from the protected
+# base branch. Never check out or execute pull-request code here.
+permissions:
+  contents: write
+  pull-requests: write
+
+concurrency:
+  group: reviewer-router-${{ github.event.pull_request.number }}
+  cancel-in-progress: true
+
+jobs:
+  route:
+    if: github.event.pull_request.draft == false
+    runs-on: ubuntu-latest
+    timeout-minutes: 5
+    steps:
+      - name: Route review and enable auto-merge
+        uses: actions/github-script@f28e40c7f34bde8b3046d885e986cb6290c5673b # v7.1.0
+        with:
+          script: |
+            const owner = context.repo.owner;
+            const repo = context.repo.repo;
+            const pullNumber = context.payload.pull_request.number;
+            const eventHeadSha = context.payload.pull_request.head.sha;
+            const reviewerPool = [
+              'sczheng189',
+              'shangguanxuan633-lab',
+              'audanye-sudo',
+              'wxianfeng',
+            ];
+
+            async function getReadyEventPull(phase) {
+              const {data: currentPull} = await github.rest.pulls.get({
+                owner,
+                repo,
+                pull_number: pullNumber,
+              });
+              if (
+                currentPull.head.sha !== eventHeadSha ||
+                currentPull.state !== 'open' ||
+                currentPull.draft ||
+                currentPull.base.ref !== 'main'
+              ) {
+                core.info(
+                  `PR #${pullNumber} state or revision no longer matches this ready-main event during ${phase}; routing stopped.`,
+                );
+                return null;
+              }
+              return currentPull;
+            }
+            const pullRequest = await getReadyEventPull('initial read');
+            if (!pullRequest) {
+              return;
+            }
+            const author = pullRequest.user.login.toLowerCase();
+            const headSha = pullRequest.head.sha;
+            const latestPusher =
+              context.payload.action === 'synchronize'
+                ? context.payload.sender?.login?.toLowerCase()
+                : author;
+
+            async function routeReview() {
+              const eligible = reviewerPool.filter(
+                reviewer =>
+                  reviewer.toLowerCase() !== author &&
+                  reviewer.toLowerCase() !== latestPusher,
+              );
+              if (eligible.length === 0) {
+                core.warning(`No eligible reviewer remains for PR #${pullNumber}.`);
+                return;
+              }
+
+              const alreadyRequested =
+                (pullRequest.requested_reviewers || []).length > 0 ||
+                (pullRequest.requested_teams || []).length > 0;
+              if (alreadyRequested) {
+                core.info(`PR #${pullNumber} already has a requested reviewer; leaving it unchanged.`);
+                return;
+              }
+
+              let reviews;
+              try {
+                reviews = await github.paginate(github.rest.pulls.listReviews, {
+                  owner,
+                  repo,
+                  pull_number: pullNumber,
+                  per_page: 100,
+                });
+              } catch (error) {
+                core.warning(
+                  `Could not inspect existing reviews for PR #${pullNumber}; skipping reviewer routing to avoid a duplicate request (${error.status || 'unknown status'}).`,
+                );
+                return;
+              }
+
+              const latestDecisionByLogin = new Map();
+              for (const review of reviews) {
+                const login = review.user?.login?.toLowerCase();
+                if (
+                  !login ||
+                  !['APPROVED', 'CHANGES_REQUESTED', 'DISMISSED'].includes(
+                    review.state,
+                  )
+                ) {
+                  continue;
+                }
+                const previous = latestDecisionByLogin.get(login);
+                if (!previous || review.id > previous.id) {
+                  latestDecisionByLogin.set(login, review);
+                }
+              }
+              const currentHeadDecision = [...latestDecisionByLogin.values()].find(
+                review =>
+                  review.commit_id === headSha &&
+                  eligible.some(
+                    reviewer =>
+                      reviewer.toLowerCase() ===
+                      review.user.login.toLowerCase(),
+                  ) &&
+                  ['APPROVED', 'CHANGES_REQUESTED'].includes(review.state),
+              );
+              if (currentHeadDecision) {
+                core.info(
+                  `PR #${pullNumber} already has a ${currentHeadDecision.state} review on its current head; leaving review ownership unchanged.`,
+                );
+                return;
+              }
+
+              const loads = new Map(eligible.map(reviewer => [reviewer, 0]));
+              try {
+                const openPullRequests = await github.paginate(github.rest.pulls.list, {
+                  owner,
+                  repo,
+                  state: 'open',
+                  per_page: 100,
+                });
+                for (const openPullRequest of openPullRequests) {
+                  for (const reviewer of openPullRequest.requested_reviewers || []) {
+                    const candidate = eligible.find(
+                      login => login.toLowerCase() === reviewer.login.toLowerCase(),
+                    );
+                    if (candidate) {
+                      loads.set(candidate, loads.get(candidate) + 1);
+                    }
+                  }
+                }
+              } catch (error) {
+                core.warning(
+                  `Could not read current reviewer load; using deterministic rotation (${error.status || 'unknown status'}).`,
+                );
+              }
+
+              const offset = pullNumber % eligible.length;
+              const rotated = eligible.slice(offset).concat(eligible.slice(0, offset));
+              const tieOrder = new Map(rotated.map((reviewer, index) => [reviewer, index]));
+              const staleChangeRequester = [...latestDecisionByLogin.values()]
+                .filter(review => review.state === 'CHANGES_REQUESTED')
+                .sort((left, right) => right.id - left.id)
+                .map(review =>
+                  eligible.find(
+                    reviewer =>
+                      reviewer.toLowerCase() === review.user.login.toLowerCase(),
+                  ),
+                )
+                .find(Boolean);
+              const ranked = [...eligible].sort(
+                (left, right) =>
+                  Number(right === staleChangeRequester) -
+                    Number(left === staleChangeRequester) ||
+                  loads.get(left) - loads.get(right) ||
+                  tieOrder.get(left) - tieOrder.get(right),
+              );
+
+              for (const reviewer of ranked) {
+                try {
+                  const currentPull = await getReadyEventPull('review request');
+                  if (!currentPull) {
+                    return;
+                  }
+                  if (
+                    (currentPull.requested_reviewers || []).length > 0 ||
+                    (currentPull.requested_teams || []).length > 0
+                  ) {
+                    core.info(
+                      `PR #${pullNumber} received a reviewer while routing; leaving it unchanged.`,
+                    );
+                    return;
+                  }
+                  await github.rest.pulls.requestReviewers({
+                    owner,
+                    repo,
+                    pull_number: pullNumber,
+                    reviewers: [reviewer],
+                  });
+                  core.info(
+                    `Requested @${reviewer} for PR #${pullNumber} (open request load: ${loads.get(reviewer)}).`,
+                  );
+                  return;
+                } catch (error) {
+                  core.warning(
+                    `Could not request @${reviewer} for PR #${pullNumber}; trying the next candidate (${error.status || 'unknown status'}).`,
+                  );
+                }
+              }
+
+              core.warning(`No reviewer request could be created for PR #${pullNumber}.`);
+            }
+
+            async function enableAutoMerge() {
+              try {
+                const currentPull = await getReadyEventPull('auto-merge enable');
+                if (!currentPull) {
+                  return;
+                }
+                if (currentPull.auto_merge) {
+                  core.info(`Auto-merge is already enabled for PR #${pullNumber}.`);
+                  return;
+                }
+                await github.graphql(
+                  `mutation EnableAutoMerge($pullRequestId: ID!) {
+                    enablePullRequestAutoMerge(
+                      input: {
+                        pullRequestId: $pullRequestId
+                        mergeMethod: MERGE
+                      }
+                    ) {
+                      pullRequest {
+                        autoMergeRequest {
+                          enabledAt
+                        }
+                      }
+                    }
+                  }`,
+                  {pullRequestId: currentPull.node_id},
+                );
+                core.info(`Enabled native auto-merge for PR #${pullNumber}.`);
+              } catch (error) {
+                core.warning(
+                  `Could not enable auto-merge for PR #${pullNumber}; checks and review can continue normally (${error.message}).`,
+                );
+              }
+            }
+
+            try {
+              await routeReview();
+            } catch (error) {
+              core.warning(
+                `Reviewer routing hit an unexpected error for PR #${pullNumber}; review can still proceed manually (${error.message}).`,
+              );
+            }
+            await enableAutoMerge();

.gitignore

@@ -62,3 +62,7 @@ dwsbin
 /docs/shortcut-comparison.html
 /docs/shortcut-gsb-eval.*
 /scripts/run_shortcut_real_read_matrix.py
+
+# Local coverage artifacts
+coverage-shortcut.txt
+coverage-*.txt

CHANGELOG.md

@@ -6,6 +6,73 @@ The format is inspired by [Keep a Changelog](https://keepachangelog.com/) and th
 
 ## [Unreleased]
 
+### Added
+
+- **Shortcut Runtime Schema delivery** — publishes all 210 public built-in shortcuts as reviewed Agent-visible leaf tools across 16 product groups, with stable canonical identities, executable `+shortcut` CLI paths, parameter and cross-parameter constraints, selection guidance, interface metadata, and runtime-aligned safety/confirmation semantics. `dws shortcut list` remains the lightweight batch-discovery view, while leaf Schema now carries the complete Agent contract; declared string-slice defaults are also preserved consistently in Cobra and Schema.
+
+## [1.0.55-beta.4] - 2026-07-27
+
+This beta validates the shortcut projection fixes for group bots, bot search,
+and mail threads, together with hardened release delivery to Gitee and npm on
+top of the `v1.0.55-beta.3` baseline.
+
+### Fixed
+
+- **Shortcut projection fixes** (#795) — `chat +chat-bots` no longer projects a non-empty `list_group_bots` response to an empty list, `+bot-find` recognizes the `search_bots` response shape (`result.bots` entries with `botOpenDingTalkId`), and mail thread listings keep `lastUpdated` when the backend returns `lastModifiedDateTime`.
+
+### Changed
+
+- **Hardened release delivery** — the Gitee mirror workflow can synchronize a specific release's assets on demand, release lookup tolerates Gitee's HTTP 200 null-body response for missing releases, npm dist-tag verification waits through slow registry CDN propagation with incremental backoff, and beta/stable release operations are role-enforced (#791).
+
+## [1.0.55-beta.3] - 2026-07-24
+
+This beta validates the HR Brain command surface, smoother guarded release
+automation, and deterministic Markdown test coverage on top of the
+`v1.0.55-beta.2` baseline.
+
+### Added
+
+- **HR Brain (`dws hrbrain`) command surface** — adds 11 commands across three groups: `talent-pool list/detail/employees` for talent pool browsing, `profile metadata/query/labels/career/performance` for employee profile data, and `search employees/employees-structured/fields` for basic and advanced (rule-based) people search. Ships with bundled mono/multi Skill guidance (`dingtalk-hrbrain`, `cli_version: ">=1.0.54"`); `search employees-structured` validates `--origin-json` as a JSON object and `--fields` as a JSON array before dispatch.
+
+### Changed
+
+- **Smoother guarded releases** — publishes verified stable and beta Homebrew Formula updates directly from the release workflow, retries transient tag-ref visibility failures, lets an exact same-run retry reuse its sealed tag, and allows machine-verified rebuild recovery without a separate approval wait.
+
+### Fixed
+
+- **Deterministic Markdown coverage** — replaces timing-dependent temporary-file deletion tests with synchronized file-stat failures so release admission no longer flakes on scheduler timing.
+
+### Changed
+
+- **Faster guarded releases** — trusts an independently revalidated, exact `CHANGELOG.md`-only successor of an already admitted `main` commit, runs cloud planning alongside governance, and executes sealed-release automation, compatibility, and multi-profile validation in parallel with artifact compilation. Normal cloud publication no longer requires an unshareable local packaging preflight.
+- **Scoped document reads and group mentions** — `doc read --content-format jsonml` can return `outline`, `range`, `section`, or custom-tag fragments with depth and block-boundary controls; document comment create, reply, and update can mention groups through `--mentioned-open-conversation-id`.
+- **Drive overwrite uploads** — `drive upload --node <fileId>` can replace an existing Drive or document-space file, is mutually exclusive with `--folder`, supports dry-run, and requires confirmation before writing.
+- **Chat nickname clearing and cross-organization todos** — omitting `--nick` from `chat group update-nick` now clears the current user's group nickname, while `todo task list --query-all` queries todos across organizations.
+
+### Fixed
+
+- **Legacy authentication compatibility** (#756) — migrates pre-v1.0.53 global and organization-scoped login state into the identity-aware token store, including all legacy organizations, while keeping unresolved accounts isolated from exact `corpId:userId` credentials so external or no-directory identities can complete login without borrowing another user's token.
+
+## [1.0.55-beta.1] - 2026-07-23
+
+This beta validates MCP Market URL resolution, the supported Wukong local-file
+send path after retiring the legacy credential-based media upload command from
+discovery, and reliable message-read rendering for rich content, forwarded
+records, encrypted messages, and media-download ID aliases.
+
+### Added
+
+- **MCP URL resolution** — adds `dws mcp url get <mcpId>` for resolving a DingTalk MCP Market ID to the current user and organization scoped Streamable HTTP URL, while keeping the helper-only `mcp-meta` endpoint out of the public product command surface.
+
+### Changed
+
+- **Chat local-file sending** — hides the open-source-only `chat media upload` compatibility command from Help, Schema, and bundled Skills, and removes its legacy AppKey/AppSecret OAPI path. Historical argv still receives an actionable migration error. Send local images and files through `chat message send --msg-type file --file-path`; callers that already hold a mediaId may continue to use `--msg-type image --media-id`.
+
+### Fixed
+
+- **Shortcut projection silent-empty returns** (#783) — a batch of read shortcuts returned an empty list with exit 0 and no error envelope even when the underlying MCP tool returned data, so agents misread "no data". The projection resolvers now probe the real container keys (`processCodeList`, `values`, `wikiSpaces`, `itemList`, `groupList`, `recentItems`, `emailAccounts`, `deptUserList`, `labelUserList`, `roles`, `report_list`, and the grouped `get_org_labels` `labels[]`), unwrap items nested under a VO wrapper (`shiftVO` / `entityVO` / `userInfo`), and `todo +created-todos` uses the shared pager (`pageSize=20`) because the backend silently returns an empty page for `pageSize>20`. Affects contact/oa/wiki/drive/minutes/calendar/attendance/chat/report/smart shortcuts, each with a guard test asserting the real response shape projects non-empty. `scripts/shortcut_real_result.py` also gains an upper-vs-lower layer comparison so an exit-0 empty projection over a non-empty backend is scored as `projection-data-loss` in the real read-audit path rather than `real-ok`.
+- **Message-read shortcut projection** (#706) — the message-list shortcuts (`chat +chat-messages` / `+messages-list` / `+messages-list-direct` / `+at-me` / `+search-msg` / `+thread-replies`) now render card and out-of-office rich-content JSON as readable text (without ever rewriting ordinary text that merely embeds a JSON fragment), expand a forwarded chat record's nested `forwardMessages` instead of collapsing to a "[卡片]" summary, and mark undecryptable encrypted card messages as `[加密消息]`; the speaker is read from the bare `sender` key, nested `{name:…}` sender objects yield their display name, and the literal string `"null"` is treated as absent. Shared projection helpers now live in `internal/shortcut/chatmsg`. `chat message download-media` also gains `--msg-id` / `--open-message-id` aliases for its `--message-id` flag so agents copying the `openMessageId`/`msgId` output field no longer hit "unknown flag".
+
 ## [1.0.54] - 2026-07-21
 
 This release promotes the validated `v1.0.54-beta.2` baseline to stable. It restores the default transport envelope for personal event output with opt-in flattening, plus Schema CLI path and plugin overlay compatibility fixes.

CONTRIBUTING.md

@@ -27,7 +27,9 @@ notes that are intentionally kept out of the repository root.
 
 ## Local Checks
 
-Run the verification commands that match the surface you changed before you hand work back.
+Run the verification commands that match the surface you changed before you
+hand work back. The goal is useful, change-specific evidence, not a second
+local execution of every CI job.
 
 Common repository checks already used here include:
 
@@ -44,19 +46,55 @@ make lint
 git diff --check
 ```
 
+Select the PR risk tier before choosing checks:
+
+| Tier | Typical scope | Developer evidence | CI expansion |
+|---|---|---|---|
+| Documentation-only | Prose and documentation assets with no executable, generated, workflow, packaging, or interface change | Links/content/rendering plus repository asset checks | Lightweight documentation validation; all nine named contexts still report |
+| Standard | Ordinary implementation work with a stable package graph | Focused unit/integration tests and observable behavior for the changed path | Race tests for changed packages and their reverse dependencies, scope-matched HEAD/base coverage, and representative Darwin/Windows compilation |
+| High-risk | Workflow/policy, package graph, generated Schema/registry, platform, auth/keychain, installer, packaging, release, transport, recovery, or an unprovable infrastructure change | Relevant full or domain suite plus focused behavior evidence | Complete race suite, native platform tests, and all affected domain gates; protected `main` uses this tier |
+
+Classification fails closed: an incomplete diff, package add/remove/rename, or
+uncertain dependency graph selects the high-risk suite. Native changed-code
+coverage is additionally selected for platform-sensitive code.
+
 ## Pull Request Checklist
 
 1. Keep implementation and tests in sync.
-2. Run `./scripts/dev/ci-local.sh`.
-3. Run `./scripts/policy/check-command-surface.sh --strict` when command paths/flags change. CI also runs `./scripts/policy/check-command-compatibility.sh --base-ref <main-ref> --stable-ref <latest-GA-tag>` against both the target branch and latest stable release.
-4. Run `./scripts/policy/check-generated-drift.sh` when generated artifacts may change.
-5. Run `./scripts/release/verify-package-managers.sh` when packaging or installer surfaces change (run `make package` first).
-6. Update docs and `CHANGELOG.md` for behavior/interface changes.
-7. Include verification evidence in your PR description.
+2. Select the documentation-only, standard, or high-risk tier and run the
+   smallest checks that prove the change. Use `./scripts/dev/ci-local.sh` when
+   a complete local pass is warranted; it is not required for every ordinary
+   PR.
+3. Include both the commands/results and user-visible or contract-level
+   behavior evidence in the PR description.
+4. Run `./scripts/policy/check-command-surface.sh --strict` when command
+   paths/flags change. CI also runs
+   `./scripts/policy/check-command-compatibility.sh --base-ref <main-ref> --stable-ref <latest-GA-tag>`
+   against both the target branch and latest stable release.
+5. Run `./scripts/policy/check-generated-drift.sh` when generated artifacts may
+   change.
+6. Run `./scripts/release/verify-package-managers.sh` when packaging or
+   installer surfaces change (run `make package` first).
+7. Update docs and `CHANGELOG.md` for behavior/interface changes.
 
 ## Submission Flow
 
 1. Make the smallest atomic change that satisfies the task.
 2. Keep doc edits factual and limited to implemented behavior.
 3. Run the relevant verification commands.
-4. Report the validation results with the handoff.
+4. Report the validation results and risk tier with the handoff.
+5. Open a ready PR against `main`. Base-owned automation assigns one eligible
+   peer reviewer, balancing the current open-review load and excluding the
+   author. A new head push re-enters the same routing flow when the latest
+   revision still needs review.
+6. After the latest push has one peer approval and the exact nine required
+   contexts are current and green, auto-merge completes the PR. If `main`
+   advances first, strict status checks revalidate the branch; no separate
+   routine merge request is needed.
+
+Contributors without repository write access stop at the PR flow. Explicitly
+authorized collaborators with `write`, `maintain`, or `admin` access can use
+[Actions → Release](https://github.com/DingTalk-Real-AI/dingtalk-workspace-cli/actions/workflows/release.yml)
+to publish beta releases without manual approval. The same internal roles may
+start a stable release, but a different repository administrator must approve
+the `release-stable` Environment deployment before publication continues.

Formula/dingtalk-workspace-cli-beta.rb

@@ -1,33 +1,33 @@
 class DingtalkWorkspaceCliBeta < Formula
   desc "Automate DingTalk workspace tasks from the terminal (beta channel)"
   homepage "https://github.com/DingTalk-Real-AI/dingtalk-workspace-cli"
-  version "1.0.54-beta.2"
+  version "1.0.55-beta.4"
   license "Apache-2.0"
   keg_only "it is the beta channel and conflicts with dingtalk-workspace-cli"
 
   on_macos do
     if Hardware::CPU.arm?
-      url "https://github.com/DingTalk-Real-AI/dingtalk-workspace-cli/releases/download/v1.0.54-beta.2/dws-darwin-arm64.tar.gz"
-      sha256 "46b57bed1f6e9f7ba007d8a86a6f5eb280fdeb557fc9bb5946f14f9b1f8f0c9f"
+      url "https://github.com/DingTalk-Real-AI/dingtalk-workspace-cli/releases/download/v1.0.55-beta.4/dws-darwin-arm64.tar.gz"
+      sha256 "05b269fe44a125ee8b368d6228c5229950b8216872fb569741d1e30a83ce952a"
     else
-      url "https://github.com/DingTalk-Real-AI/dingtalk-workspace-cli/releases/download/v1.0.54-beta.2/dws-darwin-amd64.tar.gz"
-      sha256 "1b7fd08e64b1c86bbcee217604ffe07e0e8f1b3b5c4de518534386972bcf0f9b"
+      url "https://github.com/DingTalk-Real-AI/dingtalk-workspace-cli/releases/download/v1.0.55-beta.4/dws-darwin-amd64.tar.gz"
+      sha256 "b0d7604299336c83b7805d3b1a47a90668f2e2f3fc54702b0e846bb2907b6170"
     end
   end
 
   on_linux do
     if Hardware::CPU.arm?
-      url "https://github.com/DingTalk-Real-AI/dingtalk-workspace-cli/releases/download/v1.0.54-beta.2/dws-linux-arm64.tar.gz"
-      sha256 "108d3861ef606519f9934530d29654eab55a73607d1ee6775461f98ef5a6acd4"
+      url "https://github.com/DingTalk-Real-AI/dingtalk-workspace-cli/releases/download/v1.0.55-beta.4/dws-linux-arm64.tar.gz"
+      sha256 "a9c1dd5c6171091a84fc18e5081c9f75d037826cd3966726545d715ce832ae31"
     else
-      url "https://github.com/DingTalk-Real-AI/dingtalk-workspace-cli/releases/download/v1.0.54-beta.2/dws-linux-amd64.tar.gz"
-      sha256 "6cb96ee09419bbbcc1eb336218ac2aa1d9ca0ed5cbd5a80c79bc20e1e1f03ff7"
+      url "https://github.com/DingTalk-Real-AI/dingtalk-workspace-cli/releases/download/v1.0.55-beta.4/dws-linux-amd64.tar.gz"
+      sha256 "5e97ba398f5a3e15b9d235bc53f596d31a4d6b7af7bb2185b7b48beeda6a2ebb"
     end
   end
 
   resource "skills" do
-    url "https://github.com/DingTalk-Real-AI/dingtalk-workspace-cli/releases/download/v1.0.54-beta.2/dws-skills.zip"
-    sha256 "572b93f04a10268d185ad1f8e70e0d412949ae056be8494a9949387076fd14bc"
+    url "https://github.com/DingTalk-Real-AI/dingtalk-workspace-cli/releases/download/v1.0.55-beta.4/dws-skills.zip"
+    sha256 "4ebc0294b65d90adb5c5d639a548b528af240e4117ccca3d388e2efb6030170a"
   end
 
   def install

Makefile

@@ -8,7 +8,7 @@ POLICY_GOTMPDIR ?= $(DWS_POLICY_TMPDIR)/go
 POLICY_ENV = DWS_POLICY_TMPDIR="$(DWS_POLICY_TMPDIR)" GOTMPDIR="$(POLICY_GOTMPDIR)"
 GO_SOURCE_LIST = git ls-files -z --cached --others --exclude-standard -- '*.go'
 
-.PHONY: all help build rebuild test test-plan lint format-check fmt policy edition-test interface-integrity authoritative-interface-integrity coverage-gate coverage-gate-platform update-interface-baseline reset-interface-baseline schema-compatibility skill-command-integrity cli-smoke mock-mcp-smoke test-schema-agent-examples generate-schema generate-schema-agent-metadata generate-schema-catalog package release release-pre release-stable changelog-pre changelog-stable publish-homebrew-formula setup-hooks
+.PHONY: all help build rebuild test test-plan test-auth-legacy-compat lint format-check fmt policy edition-test interface-integrity authoritative-interface-integrity coverage-gate coverage-gate-platform update-interface-baseline reset-interface-baseline schema-compatibility skill-command-integrity cli-smoke mock-mcp-smoke test-schema-agent-examples generate-schema generate-schema-agent-metadata generate-schema-catalog package release release-pre release-stable changelog-pre changelog-stable publish-homebrew-formula setup-hooks
 
 all: setup-hooks fmt lint build test rebuild
 
@@ -17,6 +17,7 @@ help:
 	@printf "  make build         - Build the dws CLI binary\n"
 	@printf "  make test          - Run the Go test suite\n"
 	@printf "  make test-plan     - Verify every default Go package belongs to one CI test shard\n"
+	@printf "  make test-auth-legacy-compat - Run stable legacy authentication compatibility regressions\n"
 	@printf "  make lint          - Run formatting checks, go vet, and staticcheck\n"
 	@printf "  make format-check  - Check all repository Go source files with gofmt\n"
 	@printf "  make fmt           - Format all repository Go source files\n"
@@ -38,8 +39,8 @@ help:
 	@printf "  make package       - Build all release artifacts locally\n"
 	@printf "  make changelog-pre VERSION=vX.Y.Z-beta.N - Prepare prerelease notes\n"
 	@printf "  make changelog-stable VERSION=vX.Y.Z FROM_BETA=vX.Y.Z-beta.N - Prepare stable notes\n"
-	@printf "  make release-pre VERSION=vX.Y.Z-beta.N [PUBLISH=1] - Validate or publish prerelease\n"
-	@printf "  make release-stable VERSION=vX.Y.Z FROM_BETA=vX.Y.Z-beta.N [PUBLISH=1] - Validate or publish stable\n"
+	@printf "  make release-pre VERSION=vX.Y.Z-beta.N - Validate prerelease; publish official releases from Actions\n"
+	@printf "  make release-stable VERSION=vX.Y.Z FROM_BETA=vX.Y.Z-beta.N - Validate stable; publish official releases from Actions\n"
 	@printf "  make publish-homebrew-formula - Push dist/homebrew/dingtalk-workspace-cli.rb to a tap repo\n"
 
 build:
@@ -54,6 +55,10 @@ test:
 test-plan:
 	@./scripts/ci/test-packages.sh verify
 
+test-auth-legacy-compat:
+	@mkdir -p "$(POLICY_GOTMPDIR)"
+	@GO="$(GO)" $(POLICY_ENV) ./scripts/policy/check-auth-legacy-compat.sh
+
 lint:
 	@./scripts/dev/lint.sh
 
@@ -76,7 +81,7 @@ fmt:
 	$(GO_SOURCE_LIST) > "$$go_files"; \
 	xargs -0 sh -c 'if [ "$$#" -gt 0 ]; then exec gofmt -w -- "$$@"; fi' sh < "$$go_files"
 
-policy:
+policy: test-auth-legacy-compat
 	@mkdir -p "$(POLICY_GOTMPDIR)"
 	@$(POLICY_ENV) ./scripts/policy/check-open-source-assets.sh
 	@$(POLICY_ENV) ./scripts/policy/check-schema-command-registry.sh

docs/architecture.md

@@ -52,7 +52,13 @@ run.
 
 ```mermaid
 flowchart TB
-  PR["Pull request"] --> CA["CI"]
+  PR["Pull request"] --> CLASSIFY["Fail-closed risk classification"]
+  CLASSIFY --> DOCS["Documentation-only<br/>asset/content validation"]
+  CLASSIFY --> STANDARD["Standard<br/>affected + reverse-dependent race<br/>scope-matched HEAD/base coverage"]
+  CLASSIFY --> HIGH["High-risk / main<br/>full race + native tests"]
+  DOCS --> CA["CI"]
+  STANDARD --> CA
+  HIGH --> CA
   subgraph CA_CHECKS["Nine required contexts"]
     L["Lint"]
     T["Test"]
@@ -72,9 +78,16 @@ flowchart TB
   PLATFORM --> RELEASE
 ```
 
-Complete Multi-profile E2E and the ordinary full native-platform matrix are
-downstream of PR admission. PRs still run primary-environment assurance and
-fast cross-platform compilation; auth, keychain, OS-specific, installer, and
-release changes additionally select native platform tests before merge. See
-[`docs/ci-pr-gates.md`](ci-pr-gates.md) for the exact context and ruleset
-contract.
+All nine named contexts are produced for every tier. Domain-specific helpers
+run when their owned surface is affected; otherwise the corresponding context
+records an explicit unaffected success. Standard code changes still receive
+representative Darwin/Windows compilation. High-risk PRs and protected `main`
+run the complete race and native test suites, while platform-sensitive diffs
+also receive native changed-code coverage.
+
+Review orchestration is also base-owned: it requests one eligible peer without
+executing PR code, re-routes an updated head when needed, and auto-merge
+completes only after the latest push has peer approval plus the current
+revision's nine strict contexts. Complete Multi-profile E2E remains downstream
+of PR admission. See [`docs/ci-pr-gates.md`](ci-pr-gates.md) for the exact
+classification, context, reviewer, and ruleset contract.

docs/automation.md

@@ -62,32 +62,27 @@ make lint
 git diff --check
 ```
 
-## Homebrew Formula PR Automation
+## Homebrew Formula Delivery
 
-Official tag releases require the repository Actions secret
-`HOMEBREW_PR_TOKEN`. Prefer a fine-grained personal access token owned by a
-maintainer or release-bot account, limited to this repository with
-`Contents: write` and `Pull requests: write`. If organization policy prevents
-that account from targeting the repository, use a dedicated classic token with
-only the `public_repo` scope. Do not reuse a broad developer token.
+Official releases use the Release workflow's built-in `GITHUB_TOKEN` to update
+exactly one tracked Formula after the immutable GitHub assets and their
+checksums have passed verification. The publisher validates the rendered Ruby,
+commits only the configured Formula path, never force-pushes `main`, and retries
+from a fresh clone up to three times when `main` advances concurrently. Normal
+stable and beta releases do not create a Formula PR or run a permission
+canary. The workflow uses the existing repository-scoped
+`HOMEBREW_PR_TOKEN` release identity because GitHub does not allow its built-in
+Actions App to bypass this repository's rulesets. That identity is the sole
+user bypass actor on the two default-branch rulesets. The workflow creates the
+nine Code Admission checks for the Formula-only commit only after proving its
+sole parent already has all nine successful checks and the committed Formula
+exactly matches this release's verified bytes.
 
-Store the dedicated token as the `HOMEBREW_PR_TOKEN` repository Actions secret
-and rotate it before its configured expiration. Replace it immediately if it is
-exposed, its owner loses repository access, or the release-bot ownership
-changes. The Release workflow uses this
-dedicated token only to push an `automation/homebrew-*` branch and open the
-stable or beta Formula PR. It does not push Formula changes directly to `main`.
-The default-branch governance preflight and every tag contract authenticate the
-token before publication, reject over-scoped classic tokens, confirm its
-identity, and run a controlled write canary. The canary pushes a unique
-`automation/homebrew-token-canary-*` branch with a `[skip ci]` commit, creates a
-draft PR, closes it, and deletes the branch with the same token. This proves both
-Contents and Pull requests write access before publication without merging
-anything. The gate also rejects reuse of `RELEASE_GOVERNANCE_TOKEN`.
-No maintainer environment variable is required when creating a tag. Using the
-built-in `GITHUB_TOKEN` is insufficient because organization policy prevents
-Actions from creating pull requests, and its generated PR events may require
-separate workflow approval.
+Keep `HOMEBREW_PR_TOKEN` repository-scoped with `Contents: write` and
+`Pull requests: write` (the latter remains necessary for withdrawal rollback),
+keep its owner as the designated ruleset bypass actor, and do not reuse
+`RELEASE_GOVERNANCE_TOKEN`. The workflow and publisher provide the Formula-only
+path restriction; GitHub rulesets do not infer that restriction from the token.
 
 ## Release Governance and Recovery
 
@@ -98,15 +93,14 @@ administration setting and cannot be read by the workflow's built-in
 contract use this same credential so a missing or expired identity is detected
 before an irreversible tag is created.
 
-Create a protected `release-recovery` environment limited to protected
-branches, with a required reviewer, self-review disabled, and administrator
-bypass disabled. The workflow reads the environment through the GitHub API and
-fails closed unless the required-reviewer, prevent-self-review, and protected-
-branch rules are present.
 Recovery is restricted to an existing annotated tag whose exact tag object,
-commit, and failed tag-push run all match; it then reuses the normal release
-jobs. Do not put publication secrets in temporary branches or create ad-hoc
-recovery workflows.
+commit, sealed metadata, original failed run/attempt, requester identity and
+Release state all match; it then reuses the normal release jobs without a
+second-person environment approval. A same-run “Re-run failed jobs” is even
+lighter: the seal job may adopt an existing tag only when its complete
+authority matches that run and its original attempt is not newer than the
+current attempt. Do not put publication secrets in temporary branches or
+create ad-hoc recovery workflows.
 
 Cloud-sealed releases mirror to OSS only when the repository variable
 `ENABLE_OSS_MIRROR` is exactly `true`. Leave the variable unset while no Bucket

docs/ci-pr-gates.md

@@ -4,9 +4,9 @@ The pull-request admission layer has exactly nine required external contexts:
 
 | Required context | Contract |
 |---|---|
-| `Lint` | Stable PR revision classification, formatting, `go vet`, and Actionlint |
-| `Test` | Race/unit/release-script tests plus fast cross-platform compilation |
-| `Coverage` | Overall non-regression and 100% changed-code coverage |
+| `Lint` | Stable PR revision/risk classification plus applicable formatting, `go vet`, and Actionlint |
+| `Test` | Tier-selected race/unit/release-script tests plus representative cross-platform compilation |
+| `Coverage` | Scope-matched overall non-regression and 100% changed-code coverage |
 | `Policy` | Repository policy and the fail-closed CHANGELOG contract |
 | `Edition` | Edition contract tests |
 | `Interface Integrity` | CLI, Schema, Skill, and stable-release compatibility |
@@ -56,8 +56,27 @@ base notes into an invalid final CHANGELOG.
 
 All nine admission contexts are still emitted and must succeed. Expensive
 implementation helpers are skipped; the named contexts record that their code
-surface is unaffected. After merge, the protected `main` push executes the
-full admission suite.
+surface is unaffected.
+
+The protected `main` push keeps that fast path only when all of these
+fail-closed conditions hold:
+
+- the event is a non-forced update of the existing `refs/heads/main`;
+- the event `after` SHA is the exact workflow SHA, and both event SHAs are
+  complete, non-zero commit IDs;
+- GitHub's comparison reports the previous main tip as the unique linear merge
+  base, with no commits behind it;
+- the complete resulting tree diff is exactly one in-place modification of
+  `CHANGELOG.md`;
+- the previous main tip already has successful GitHub Actions checks for all
+  nine Code Admission contexts.
+
+`Policy` then independently checks out the pushed revision and runs the same
+`check-changelog-pr.sh --fast-path` contract from the event's `before` SHA to
+its `after` SHA. If identity, ancestry, file scope, tree mode, CHANGELOG
+content, or predecessor admission cannot be proved, classification falls back
+to the complete main admission suite. A source change can therefore never
+inherit the CHANGELOG-only result.
 
 Any PR that touches `CHANGELOG.md` but also changes another file runs the same
 content contract in `Policy` with `--content-only`. That mode permits the
@@ -65,13 +84,28 @@ second file but still rejects invalid dates or versions, missing bullets,
 placeholder `TODO`/`TBD`, unmanaged-section changes, and unsafe tree modes.
 Adding a second file therefore cannot bypass CHANGELOG validation.
 
-## Platform and downstream boundaries
+## Risk tiers and downstream boundaries
 
-Ordinary PRs run the primary Linux assurance plus fast Darwin/Windows compile
-checks. Full native macOS/Windows tests and platform coverage run on a PR only
-when its diff touches auth, keychain, OS-specific Go files, installers,
-packaging, Formulae, or release automation. Protected `main` pushes run the
-complete native matrix.
+`Lint` resolves the complete base/head diff before any helper is skipped.
+Unknown or truncated input fails closed into the high-risk tier.
+
+| Tier | Selection | Admission work |
+|---|---|---|
+| Documentation-only | Only prose/documentation assets; no executable, generated, workflow, packaging, or interface surface | Documentation and repository-asset validation; expensive code helpers skip while every required context still succeeds |
+| Standard | Ordinary code change with a stable package graph | Race tests for changed Go packages and their reverse dependencies; candidate and merge-base coverage over the same impacted scope and `coverpkg`; representative Darwin/Windows compilation |
+| High-risk / protected `main` | Workflow/policy, package add/remove/rename, generated Schema/registry, platform, auth/keychain, installer, packaging, release, transport, recovery, or an unprovable infrastructure classification | Complete race suite and full native macOS/Windows tests, plus every affected domain gate |
+
+Domain helpers (`Edition`, `Interface Integrity`, `CLI Smoke`, and `Mock MCP`,
+for example) execute their substantive suites when the diff can affect that
+contract or when the high-risk tier is selected. Otherwise their stable named
+contexts still report a successful, explicit unaffected result. Release-script
+tests follow the same impact rule. This preserves the ruleset contract without
+charging every developer for unrelated work.
+
+Platform-sensitive changes additionally run native changed-code coverage.
+Protected `main` always runs native tests; generic portable changes are held to
+the Linux changed-code gate rather than being forced to manufacture
+platform-only coverage.
 
 Complete `Multi-profile E2E` is not a PR admission context. It belongs to the
 `Main Integration — 主干集成` workflow and runs only after a push to `main` (or
@@ -97,9 +131,28 @@ flowchart TB
   MAIN --> RELEASE["Release delivery"]
 ```
 
+## Review ownership and auto-merge
+
+A base-owned `pull_request_target` workflow routes newly opened, updated,
+reopened, or newly ready PRs targeting `main` to one eligible peer reviewer. It
+does not check out or execute PR code, excludes both the author and the known
+latest pusher, and balances the open requested-review load across the reviewed
+maintainer pool. A current-head approval or change request is preserved; after
+a new push, stale activity does not suppress a fresh request, and an
+outstanding change requester is preferred for continuity.
+
+The branch ruleset keeps one human approval and all nine strict required
+contexts, and requires someone other than the latest pusher to approve after
+the most recent head update. Repository auto-merge is enabled for ready PRs,
+so a PR merges after that approval and the current revision's nine checks are
+green. If `main` advances, strict checks rerun before merge. The reviewer
+router is orchestration, not a quality context, and must not be added to the
+ruleset.
+
 ## Running focused gates locally
 
-Run the contracts relevant to the change:
+Run the contracts relevant to the change. Ordinary contributors are not
+expected to repeat every CI job locally:
 
 ```sh
 make build
@@ -120,15 +173,18 @@ base_ref=$(git merge-base HEAD origin/main)
 ./scripts/policy/check-changelog-pr.sh --fast-path "$base_ref" HEAD
 ```
 
-`make coverage-gate` is an enforcement step, not a profile generator. CI
-generates the candidate, supporting, merge-base, and (when risk-selected)
-native profiles before the aggregate `Coverage` context evaluates them. The
+`make coverage-gate` is an enforcement step, not a profile generator. For a
+standard PR, CI derives changed packages and their reverse-dependency test
+closure, then generates candidate and merge-base profiles with the same test
+scope and `coverpkg`. High-risk and protected-main runs use the complete
+profiles. Supporting and (when platform-selected) native profiles are
+generated before the aggregate `Coverage` context evaluates them. The
 aggregate and native gates require 100% coverage for changed executable Go
-statements. Overall coverage remains an unrounded, zero-tolerance merge-base
-non-regression check. Candidate and baseline profiles are evaluated by the
-same block-deduplicating checker; supporting policy and shortcut profiles
-contribute to changed-code coverage only. The checked-in badge is presentation
-only and is never read as a gate input.
+statements. Overall coverage remains an unrounded, zero-tolerance,
+scope-matched merge-base non-regression check. Candidate and baseline profiles
+are evaluated by the same block-deduplicating checker; supporting policy and
+shortcut profiles contribute to changed-code coverage only. The checked-in
+badge is presentation only and is never read as a gate input.
 
 Compatibility checks derive authoritative Interface snapshots from the PR
 merge-base and the latest reachable stable release. The candidate cannot bless
@@ -156,3 +212,7 @@ Do not require helper jobs, `Multi-profile E2E`, or an aggregate admission
 alias. Update ruleset contexts only after the new names have appeared on the
 protected branch, so a rename cannot silently remove enforcement or leave an
 unproducible required context.
+
+The branch ruleset also requires one approval after the latest push. Enable
+repository auto-merge and automatic head-branch deletion; keep the base-owned
+reviewer router outside the required-context list.

docs/releasing.md

@@ -1,19 +1,28 @@
 # 发布手册（预发 / 正式）
 
-发布只走一条受控链路：GitHub Actions 的 `Release` workflow 负责版本分配、封板、构建、签名和下游发布；Homebrew 以 workflow 自动创建的 Formula PR 经独立审核合入为交付边界。本地 `dws-release` 仍是兼容入口，但不再要求某一台固定电脑承担打包；不要直接运行 `goreleaser release`，也不要手工补打、移动或复用 tag。
+发布只走一条受控链路：GitHub Actions 的 `Release` workflow 负责版本分配、封板、构建、签名和下游发布；Homebrew Formula 在不可变 Release 资产及 checksum 通过校验后，由同一 workflow 直接写入 `main`，不再创建二次 PR。本地 `dws-release` 仍是兼容入口，但不再要求某一台固定电脑承担打包；不要直接运行 `goreleaser release`，也不要手工补打、移动或复用 tag。
 
-发布前必须完成平台治理：目标 GitHub 仓库已启用 immutable releases，`main` 精确要求 `CI` workflow 的九个 context：`Lint`、`Test`、`Coverage`、`Policy`、`Edition`、`Interface Integrity`、`AI Behavior`、`CLI Smoke`、`Mock MCP`。云端和本地入口都会在封 tag 前检查 immutable releases、当前 SHA 的全部九个 context 和在途 Release；`v*` tag ruleset 仍需仓库管理员预先配置。
+发布前必须完成平台治理：目标 GitHub 仓库已启用 immutable releases，`main` 精确要求 `CI` workflow 的九个 context：`Lint`、`Test`、`Coverage`、`Policy`、`Edition`、`Interface Integrity`、`AI Behavior`、`CLI Smoke`、`Mock MCP`。云端入口会在封 tag 前检查 immutable releases、当前 SHA 的全部九个 context、Environment 保护规则和在途 Release；`v*` tag ruleset 仍需仓库管理员预先配置。
 
 ## 推荐入口：GitHub 云端发布
 
-任何具有仓库写权限、因而可以手动运行 Actions workflow 的成员，都可以基于当时最新的 `main` 发起发布：
+入口页面是 [GitHub Actions → Release](https://github.com/DingTalk-Real-AI/dingtalk-workspace-cli/actions/workflows/release.yml)。在仓库页面依次点击 `Actions` → `Release` → `Run workflow` 即可操作；不需要通过 Agent 或本地机器触发。
 
-1. 在 GitHub Actions 打开 `Release`，选择 `Run workflow`，分支必须是默认分支 `main`。
+只有得到该仓库明确授权、最终权限为 `write`、`maintain` 或 `admin` 的协作者可以运行发布操作，workflow 还会对发起人和重新运行者做同样的权限复核。没有仓库写权限的外部贡献者以及仅有 `read` / `triage` 权限的成员不能规划或发布版本。两个渠道的授权边界如下：
+
+- beta：上述任一内部成员都可以直接规划和发布，不需要人工审批。
+- stable：上述任一内部成员都可以规划并发起发布；完成只读规划和治理检查后，workflow 会在 `release-stable` Environment 等待另一名仓库管理员通过 `Review deployments` 签收。申请人不能批准自己的请求，批准后原 run 自动继续，无需重新触发。
+
+基于当时最新的 `main` 发起发布：
+
+1. 在上述 `Release` 页面选择 `Run workflow`，分支必须是默认分支 `main`。
 2. `release_operation=plan`，选择 `release_channel=beta|stable`；仅在开始新 beta 线时选择 `release_bump=patch|minor|major`。
 3. workflow summary 会给出唯一的下一版本。把对应的精确 `CHANGELOG.md` 章节通过 PR 合入 `main`。
-4. 再次运行，改为 `release_operation=publish`，并输入 `PUBLISH beta` 或 `PUBLISH stable`。
+4. 再次运行，改为 `release_operation=publish`。beta 会直接进入自动化发布；stable 会在封 tag 前等待管理员签收。
 
-`plan` 是纯只读操作，不创建 tag、预留版本号或生成包。CHANGELOG 合入期间若另一个发布先占用了该版本，`publish` 会重新分配并因 CHANGELOG 章节不匹配而拒绝，需要重新 plan。`publish` 会先再次确认 dispatch SHA 仍是当前 `main`、Code Admission 和平台治理均通过，再由唯一的 write job 使用 GitHub API 原子创建 annotated tag；同一次 run 随即进入既有的跨平台构建、GitHub/npm、可选 OSS/Gitee 发布和 Homebrew PR DAG。内置 `GITHUB_TOKEN` 创建的 tag 不依赖第二条 workflow 被再次触发。
+`plan` 是纯只读操作，不创建 tag、预留版本号或生成包。CHANGELOG 合入期间若另一个发布先占用了该版本，`publish` 会重新分配并因 CHANGELOG 章节不匹配而拒绝，需要重新 plan。`publish` 会先再次确认 dispatch SHA 仍是当前 `main`、Code Admission 和平台治理均通过，再由唯一的 write job 使用 GitHub API 原子创建 annotated tag；同一次 run 随即进入既有的跨平台构建、GitHub/npm、可选 OSS/Gitee 发布和 Homebrew 直交付 DAG。内置 `GITHUB_TOKEN` 创建的 tag 不依赖第二条 workflow 被再次触发。
+
+为缩短封板前后的关键路径，`publish` 的只读版本规划会与平台治理检查并行，seal 仍严格等待二者成功；plan 在 candidate annotated tag 上验证过的 contract 和 stable/beta baseline 会绑定进 seal，并由 seal 后的 tag authority 检查复用。Code Admission 状态与 immutable-releases 治理仍会在 seal 后再次读取，避免 preflight 与发布之间的状态变化被忽略。随后三类只读门禁（release automation、命令兼容性、multi-profile E2E）与 GoReleaser 构建并行；Node/archive 等仅供后处理使用的工具也延后到构建完成后安装。并行和已验证结果复用只改变调度，不降低发布门禁：任何一条验证失败都会阻止 GitHub Release、npm、镜像和 Homebrew 发布，delivery proof 也要求三条验证 job 全部成功。
 
 OSS 镜像默认不参与发布 DAG，适用于尚未创建 Bucket 的仓库。云端封板会把当时的仓库变量 `ENABLE_OSS_MIRROR=true` 记录为不可变 tag 元数据 `OSS-Mirror: enabled`，否则记录为 `deferred`；后续发布和撤回只读取该 sealed policy，不读取变量的当前值。`enabled` 继续对缺失凭据、无效 Bucket、上传、pointer 和撤回失败保持 fail-closed；`deferred` 明确跳过不存在的渠道。为避免补发后撤回遗漏，deferred 版本暂不接受 `repair_oss_version`，启用 OSS 只影响后续新 tag，直到补齐可审计的不可变 repair 证明。
 
@@ -62,7 +71,9 @@ dws-release
 dws-release config --remote origin
 ```
 
-之后命令按仓库状态自动走到正确步骤：缺少精确 CHANGELOG 章节时只生成模板并停止；补全、提交并合入 `main` 后，再运行同一条命令就会安全快进本地 `main` 并执行完整预检。若同名 remote 后续被改指向其他仓库会直接拒绝。只有显式增加 `--publish` 才会进入 tag 发布，且底层仍要求最终版本确认。
+之后命令按仓库状态自动走到正确步骤：缺少精确 CHANGELOG 章节时只生成模板并停止；补全、提交并合入 `main` 后，再运行同一条命令就会安全快进本地 `main` 并执行完整预检。若同名 remote 后续被改指向其他仓库会直接拒绝。官方仓库不再接受本地 `--publish`，命令会直接提示上述 Actions 页面；本地入口不能绕过 beta/stable 的统一授权。
+
+Release workflow 不再监听新建的 `v*` tag；直接推 tag 不会发布 GitHub Release、npm 或镜像渠道。所有新 beta/stable 都必须从云端页面进入统一授权和审计链路；历史失败版本仍可通过受保护的 recovery 兼容处理。
 
 ## 发布模型
 
@@ -89,13 +100,7 @@ dws-release v1.2.3-beta.1
 dws-release v1.2.3-beta.1
 ```
 
-预检包含测试、策略检查、旧正式版命令树兼容检查、全平台打包、npm 安装验证，以及 macOS 环境下的 Homebrew 安装验证。它还会从默认分支触发一次无发布权限的 `Release governance preflight`，用正式流水线相同的身份检查该精确 commit 的九个 Code Admission context 和 immutable releases。通过后会在当前 Git worktree 的私有 Git 状态目录写入一个有效期六小时的证明，绑定版本、精确 commit、发布仓库、beta/stable 基线和远端 `main`：
-
-```bash
-dws-release v1.2.3-beta.1 --publish
-```
-
-若源码、版本、远端身份和 stable 基线均未变化，`--publish` 会复用该证明，只执行远端契约、发布身份和最终治理复核，不再重复测试与打包。也可以直接运行 `--publish`；没有可复用证明时只会完整执行一次预检。命令在封 tag 前仍要求再次输入完整版本号，统一入口不提供跳过确认的参数。
+预检包含测试、策略检查、旧正式版命令树兼容检查、全平台打包、npm 安装验证，以及 macOS 环境下的 Homebrew 安装验证。它还会从默认分支触发一次无发布权限的 `Release governance preflight`，用正式流水线相同的身份检查该精确 commit 的九个 Code Admission context 和 immutable releases。通过后回到上述 Actions 页面选择 beta 和 `release_operation=publish`；云端会重新绑定当前 `main`，然后直接进入 beta 自动发布，不需要人工审批或输入确认短语。
 
 ## 正式发布
 
@@ -105,14 +110,13 @@ beta 验证通过后，运行正式版入口：
 dws-release v1.2.3 --from-beta v1.2.3-beta.1
 ```
 
-首次运行只生成正式版 CHANGELOG 并停止。补全内容、删除 `TODO`，提交后通过 PR 合入 `main`；重新运行同一条命令做完整预检，确认后增加 `--publish`：
+首次运行只生成正式版 CHANGELOG 并停止。补全内容、删除 `TODO`，提交后通过 PR 合入 `main`；重新运行同一条命令做完整预检：
 
 ```bash
 dws-release v1.2.3 --from-beta v1.2.3-beta.1
-dws-release v1.2.3 --from-beta v1.2.3-beta.1 --publish
 ```
 
-本地入口的 `FROM_BETA` 不会自动推断；云端入口会按上述规则唯一选择。两种入口都会把它写入 stable annotated tag 的 `From-Beta` 元数据，CI 会再次读取和验证。
+预检通过后，在 Actions 页面选择 stable 和 `release_operation=publish`。云端入口会按上述规则唯一选择 beta，并把它写入 stable annotated tag 的 `From-Beta` 元数据；在创建 tag 前必须由另一名仓库管理员签收。
 
 ## CHANGELOG 契约
 
@@ -130,14 +134,14 @@ dws-release v1.2.3 --from-beta v1.2.3-beta.1 --publish
 
 ## CI/CD 保证
 
-- 只接受 `vX.Y.Z-beta.N` 和 `vX.Y.Z`，且新版本必须高于上一正式版。这里的“上一正式版”必须同时具备公开非草稿 GitHub Release 和同 tag/commit 的成功 Release workflow；只有 tag、没有交付成功的孤儿版本会阻断后续发布，要求走受保护恢复补齐。云端 tag 会固定 `Release-Run`、requester、commit 和版本分配指纹，交付验证按该精确 run/attempt 及完整 job graph 取证，不接受任意 `workflow_dispatch`。历史版本若曾通过专用 recovery workflow 完成交付，只能使用仓库内 `delivered-stable-recoveries.json` 中精确到 tag、commit、run、workflow SHA 与 attempt 的 reviewed 证据。
-- tag 必须是 annotated tag；本地脚本要求封板提交已通过 PR 合入并包含在远端 `main` 历史中，发布只推送 tag。CI 允许其后 `main` 继续前进，但始终要求封板提交位于 `main` 历史中。
+- 只接受 `vX.Y.Z-beta.N` 和 `vX.Y.Z`，且新版本必须高于上一正式版。这里的“上一正式版”必须同时具备公开非草稿 GitHub Release 和同 tag/commit 的成功 Release workflow；只有 tag、没有交付成功的孤儿版本会阻断后续发布，要求走机器核验恢复补齐。云端 tag 会固定 `Release-Run`、requester、commit 和版本分配指纹，交付验证按该精确 run/attempt 及完整 job graph 取证，不接受任意 `workflow_dispatch`。历史版本若曾通过专用 recovery workflow 完成交付，只能使用仓库内 `delivered-stable-recoveries.json` 中精确到 tag、commit、run、workflow SHA 与 attempt 的 reviewed 证据。
+- tag 必须由云端 seal job 创建为 annotated tag；封板提交必须已通过 PR 合入并包含在远端 `main` 历史中。流水线允许其后 `main` 继续前进，但始终要求封板提交位于 `main` 历史中。
 - 日常 CI 和发布前都会对比“最新已交付正式版”的完整命令树；若长时间预检期间该 baseline 发生变化，会针对新的 baseline 重新比较。
 - GoReleaser 只构建；Darwin 重签、checksums 重算和 npm 安装验证通过后，才统一上传 GitHub Release 的最终产物。
 - 六个平台归档会逐个解包并核验二进制内嵌版本；公开资产集合、checksums 集合和 npm tarball integrity 都必须精确一致。npm tarball 固定由 npm `10.9.2` 打包，避免重跑时因 runner 自带 npm 漂移产生不同字节。
 - stable 发布到 npm `latest`；prerelease 发布到 npm `beta`。启用 `ENABLE_OSS_MIRROR=true` 后，stable 同步 OSS `latest.txt` 和共享安装脚本，prerelease 只同步 OSS `beta.txt`，不会覆盖稳定入口。
 - Release workflow 使用一个最多容纳 100 个 pending run 的串行 publication queue；版本规划、云端封板、发布、恢复、修复和撤回共享同一发布锁。
-- 本地 tag push 失败时会删除本次新建的本地 tag。远端 tag 一旦创建，后续发布归 CI 所有；发布中途失败时走受保护恢复，禁止改 tag 指向或复用版本号。只有已经公开版本经过受保护的全渠道撤回并留下永久 `withdrawn/...` 墓碑后，撤回 workflow 才会在最后一步删除原 tag。
+- 云端 seal 创建远端 tag 后，后续发布归同一 run 所有；发布中途失败时先重跑同一 run 的失败 jobs，必须跨 run 时走机器核验恢复，禁止改 tag 指向或复用版本号。只有已经公开版本经过受保护的全渠道撤回并留下永久 `withdrawn/...` 墓碑后，撤回 workflow 才会在最后一步删除原 tag。
 
 npm 补发只允许从默认分支触发 Release workflow 的 `repair_npm_version`。它只支持启用 immutable releases 后、由本流水线成功产出的公开 immutable release：目标必须是 `main` 历史中的 annotated tag，并且同 commit 的 `Build immutable GitHub Release` job 已成功。即使后续 npm 分发失败，这个独立的产物封存边界仍可作为补发依据。补发会用目标 commit 的 npm 模板重组包，逐平台核验资产和二进制版本，再发布到隔离的 `backfill` dist-tag，不会回滚 `latest` / `beta`。历史 mutable release 不进入自动补发路径，避免把可被替换的资产带入 npm。
 
@@ -164,32 +168,35 @@ dws-release recover v1.2.3-beta.1
 
 - 输入精确绑定原 annotated tag object、commit 和失败的 sealed `Release` run；云端 run 还必须与 tag 内的 run ID、attempt、requester 完全一致，commit 必须仍在 `main` 历史中。
 - 目标只允许不存在 GitHub Release 或仍为 Draft；已经公开的版本不能全量重建：单个下游故障走对应的 channel repair，版本本身有问题则走受保护的全平台 withdrawal。
-- `release-recovery` environment 必须限制为受保护分支、配置至少一名 required reviewer，并禁止自审；workflow 会通过 API 复核这些设置，未配置时 fail closed。
+- 恢复不再进入人工审批 environment。workflow 会机器核验 tag object、commit、原失败 run/attempt、请求人、完整 seal metadata、`main` 祖先关系以及 Release 状态；任一事实不一致都会在构建前 fail closed。
 - 恢复复用正常的 contract、构建、Developer ID 签名、资产校验、immutable 发布、Homebrew、npm，以及已启用的 OSS jobs，不存在 recovery 专用 publisher 或门禁跳过。
 - 如果 GitHub Release 已在 recovery 中封存、后续 Homebrew/npm 校验发生瞬时失败，只重跑该 run 的 failed jobs；流水线仅在隐藏 run marker、tag object、commit 和 finalized artifact 字节全部精确一致时复用公开 Release。
 
 成功的默认分支恢复 run 会成为后续 beta → stable 和 stable baseline 验证的可审计交付证据；历史临时分支恢复仍只接受 reviewed manifest 中的固定证据。
 
-云端 seal 后不要使用 GitHub 的 “Re-run failed jobs” 作为交付修复：annotated tag 永久绑定最初的 run attempt，普通 rerun 不会成为可接受的交付证据。GitHub Release 尚未公开时走上述 protected recovery；已经公开且仅 npm/OSS/Gitee 某一渠道失败时走对应 repair；版本内容本身有问题时走 withdrawal。
+seal job 写入 tag 后如果只因 GitHub API 瞬时 404/429/5xx 或后续 job 失败，可直接使用 GitHub 的 “Re-run failed jobs”。同一 run 会精确复用原 release-plan；seal 只在 version、tag object、commit、channel、beta 来源、OSS policy、请求人、run ID 和完整 message 全部匹配且原 attempt 不大于当前 attempt 时认领已有 tag。不同 run 或任一字段不匹配时不会认领。GitHub Release 尚未公开且必须跨 run 重建时走上述机器核验 recovery；已经公开且仅 npm/OSS/Gitee 某一渠道失败时走对应 repair；版本内容本身有问题时走 withdrawal。
 
 OSS 的 `latest.txt` / `beta.txt` 是镜像频道元数据；当前仓库安装器仍主要从 GitHub/Gitee 解析版本。启用 OSS 后，发布和撤回把它作为受控分发渠道处理，保证一旦外部消费者接入该 pointer，也不会继续解析到已撤回版本；未启用时两条流程都明确跳过不存在的 OSS 渠道。
 
-Release workflow 会生成 Darwin/Linux 双架构 Formula，并分别为 stable/beta 打开 Homebrew PR；tap 的默认分支仍以独立审核合入为交付边界。撤回 workflow 使用相同模板和回退版本 checksums 打开反向 PR；问题 GitHub Release 会先被移除以阻止新安装，永久墓碑和 workflow 日志承担审计/续跑依据。
+Release workflow 会生成 Darwin/Linux 双架构 Formula，并在不可变资产逐个校验后，由 `HOMEBREW_PR_TOKEN` 所属的受控发布身份只提交对应 stable 或 beta Formula 文件到 `main`，不再创建二次发布 PR；并发 `main` 更新会以全新 clone 最多重试三次，绝不 force push。该身份的提交不会依赖另一轮 CI 来补齐证明：workflow 只在确认该 commit 单父、唯一改动为目标 Formula、内容与本次已验证产物逐字节一致，且父 commit 九项 Code Admission 全绿后，直接为 Formula-only commit 封存同名九项成功 checks，避免下一次发布因缺失 contexts 被卡住。撤回 workflow 暂时仍使用相同模板和回退版本 checksums 打开反向 PR；问题 GitHub Release 会先被移除以阻止新安装，永久墓碑和 workflow 日志承担审计/续跑依据。
 
 ## 平台治理前置
 
 仓库管理员还需要在 GitHub 平台配置以下不可由脚本替代的规则：
 
-- `main` 必须精确要求 `Lint`、`Test`、`Coverage`、`Policy`、`Edition`、`Interface Integrity`、`AI Behavior`、`CLI Smoke`、`Mock MCP` 九个 Code Admission context；tag workflow 也会通过 Checks API 再确认该封板 SHA 上九项全部成功。
+- `main` 必须精确要求 `Lint`、`Test`、`Coverage`、`Policy`、`Edition`、`Interface Integrity`、`AI Behavior`、`CLI Smoke`、`Mock MCP` 九个 Code Admission context；Release workflow 也会通过 Checks API 再确认该封板 SHA 上九项全部成功。
 - 必须启用 immutable releases；它只保护启用后发布的 release，因此应在第一次使用新流水线前配置。为 `v*` 增加 tag ruleset，限制创建权限，并在 release 发布前保护 tag 的短暂窗口。
 - tag ruleset 还必须覆盖 `withdrawn/v*`：只允许受保护的撤回 workflow 创建墓碑，禁止更新或删除墓碑；同时应允许 Release workflow 创建新的 `v*`，允许撤回 workflow 在全部渠道回退后删除精确的问题 `v*`。若组织级规则阻止这两个 workflow 的预期动作，发布或撤回会 fail closed，不能靠手工移动 tag 绕过。
-- 配置 `RELEASE_GOVERNANCE_TOKEN` Actions secret，只授予目标仓库 `Administration: read`；内置 `GITHUB_TOKEN` 不具备 immutable-releases API 所需的仓库治理权限。每次本地预检和 tag workflow 都使用这一个身份进行 fail-closed 验证。
+- 配置 `RELEASE_GOVERNANCE_TOKEN` Actions secret，只授予目标仓库 `Administration: read`；内置 `GITHUB_TOKEN` 不具备 immutable-releases API 所需的仓库治理权限。每次本地预检和云端发布都使用这一个身份进行 fail-closed 验证。
 - 配置 `APPLE_CERTIFICATE_P12_BASE64`、`APPLE_CERTIFICATE_PASSWORD` 和具备发布权限的 `NPM_TOKEN`；撤回还要求该 npm 身份能够执行 `deprecate` 和修改 dist-tag。
 - 启用 OSS 镜像时，先创建有效 Bucket，再设置仓库变量 `ENABLE_OSS_MIRROR=true`，并配置 `OSS_ACCESS_KEY_ID`、`OSS_ACCESS_KEY_SECRET`、`OSS_ENDPOINT`、`OSS_BUCKET`，按需配置 `OSS_PREFIX`。启用后发布保持 fail-closed；撤回身份必须能够补齐安全版本资产、写 `latest.txt` / `beta.txt` 并删除问题版本前缀。尚未 provision Bucket 时保持该变量未设置或不等于 `true`，新 tag 会封存 `OSS-Mirror: deferred` 并跳过 OSS；该版本不能通过现有 repair 流程事后改成启用。
 - 若启用 Gitee fallback，设置 `ENABLE_GITEE_UPLOAD_FALLBACK=true`，并配置 `GITEE_TOKEN`、`GITEE_USER`、`GITEE_REPO`；该身份必须能够创建和删除目标仓库的 Release 与 tag。
-- 单独配置 `HOMEBREW_PR_TOKEN`，优先使用仅授权本仓库且具备 `Contents: write`、`Pull requests: write` 的 fine-grained PAT；若组织策略不允许该账号使用 fine-grained PAT，则回退到仅带 `public_repo` scope 的专用 classic PAT。治理预检和 tag contract 会验证 token 身份、classic scope，并用 `[skip ci]` 临时分支和 draft PR 完成真实写权限 canary，随后立即关闭 PR、删除分支；任何清理失败都会 fail closed。门禁也会拒绝与治理 token 复用。
-- 创建 `release-recovery` environment，只允许受保护分支，设置 required reviewer、禁止自审并关闭管理员绕过。workflow 会读取 environment 的 required-reviewer、prevent-self-review 和 protected-branch 规则；规则缺失时紧急恢复会失败，正常 beta/stable tag 发布不受影响。
+- 正常 Homebrew 发布使用现有的 `HOMEBREW_PR_TOKEN` 直接提交 Formula-only commit，不再创建 Homebrew PR，也不跑权限 canary。GitHub 不允许内置 Actions App 作为当前仓库 ruleset 的 bypass actor，因此两个默认分支 ruleset 都只给该 token 所属的指定发布管理员用户 `always` bypass；仓库脚本仍会限制提交路径、校验 Ruby、禁止 force push，并在并发更新时重新基于最新 `main`。
+- `HOMEBREW_PR_TOKEN` 应保持仓库范围的 `Contents: write` 与 `Pull requests: write` 权限；后者仅供撤回流程创建回退 PR。不要与 `RELEASE_GOVERNANCE_TOKEN` 复用，并定期审计 token owner 与 ruleset bypass actor 一致。
+- 创建 `release-beta` environment，只允许受保护分支且不配置 required reviewer；仓库内部 `write`、`maintain`、`admin` 成员的 beta 发布会直接通过该边界。
+- 创建 `release-stable` environment，只允许受保护分支，以仓库管理员为 required reviewer，禁止申请人自审并关闭管理员绕过。内部成员可以发起 stable，但必须由另一名管理员签收后才能封 tag 和写入任何发布渠道。
+- Release workflow 会在封 tag 前回读并验证上述两套 Environment 规则；规则缺失、stable reviewer 不再是仓库管理员、或 beta 被误加人工审批时都会 fail closed。
 - 创建 `release-withdrawal` environment，只允许受保护分支，设置至少一名 required reviewer、禁止申请人自审并关闭管理员绕过。撤回 workflow 会通过 API 复核这些规则；任何一项缺失都会在触碰 npm、OSS、Gitee、Homebrew 或 GitHub Release 前失败。
-- 仓库或组织的 Actions 策略必须允许 `Release` 与 `Withdraw release` workflow 的 `GITHUB_TOKEN` 获得各 job 声明的 `contents: write`。若上述发布凭证采用 environment secret，确认 `release-withdrawal` 审批完成后能够读取撤回所需的 npm、OSS、Gitee 和 Homebrew 凭证。
+- 仓库或组织的 Actions 策略必须允许 `Release` 与 `Withdraw release` workflow 的 `GITHUB_TOKEN` 获得各 job 声明的权限；正常发布由 `HOMEBREW_PR_TOKEN` 更新两个受控 Formula 路径，内置 token 只承担 workflow 自身声明的封板与校验写入。若撤回凭证采用 environment secret，确认 `release-withdrawal` 审批完成后能够读取撤回所需的 npm、OSS、Gitee 和 Homebrew 凭证。
 
 immutable releases，或任一 Code Admission context 缺失、未成功时，发布脚本会自动拒绝封 tag。tag ruleset 可能来自组织层，脚本不自动推断其最终作用范围；管理员确认不能省略，脚本约定也不能替代平台强制。

internal/app/auth_command.go

@@ -40,12 +40,14 @@ import (
 )
 
 type authLoginConfig struct {
-	Token        string
-	Force        bool
-	Device       bool
-	Recommend    bool
-	Yes          bool
-	TargetCorpID string
+	Token                          string
+	Force                          bool
+	Device                         bool
+	Recommend                      bool
+	Yes                            bool
+	TargetCorpID                   string
+	HistoryProfileSelector         string
+	HistoryProfileSelectorExplicit bool
 }
 
 type authLoginGuideAction string
@@ -148,7 +150,10 @@ func newAuthLoginCommand(patCaller edition.ToolCaller) *cobra.Command {
 				provider.Output = cmd.ErrOrStderr()
 				provider.NoBrowser, _ = cmd.Flags().GetBool("no-browser")
 				provider.IdentityEnricher = func(ctx context.Context, data *authpkg.TokenData) error {
-					return enrichAuthLoginProfileFromContact(ctx, configDir, patCaller, data)
+					return enrichAuthLoginProfileFromContact(ctx, configDir, patCaller, data, authLoginHistoryHint{
+						Selector: cfg.HistoryProfileSelector,
+						Explicit: cfg.HistoryProfileSelectorExplicit,
+					})
 				}
 				tokenData, err = authDeviceLogin(provider, loginCtx)
 				if err != nil {
@@ -163,7 +168,10 @@ func newAuthLoginCommand(patCaller edition.ToolCaller) *cobra.Command {
 				provider.NoBrowser, _ = cmd.Flags().GetBool("no-browser")
 				provider.TargetCorpID = cfg.TargetCorpID
 				provider.IdentityEnricher = func(ctx context.Context, data *authpkg.TokenData) error {
-					return enrichAuthLoginProfileFromContact(ctx, configDir, patCaller, data)
+					return enrichAuthLoginProfileFromContact(ctx, configDir, patCaller, data, authLoginHistoryHint{
+						Selector: cfg.HistoryProfileSelector,
+						Explicit: cfg.HistoryProfileSelectorExplicit,
+					})
 				}
 				configureOAuthProviderCompatibility(provider, configDir)
 				tokenData, err = authOAuthLogin(provider, loginCtx, authLoginForcesAuthorization(cfg))
@@ -175,11 +183,23 @@ func newAuthLoginCommand(patCaller edition.ToolCaller) *cobra.Command {
 			ResetRuntimeTokenCache()
 			clearCompatCache()
 			w := cmd.OutOrStdout()
+			postLoginSelector := authpkg.TokenProfileSelector(tokenData)
+			if tokenData != nil && strings.TrimSpace(tokenData.CorpID) != "" && strings.TrimSpace(tokenData.UserID) == "" {
+				if profiles, loadErr := authLoadProfiles(configDir); loadErr == nil && profiles != nil {
+					for i := range profiles.Profiles {
+						profile := profiles.Profiles[i]
+						if strings.TrimSpace(profile.CorpID) == strings.TrimSpace(tokenData.CorpID) && strings.TrimSpace(profile.UserID) == "" {
+							postLoginSelector = profileCLISelector(profile, profiles)
+							break
+						}
+					}
+				}
+			}
 			runPostLoginAuthorization := func() error {
 				if !recommendAuthMode {
 					return nil
 				}
-				restoreProfile := replaceRuntimeProfile(authpkg.TokenProfileSelector(tokenData))
+				restoreProfile := replaceRuntimeProfile(postLoginSelector)
 				defer restoreProfile()
 				recommendScopeMode := pat.LoginRecommendScopeRecommended
 				var initialPlan *pat.LoginRecommendPlan
@@ -287,7 +307,7 @@ var (
 	migrateKeychainToFileDEK         = authpkg.MigrateKeychainToFileDEK
 	authMigrateTarget                = func(cmd *cobra.Command) (string, error) { return cmd.Flags().GetString("to") }
 	authRunForm                      = (*huh.Form).Run
-	authSaveTokenData                = authpkg.SaveTokenData
+	authSaveTokenData                = authpkg.SaveLoginTokenData
 	authSaveAppConfig                = authpkg.SaveAppConfig
 	authDeviceLogin                  = func(provider *authpkg.DeviceFlowProvider, ctx context.Context) (*authpkg.TokenData, error) {
 		return provider.Login(ctx)
@@ -494,7 +514,9 @@ func newAuthStatusCommand() *cobra.Command {
 				if selected == nil {
 					return apperrors.NewValidation(fmt.Sprintf("profile %q not found", profileSelector))
 				}
-				profileSelector = authpkg.ProfileSelector(*selected)
+				if strings.TrimSpace(selected.UserID) != "" {
+					profileSelector = authpkg.ProfileSelector(*selected)
+				}
 			}
 			restoreProfile := pushRuntimeProfile(profileSelector)
 			defer restoreProfile()
@@ -522,7 +544,11 @@ func newAuthStatusCommand() *cobra.Command {
 						_ = authDeleteTokenData(configDir)
 					} else if tokenData != nil {
 						refreshFailure = refreshErr
-						_ = authMarkProfileStatus(configDir, authpkg.TokenProfileSelector(tokenData), authpkg.ProfileStatusExpired)
+						markSelector := profileSelector
+						if markSelector == "" {
+							markSelector = authpkg.StableTokenProfileSelector(configDir, tokenData)
+						}
+						_ = authMarkProfileStatus(configDir, markSelector, authpkg.ProfileStatusExpired)
 					}
 				}
 				if refreshFailure == nil && authStatusAuthenticated(tokenData) {
@@ -652,7 +678,14 @@ func logoutOneProfile(_ *cobra.Command, ctx context.Context, configDir, selector
 	}
 	stableSelector := selected.CorpID
 	if exact {
-		stableSelector = authpkg.ProfileSelector(*selected)
+		if strings.TrimSpace(selected.UserID) == "" {
+			// A blank historical profile can coexist with exact accounts in the
+			// same organization. Preserve the exact local-name selector; reducing
+			// it to corpId would log out the entire organization.
+			stableSelector = strings.TrimSpace(selector)
+		} else {
+			stableSelector = authpkg.ProfileSelector(*selected)
+		}
 		if data, loadErr := authLoadTokenForProfile(configDir, stableSelector); loadErr == nil {
 			_ = authRevokeTokenForData(ctx, data)
 		}
@@ -661,7 +694,7 @@ func logoutOneProfile(_ *cobra.Command, ctx context.Context, configDir, selector
 			if profile.CorpID != selected.CorpID {
 				continue
 			}
-			if data, tokenErr := authLoadTokenForProfile(configDir, authpkg.ProfileSelector(profile)); tokenErr == nil {
+			if data, tokenErr := authLoadTokenForProfile(configDir, profileCLISelector(profile, cfg)); tokenErr == nil {
 				_ = authRevokeTokenForData(ctx, data)
 			}
 		}
@@ -687,7 +720,7 @@ func logoutAllProfiles(_ *cobra.Command, ctx context.Context, configDir string)
 		_ = authRevokeToken(ctx)
 	} else {
 		for _, profile := range cfg.Profiles {
-			if data, tokenErr := authLoadTokenForProfile(configDir, authpkg.ProfileSelector(profile)); tokenErr == nil {
+			if data, tokenErr := authLoadTokenForProfile(configDir, profileCLISelector(profile, cfg)); tokenErr == nil {
 				_ = authRevokeTokenForData(ctx, data)
 			}
 		}
@@ -1206,7 +1239,7 @@ func resolveAuthLoginConfig(cmd *cobra.Command) (authLoginConfig, error) {
 		yes, _ = cmd.Root().PersistentFlags().GetBool("yes")
 		profileSelector, _ = cmd.Root().PersistentFlags().GetString("profile")
 	}
-	targetCorpID, err := resolveAuthLoginTargetCorpID(defaultConfigDir(), profileSelector)
+	targetCorpID, historyProfileSelector, historyProfileSelectorExplicit, err := resolveAuthLoginTarget(defaultConfigDir(), profileSelector)
 	if err != nil {
 		return authLoginConfig{}, err
 	}
@@ -1221,15 +1254,18 @@ func resolveAuthLoginConfig(cmd *cobra.Command) (authLoginConfig, error) {
 		"flow", flow,
 		"profile_selector", strings.TrimSpace(profileSelector),
 		"target_corp_id", targetCorpID,
+		"history_profile_selector", historyProfileSelector,
 		"recommend", recommend,
 	)
 	return authLoginConfig{
-		Token:        strings.TrimSpace(token),
-		Force:        force,
-		Device:       device,
-		Recommend:    recommend,
-		Yes:          yes,
-		TargetCorpID: targetCorpID,
+		Token:                          strings.TrimSpace(token),
+		Force:                          force,
+		Device:                         device,
+		Recommend:                      recommend,
+		Yes:                            yes,
+		TargetCorpID:                   targetCorpID,
+		HistoryProfileSelector:         historyProfileSelector,
+		HistoryProfileSelectorExplicit: historyProfileSelectorExplicit,
 	}, nil
 }
 
@@ -1238,17 +1274,57 @@ func authLoginForcesAuthorization(_ authLoginConfig) bool {
 }
 
 func resolveAuthLoginTargetCorpID(configDir, selector string) (string, error) {
+	targetCorpID, _, _, err := resolveAuthLoginTarget(configDir, selector)
+	return targetCorpID, err
+}
+
+// resolveAuthLoginTarget keeps the authorization target separate from the
+// local identity hint used only when contact cannot resolve the logged-in
+// account. An implicit current profile must never constrain a fresh OAuth
+// authorization to that profile's organization.
+func resolveAuthLoginTarget(configDir, selector string) (targetCorpID, historySelector string, explicit bool, err error) {
 	selector = strings.TrimSpace(selector)
 	if selector == "" {
-		return "", nil
+		if profile, resolveErr := authResolveProfile(configDir, ""); resolveErr == nil && profile != nil {
+			return "", authLoginHistorySelector(configDir, profile), false, nil
+		}
+		return "", "", false, nil
 	}
 	if profile, err := authResolveProfile(configDir, selector); err == nil && profile != nil {
-		return strings.TrimSpace(profile.CorpID), nil
+		historySelector := authLoginHistorySelector(configDir, profile)
+		_, _, identityExact := authpkg.ParseIdentitySelector(selector)
+		if selector != strings.TrimSpace(profile.CorpID) && selector != strings.TrimSpace(profile.CorpName) {
+			identityExact = true
+		}
+		return strings.TrimSpace(profile.CorpID), historySelector, identityExact, nil
+	}
+	if _, _, exact := authpkg.ParseIdentitySelector(selector); exact || strings.Contains(selector, ":") {
+		return "", "", true, apperrors.NewValidation(fmt.Sprintf("profile %q not found", selector))
 	}
 	if strings.HasPrefix(selector, "ding") {
-		return selector, nil
+		// A known organization that failed resolution is ambiguous (for
+		// example, two local accounts without an org-current pointer), not a
+		// request to invent a new corpId.
+		if cfg, loadErr := authLoadProfiles(configDir); loadErr == nil && cfg != nil {
+			for i := range cfg.Profiles {
+				if strings.TrimSpace(cfg.Profiles[i].CorpID) == selector {
+					return "", "", true, apperrors.NewValidation(fmt.Sprintf("profile %q is ambiguous; use an exact corpId:userId selector", selector))
+				}
+			}
+		}
+		return selector, "", false, nil
 	}
-	return "", apperrors.NewValidation(fmt.Sprintf("profile %q not found", selector))
+	return "", "", true, apperrors.NewValidation(fmt.Sprintf("profile %q not found", selector))
+}
+
+func authLoginHistorySelector(configDir string, profile *authpkg.Profile) string {
+	if profile == nil {
+		return ""
+	}
+	if cfg, err := authLoadProfiles(configDir); err == nil && cfg != nil {
+		return authpkg.ProfileSelectionSelector(*profile, cfg)
+	}
+	return authpkg.ProfileSelector(*profile)
 }
 
 type contactProfileIdentity struct {
@@ -1258,12 +1334,23 @@ type contactProfileIdentity struct {
 	UserName string
 }
 
+type authLoginHistoryHint struct {
+	Selector string
+	Explicit bool
+}
+
 type tokenOverrideToolCaller interface {
 	CallToolWithToken(ctx context.Context, token, productID, toolName string, args map[string]any) (*edition.ToolResult, error)
 }
 
-func enrichAuthLoginProfileFromContact(ctx context.Context, _ string, caller edition.ToolCaller, data *authpkg.TokenData) error {
-	if caller == nil || data == nil {
+func enrichAuthLoginProfileFromContact(
+	ctx context.Context,
+	configDir string,
+	caller edition.ToolCaller,
+	data *authpkg.TokenData,
+	hints ...authLoginHistoryHint,
+) error {
+	if data == nil {
 		return nil
 	}
 	corpID := strings.TrimSpace(data.CorpID)
@@ -1288,6 +1375,27 @@ func enrichAuthLoginProfileFromContact(ctx context.Context, _ string, caller edi
 		)
 		return nil
 	}
+	hint := authLoginHistoryHint{}
+	if len(hints) > 0 {
+		hint = hints[0]
+	}
+	tryHistory := func() bool {
+		reused, historyErr := enrichAuthLoginProfileFromHistory(configDir, data, hint)
+		if historyErr != nil {
+			logging.AuthDebug(
+				"auth.login.identity.history.error",
+				"corp_id", corpID,
+				"error", historyErr,
+			)
+		}
+		return reused
+	}
+	if caller == nil {
+		if strings.TrimSpace(data.UserID) == "" {
+			tryHistory()
+		}
+		return nil
+	}
 
 	var (
 		result *edition.ToolResult
@@ -1297,7 +1405,7 @@ func enrichAuthLoginProfileFromContact(ctx context.Context, _ string, caller edi
 		result, err = tokenCaller.CallToolWithToken(ctx, data.AccessToken, "contact", "get_current_user_profile", nil)
 	} else {
 		if strings.TrimSpace(data.UserID) == "" {
-			return fmt.Errorf("login identity lookup requires an in-memory token override")
+			tryHistory()
 		}
 		return nil
 	}
@@ -1311,11 +1419,15 @@ func enrichAuthLoginProfileFromContact(ctx context.Context, _ string, caller edi
 		if strings.TrimSpace(data.UserID) != "" {
 			return nil
 		}
-		return err
+		tryHistory()
+		return nil
 	}
-	identity, ok := contactProfileIdentityFromToolResult(result)
+	identity, ok := contactProfileIdentityFromToolResult(result, corpID)
 	if !ok {
 		logging.AuthDebug("auth.login.identity.lookup.empty", "corp_id", corpID)
+		if strings.TrimSpace(data.UserID) == "" {
+			tryHistory()
+		}
 		return nil
 	}
 	logging.AuthDebug(
@@ -1327,19 +1439,42 @@ func enrichAuthLoginProfileFromContact(ctx context.Context, _ string, caller edi
 		"corp_name", strings.TrimSpace(identity.CorpName),
 	)
 	if identity.CorpID != "" && identity.CorpID != corpID {
-		return fmt.Errorf("contact profile corpId %q does not match login corpId %q", identity.CorpID, corpID)
+		logging.AuthDebug(
+			"auth.login.identity.lookup.mismatch",
+			"login_corp_id", corpID,
+			"contact_corp_id", strings.TrimSpace(identity.CorpID),
+		)
+		if strings.TrimSpace(data.UserID) == "" {
+			tryHistory()
+		}
+		return nil
 	}
 
 	updated := *data
+	exchangeUserID := strings.TrimSpace(data.UserID)
 	if identity.CorpName != "" {
 		updated.CorpName = identity.CorpName
 	}
-	if identity.UserID != "" {
+	if exchangeUserID == "" && identity.UserID != "" {
 		updated.UserID = identity.UserID
 	}
-	if identity.UserName != "" {
+	if identity.UserName != "" && (exchangeUserID == "" || identity.UserID == "" || identity.UserID == exchangeUserID) {
 		updated.UserName = identity.UserName
 	}
+	if strings.TrimSpace(updated.UserID) == "" {
+		reused, historyErr := enrichAuthLoginProfileFromHistory(configDir, &updated, hint)
+		if historyErr != nil {
+			logging.AuthDebug(
+				"auth.login.identity.history.error",
+				"corp_id", corpID,
+				"error", historyErr,
+			)
+		}
+		if reused {
+			*data = updated
+			return nil
+		}
+	}
 	if updated.CorpName == data.CorpName && updated.UserID == data.UserID && updated.UserName == data.UserName {
 		logging.AuthDebug(
 			"auth.login.identity.resolved",
@@ -1361,7 +1496,144 @@ func enrichAuthLoginProfileFromContact(ctx context.Context, _ string, caller edi
 	return nil
 }
 
-func contactProfileIdentityFromToolResult(result *edition.ToolResult) (contactProfileIdentity, bool) {
+// enrichAuthLoginProfileFromHistory recovers display metadata when the contact
+// service cannot describe an external-worker account. Historical profile
+// selection is never proof of the user who completed a fresh authorization:
+// only the token exchange or contact service may supply UserID.
+//
+// An explicit profile remains useful as a storage/selection hint. Keeping it in
+// LegacyOrgScopedProfile prevents the login from switching the process-global
+// current profile while SaveTokenData publishes the UID-less credential to the
+// unresolved organization slot. A historical blank profile is updated in
+// place; an exact historical profile and its token remain untouched.
+func enrichAuthLoginProfileFromHistory(configDir string, data *authpkg.TokenData, hints ...authLoginHistoryHint) (bool, error) {
+	if data == nil || strings.TrimSpace(data.UserID) != "" {
+		return false, nil
+	}
+	corpID := strings.TrimSpace(data.CorpID)
+	if corpID == "" {
+		return false, nil
+	}
+	cfg, err := authLoadProfiles(configDir)
+	if err != nil {
+		return false, err
+	}
+	if cfg == nil {
+		return false, nil
+	}
+	sameCorp := make([]*authpkg.Profile, 0, len(cfg.Profiles))
+	for i := range cfg.Profiles {
+		profile := &cfg.Profiles[i]
+		if strings.TrimSpace(profile.CorpID) != corpID {
+			continue
+		}
+		sameCorp = append(sameCorp, profile)
+	}
+	if len(sameCorp) == 0 {
+		return false, nil
+	}
+
+	hint := authLoginHistoryHint{}
+	if len(hints) > 0 {
+		hint = hints[0]
+	}
+	var candidate *authpkg.Profile
+	if hint.Explicit {
+		candidate = historicalProfileForSelector(corpID, hint.Selector, sameCorp)
+		if candidate == nil {
+			// An explicit account is a hard identity boundary. If that exact
+			// historical hint no longer matches the token's organization, do
+			// not silently substitute org-current, sole, or global-current.
+			return false, nil
+		}
+	} else if len(sameCorp) > 1 {
+		// Organization-current is a storage preference, not proof of which user
+		// completed a fresh authorization. With multiple accounts, only an exact
+		// user selection may be used when the token/contact response has no UID.
+		return false, nil
+	} else {
+		candidate = sameCorp[0]
+	}
+
+	updated := *data
+	if hint.Explicit {
+		updated.LegacyOrgScopedProfile = strings.TrimSpace(hint.Selector)
+	}
+	if strings.TrimSpace(updated.CorpName) == "" {
+		updated.CorpName = strings.TrimSpace(candidate.CorpName)
+	}
+	if strings.TrimSpace(updated.UserName) == "" {
+		updated.UserName = strings.TrimSpace(candidate.UserName)
+	}
+	*data = updated
+	logging.AuthDebug(
+		"auth.login.identity.resolved",
+		"source", "local_profile_history_display_only",
+		"corp_id", corpID,
+		"user_id", updated.UserID,
+		"user_name", updated.UserName,
+		"corp_name", updated.CorpName,
+		"identity_proven", false,
+	)
+	return true, nil
+}
+
+func historicalProfileForSelector(corpID, selector string, profiles []*authpkg.Profile) *authpkg.Profile {
+	selector = strings.TrimSpace(selector)
+	if selector == "" {
+		return nil
+	}
+	cfg := &authpkg.ProfilesConfig{Profiles: make([]authpkg.Profile, 0, len(profiles))}
+	for _, profile := range profiles {
+		if profile != nil {
+			cfg.Profiles = append(cfg.Profiles, *profile)
+		}
+	}
+	var stableMatch *authpkg.Profile
+	for _, profile := range profiles {
+		if profile == nil || strings.TrimSpace(profile.CorpID) != strings.TrimSpace(corpID) ||
+			authpkg.ProfileSelectionSelector(*profile, cfg) != selector {
+			continue
+		}
+		if stableMatch != nil {
+			return nil
+		}
+		stableMatch = profile
+	}
+	if stableMatch != nil {
+		return stableMatch
+	}
+	if selectedCorpID, userID, exact := authpkg.ParseIdentitySelector(selector); exact {
+		if strings.TrimSpace(selectedCorpID) != strings.TrimSpace(corpID) {
+			return nil
+		}
+		for _, profile := range profiles {
+			if profile != nil && strings.TrimSpace(profile.UserID) == strings.TrimSpace(userID) {
+				return profile
+			}
+		}
+		return nil
+	}
+	var named *authpkg.Profile
+	for _, profile := range profiles {
+		if profile == nil || strings.TrimSpace(profile.Name) != selector {
+			continue
+		}
+		if named != nil {
+			return nil
+		}
+		named = profile
+	}
+	if named != nil {
+		return named
+	}
+	if selector == strings.TrimSpace(corpID) && len(profiles) == 1 {
+		return profiles[0]
+	}
+	return nil
+}
+
+func contactProfileIdentityFromToolResult(result *edition.ToolResult, expectedCorpIDs ...string) (contactProfileIdentity, bool) {
 	if result == nil {
 		return contactProfileIdentity{}, false
 	}
@@ -1369,14 +1641,14 @@ func contactProfileIdentityFromToolResult(result *edition.ToolResult) (contactPr
 		if strings.TrimSpace(block.Text) == "" {
 			continue
 		}
-		if identity, ok := contactProfileIdentityFromJSON([]byte(block.Text)); ok {
+		if identity, ok := contactProfileIdentityFromJSON([]byte(block.Text), expectedCorpIDs...); ok {
 			return identity, true
 		}
 	}
 	return contactProfileIdentity{}, false
 }
 
-func contactProfileIdentityFromJSON(data []byte) (contactProfileIdentity, bool) {
+func contactProfileIdentityFromJSON(data []byte, expectedCorpIDs ...string) (contactProfileIdentity, bool) {
 	var payload struct {
 		Result []struct {
 			OrgEmployeeModel struct {
@@ -1396,14 +1668,40 @@ func contactProfileIdentityFromJSON(data []byte) (contactProfileIdentity, bool)
 	if len(payload.Result) == 0 {
 		return contactProfileIdentity{}, false
 	}
-	org := payload.Result[0].OrgEmployeeModel
-	identity := contactProfileIdentity{
-		CorpID:   strings.TrimSpace(org.CorpID),
-		CorpName: strings.TrimSpace(org.OrgName),
-		UserID:   firstNonEmptyString(org.UserID, org.UserIDLower, org.OrgUserID),
-		UserName: firstNonEmptyString(org.OrgUserName, org.Name),
+	identities := make([]contactProfileIdentity, 0, len(payload.Result))
+	for i := range payload.Result {
+		org := payload.Result[i].OrgEmployeeModel
+		identity := contactProfileIdentity{
+			CorpID:   strings.TrimSpace(org.CorpID),
+			CorpName: strings.TrimSpace(org.OrgName),
+			UserID:   firstNonEmptyString(org.UserID, org.UserIDLower, org.OrgUserID),
+			UserName: firstNonEmptyString(org.OrgUserName, org.Name),
+		}
+		if identity.CorpID != "" || identity.CorpName != "" || identity.UserID != "" || identity.UserName != "" {
+			identities = append(identities, identity)
+		}
 	}
-	return identity, identity.CorpID != "" || identity.CorpName != "" || identity.UserID != "" || identity.UserName != ""
+	if len(identities) == 0 {
+		return contactProfileIdentity{}, false
+	}
+	expectedCorpID := ""
+	if len(expectedCorpIDs) > 0 {
+		expectedCorpID = strings.TrimSpace(expectedCorpIDs[0])
+	}
+	if expectedCorpID != "" {
+		for _, identity := range identities {
+			if identity.CorpID == expectedCorpID {
+				return identity, true
+			}
+		}
+		// Older contact responses omit corpId. A single result is still
+		// unambiguous; multiple organization records without a target match
+		// must fall back to local history instead of choosing result[0].
+		if len(payload.Result) != 1 {
+			return contactProfileIdentity{}, false
+		}
+	}
+	return identities[0], true
 }
 
 func firstNonEmptyString(values ...string) string {

internal/app/auth_command_full_coverage_test.go

@@ -262,7 +262,10 @@ func TestCrossPlatformCoverageAuthCoverageLoginFlows(t *testing.T) {
 	if _, _, err := authCoverageRunLogin(t, nil, "table", true, map[string]string{"device": "true"}); err == nil {
 		t.Fatal("device error should propagate")
 	}
-	authDeviceLogin = func(*authpkg.DeviceFlowProvider, context.Context) (*authpkg.TokenData, error) {
+	authDeviceLogin = func(provider *authpkg.DeviceFlowProvider, _ context.Context) (*authpkg.TokenData, error) {
+		if provider.IdentityEnricher == nil {
+			t.Error("device login missing shared identity enricher")
+		}
 		return &authpkg.TokenData{AccessToken: "a", ExpiresAt: time.Now().Add(time.Hour)}, nil
 	}
 	if _, _, err := authCoverageRunLogin(t, nil, "table", true, map[string]string{"device": "true", "no-browser": "true"}); err != nil {
@@ -275,7 +278,10 @@ func TestCrossPlatformCoverageAuthCoverageLoginFlows(t *testing.T) {
 	if _, _, err := authCoverageRunLogin(t, nil, "table", true, nil); err == nil {
 		t.Fatal("oauth error should propagate")
 	}
-	authOAuthLogin = func(*authpkg.OAuthProvider, context.Context, bool) (*authpkg.TokenData, error) {
+	authOAuthLogin = func(provider *authpkg.OAuthProvider, _ context.Context, _ bool) (*authpkg.TokenData, error) {
+		if provider.IdentityEnricher == nil {
+			t.Error("OAuth login missing shared identity enricher")
+		}
 		return &authpkg.TokenData{
 			AccessToken: "a", ExpiresAt: time.Now().Add(time.Hour), RefreshToken: "r", RefreshExpAt: time.Now().Add(48 * time.Hour),
 			CorpName: "Corp", CorpID: "ding1", UserName: "User", UserID: "u",
@@ -356,8 +362,8 @@ func TestCrossPlatformCoverageAuthCoverageContactEnrichment(t *testing.T) {
 	if err := enrichAuthLoginProfileFromContact(ctx, "cfg", &authCoverageCaller{}, complete); err != nil {
 		t.Fatal(err)
 	}
-	if err := enrichAuthLoginProfileFromContact(ctx, "cfg", &authCoverageCaller{err: errors.New("call")}, &authpkg.TokenData{CorpID: "ding"}); err == nil {
-		t.Fatal("caller error should propagate")
+	if err := enrichAuthLoginProfileFromContact(ctx, "cfg", &authCoverageCaller{err: errors.New("call")}, &authpkg.TokenData{CorpID: "ding"}); err != nil {
+		t.Fatalf("contact failure must remain best effort: %v", err)
 	}
 	if err := enrichAuthLoginProfileFromContact(
 		ctx,
@@ -374,8 +380,8 @@ func TestCrossPlatformCoverageAuthCoverageContactEnrichment(t *testing.T) {
 		}
 	}
 	mismatch := &authCoverageCaller{result: &edition.ToolResult{Content: []edition.ContentBlock{{Text: `{"result":[{"orgEmployeeModel":{"corpId":"other"}}]}`}}}}
-	if err := enrichAuthLoginProfileFromContact(ctx, "cfg", mismatch, &authpkg.TokenData{CorpID: "ding", AccessToken: "token"}); err == nil {
-		t.Fatal("corp mismatch should fail")
+	if err := enrichAuthLoginProfileFromContact(ctx, "cfg", mismatch, &authpkg.TokenData{CorpID: "ding", AccessToken: "token"}); err != nil {
+		t.Fatalf("contact corp mismatch must remain best effort: %v", err)
 	}
 	same := &authCoverageCaller{result: &edition.ToolResult{Content: []edition.ContentBlock{{Text: `{"result":[{"orgEmployeeModel":{"corpId":"ding","orgName":"Corp","userid":"u","name":"User"}}]}`}}}}
 	if err := enrichAuthLoginProfileFromContact(ctx, "cfg", same, complete); err != nil {
@@ -390,6 +396,25 @@ func TestCrossPlatformCoverageAuthCoverageContactEnrichment(t *testing.T) {
 	if err := enrichAuthLoginProfileFromContact(ctx, "cfg", same, data); err != nil || data.CorpName != "Corp" || data.UserID != "u" {
 		t.Fatalf("enriched = %#v, %v", data, err)
 	}
+	known := &authpkg.TokenData{CorpID: "ding", UserID: "exchange-user", AccessToken: "token"}
+	differentContactUser := &authCoverageCaller{result: &edition.ToolResult{Content: []edition.ContentBlock{{Text: `{"result":[{"orgEmployeeModel":{"corpId":"ding","orgName":"Corp","userid":"other-user","name":"Other User"}}]}`}}}}
+	if err := enrichAuthLoginProfileFromContact(ctx, "cfg", differentContactUser, known); err != nil {
+		t.Fatal(err)
+	}
+	if known.UserID != "exchange-user" || known.UserName != "" || known.CorpName != "Corp" {
+		t.Fatalf("token-exchange identity was overwritten: %#v", known)
+	}
+	multiOrg := &authCoverageCaller{result: &edition.ToolResult{Content: []edition.ContentBlock{{Text: `{"result":[{"orgEmployeeModel":{"corpId":"other","userid":"other-user"}},{"orgEmployeeModel":{"corpId":"ding","orgName":"Target Corp","userid":"target-user","name":"Target User"}}]}`}}}}
+	multiOrgData := &authpkg.TokenData{CorpID: "ding", AccessToken: "token"}
+	if err := enrichAuthLoginProfileFromContact(ctx, "cfg", multiOrg, multiOrgData); err != nil || multiOrgData.UserID != "target-user" || multiOrgData.CorpName != "Target Corp" {
+		t.Fatalf("multi-org contact selection = %#v, %v", multiOrgData, err)
+	}
+	if _, ok := contactProfileIdentityFromJSON(
+		[]byte(`{"result":[{"orgEmployeeModel":{"corpId":"other-a","userid":"user-a"}},{"orgEmployeeModel":{"corpId":"other-b","userid":"user-b"}}]}`),
+		"ding",
+	); ok {
+		t.Fatal("multiple nonmatching organizations must not select an arbitrary contact identity")
+	}
 	if _, ok := contactProfileIdentityFromToolResult(nil); ok {
 		t.Fatal("nil result should not parse")
 	}
@@ -398,6 +423,570 @@ func TestCrossPlatformCoverageAuthCoverageContactEnrichment(t *testing.T) {
 	}
 }
 
+func TestCrossPlatformCoverageContactFailureReusesOnlySameCorpHistoricalDisplayMetadata(t *testing.T) {
+	configDir := t.TempDir()
+	if err := authpkg.SaveProfiles(configDir, &authpkg.ProfilesConfig{
+		Version: 1,
+		Profiles: []authpkg.Profile{{
+			CorpID:   "ding_ecological_worker",
+			CorpName: "Historical Corp",
+			UserID:   "external-user",
+			UserName: "Historical Worker",
+		}},
+	}); err != nil {
+		t.Fatalf("SaveProfiles() error = %v", err)
+	}
+
+	for _, tc := range []struct {
+		name     string
+		caller   edition.ToolCaller
+		wantCorp string
+	}{
+		{
+			name: "contact business error",
+			caller: &authCoverageCaller{err: apperrors.NewAPI(
+				"business error: success=false",
+				apperrors.WithReason("business_error"),
+			)},
+			wantCorp: "Fresh Corp",
+		},
+		{
+			name:     "contact has no identity",
+			caller:   &authCoverageCaller{result: &edition.ToolResult{Content: []edition.ContentBlock{{Text: `{"success":false}`}}}},
+			wantCorp: "Fresh Corp",
+		},
+		{
+			name: "contact identity is missing user id",
+			caller: &authCoverageCaller{result: &edition.ToolResult{Content: []edition.ContentBlock{{
+				Text: `{"result":[{"orgEmployeeModel":{"corpId":"ding_ecological_worker","orgName":"Contact Corp"}}]}`,
+			}}}},
+			wantCorp: "Contact Corp",
+		},
+		{
+			name:     "ordinary contact error",
+			caller:   &authCoverageCaller{err: errors.New("network failure")},
+			wantCorp: "Fresh Corp",
+		},
+		{
+			name: "other contact business error",
+			caller: &authCoverageCaller{err: apperrors.NewAPI(
+				"permission denied",
+				apperrors.WithReason("business_error"),
+			)},
+			wantCorp: "Fresh Corp",
+		},
+		{
+			name:     "contact caller unavailable",
+			caller:   nil,
+			wantCorp: "Fresh Corp",
+		},
+		{
+			name: "contact returns another organization",
+			caller: &authCoverageCaller{result: &edition.ToolResult{Content: []edition.ContentBlock{{
+				Text: `{"result":[{"orgEmployeeModel":{"corpId":"ding_other","userid":"other-user"}}]}`,
+			}}}},
+			wantCorp: "Fresh Corp",
+		},
+	} {
+		t.Run(tc.name, func(t *testing.T) {
+			data := &authpkg.TokenData{
+				AccessToken:  "new-access",
+				RefreshToken: "new-refresh",
+				CorpID:       "ding_ecological_worker",
+				CorpName:     "Fresh Corp",
+			}
+			if err := enrichAuthLoginProfileFromContact(context.Background(), configDir, tc.caller, data); err != nil {
+				t.Fatalf("contact failure blocked historical identity recovery: %v", err)
+			}
+			if data.UserID != "" || data.UserName != "Historical Worker" {
+				t.Fatalf("historical metadata supplied UID evidence: %#v", data)
+			}
+			if data.CorpName != tc.wantCorp {
+				t.Fatalf("corp name = %q, want %q", data.CorpName, tc.wantCorp)
+			}
+			if data.AccessToken != "new-access" || data.RefreshToken != "new-refresh" {
+				t.Fatalf("new token material was changed: %#v", data)
+			}
+		})
+	}
+}
+
+func TestCrossPlatformCoverageContactFailureDoesNotGuessHistoricalIdentity(t *testing.T) {
+	businessErr := apperrors.NewAPI(
+		"business error: success=false",
+		apperrors.WithReason("business_error"),
+	)
+	for _, tc := range []struct {
+		name     string
+		corpID   string
+		profiles []authpkg.Profile
+		callErr  error
+	}{
+		{
+			name:   "same corp has two identities",
+			corpID: "ding_ecological_worker",
+			profiles: []authpkg.Profile{
+				{CorpID: "ding_ecological_worker", UserID: "external-user"},
+				{CorpID: "ding_ecological_worker", UserID: "external-user-b"},
+			},
+			callErr: businessErr,
+		},
+		{
+			name:   "same corp has one identity and one blank profile",
+			corpID: "ding_ecological_worker",
+			profiles: []authpkg.Profile{
+				{CorpID: "ding_ecological_worker", UserID: "external-user"},
+				{CorpID: "ding_ecological_worker"},
+			},
+			callErr: businessErr,
+		},
+		{
+			name:   "identity belongs to another corp",
+			corpID: "ding_ecological_worker",
+			profiles: []authpkg.Profile{
+				{CorpID: "ding_other", UserID: "external-user"},
+			},
+			callErr: businessErr,
+		},
+		{
+			name:    "no historical identity",
+			corpID:  "ding_ecological_worker",
+			callErr: businessErr,
+		},
+	} {
+		t.Run(tc.name, func(t *testing.T) {
+			configDir := t.TempDir()
+			if err := authpkg.SaveProfiles(configDir, &authpkg.ProfilesConfig{Version: 2, Profiles: tc.profiles}); err != nil {
+				t.Fatalf("SaveProfiles() error = %v", err)
+			}
+			data := &authpkg.TokenData{AccessToken: "new-access", CorpID: tc.corpID}
+			err := enrichAuthLoginProfileFromContact(
+				context.Background(),
+				configDir,
+				&authCoverageCaller{err: tc.callErr},
+				data,
+			)
+			if err != nil {
+				t.Fatalf("contact failure must not block unresolved legacy login: %v", err)
+			}
+			if data.UserID != "" {
+				t.Fatalf("ambiguous/cross-corp identity was reused: %#v", data)
+			}
+		})
+	}
+}
+
+func TestCrossPlatformCoverageContactHistoryFallbackEdges(t *testing.T) {
+	for _, data := range []*authpkg.TokenData{
+		nil,
+		{UserID: "known"},
+		{},
+	} {
+		reused, err := enrichAuthLoginProfileFromHistory(t.TempDir(), data)
+		if reused || err != nil {
+			t.Fatalf("ineligible history fallback = %v, %v", reused, err)
+		}
+	}
+
+	configDir := t.TempDir()
+	if err := authpkg.SaveProfiles(configDir, &authpkg.ProfilesConfig{
+		Version: 2,
+		Profiles: []authpkg.Profile{{
+			CorpID:   "ding_external",
+			CorpName: "Historical Corp",
+			UserID:   "external-user",
+			UserName: "Historical Worker",
+		}},
+	}); err != nil {
+		t.Fatalf("SaveProfiles() error = %v", err)
+	}
+	data := &authpkg.TokenData{CorpID: "ding_external"}
+	reused, err := enrichAuthLoginProfileFromHistory(configDir, data)
+	if err != nil || !reused {
+		t.Fatalf("history fallback = %v, %v", reused, err)
+	}
+	if data.CorpName != "Historical Corp" || data.UserName != "Historical Worker" || data.UserID != "" {
+		t.Fatalf("history metadata = %#v", data)
+	}
+
+	corruptDir := t.TempDir()
+	if err := os.Mkdir(authpkg.ProfilesPath(corruptDir), 0o700); err != nil {
+		t.Fatalf("create unreadable profiles path: %v", err)
+	}
+	if reused, err := enrichAuthLoginProfileFromHistory(corruptDir, &authpkg.TokenData{CorpID: "ding_external"}); reused || err == nil {
+		t.Fatalf("corrupt history fallback = %v, %v; want load error", reused, err)
+	}
+
+	businessErr := apperrors.NewAPI(
+		"business error: success=false",
+		apperrors.WithReason("business_error"),
+	)
+	for _, tc := range []struct {
+		name   string
+		caller *authCoverageCaller
+	}{
+		{
+			name:   "contact business error",
+			caller: &authCoverageCaller{err: businessErr},
+		},
+		{
+			name:   "contact has no identity",
+			caller: &authCoverageCaller{result: &edition.ToolResult{}},
+		},
+		{
+			name: "contact identity is missing user id",
+			caller: &authCoverageCaller{result: &edition.ToolResult{Content: []edition.ContentBlock{{
+				Text: `{"result":[{"orgEmployeeModel":{"corpId":"ding_external"}}]}`,
+			}}}},
+		},
+	} {
+		t.Run(tc.name, func(t *testing.T) {
+			err := enrichAuthLoginProfileFromContact(
+				context.Background(),
+				corruptDir,
+				tc.caller,
+				&authpkg.TokenData{CorpID: "ding_external", AccessToken: "new-access"},
+			)
+			if err != nil {
+				t.Fatalf("best-effort contact/history lookup blocked login: %v", err)
+			}
+		})
+	}
+}
+
+func TestCrossPlatformCoverageAuthLoginConfigPreservesHistoryIdentityHint(t *testing.T) {
+	t.Setenv("DWS_CONFIG_DIR", t.TempDir())
+	oldResolve := authResolveProfile
+	oldLoad := authLoadProfiles
+	t.Cleanup(func() {
+		authResolveProfile = oldResolve
+		authLoadProfiles = oldLoad
+	})
+
+	explicit := &authpkg.Profile{CorpID: "ding_same", UserID: "user_2", Name: "second"}
+	current := &authpkg.Profile{CorpID: "ding_current", UserID: "current_user"}
+	authResolveProfile = func(_ string, selector string) (*authpkg.Profile, error) {
+		switch selector {
+		case "ding_same:user_2":
+			clone := *explicit
+			return &clone, nil
+		case "external-worker":
+			return &authpkg.Profile{Name: "external-worker", CorpID: "ding_external"}, nil
+		case "":
+			clone := *current
+			return &clone, nil
+		default:
+			return nil, errors.New("missing")
+		}
+	}
+	authLoadProfiles = func(string) (*authpkg.ProfilesConfig, error) {
+		return &authpkg.ProfilesConfig{}, nil
+	}
+
+	cmd := newAuthLoginCommand(nil)
+	root, _, _ := authCoverageRoot(cmd, "table", true)
+	if err := root.PersistentFlags().Set("profile", "ding_same:user_2"); err != nil {
+		t.Fatal(err)
+	}
+	cfg, err := resolveAuthLoginConfig(cmd)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if cfg.TargetCorpID != "ding_same" || cfg.HistoryProfileSelector != "ding_same:user_2" || !cfg.HistoryProfileSelectorExplicit {
+		t.Fatalf("explicit login config = %#v", cfg)
+	}
+	if target, hint, exact, err := resolveAuthLoginTarget("cfg", "external-worker"); err != nil ||
+		target != "ding_external" || hint != "ding_external" || !exact {
+		t.Fatalf("blank-userId profile target = %q/%q/%v, %v", target, hint, exact, err)
+	}
+
+	if err := root.PersistentFlags().Set("profile", ""); err != nil {
+		t.Fatal(err)
+	}
+	cfg, err = resolveAuthLoginConfig(cmd)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if cfg.TargetCorpID != "" || cfg.HistoryProfileSelector != "ding_current:current_user" || cfg.HistoryProfileSelectorExplicit {
+		t.Fatalf("implicit login config constrained authorization target: %#v", cfg)
+	}
+
+	if _, _, _, err := resolveAuthLoginTarget("cfg", "ding_same:missing"); err == nil {
+		t.Fatal("missing exact profile must not be reinterpreted as a corpId")
+	}
+	if target, hint, explicitHint, err := resolveAuthLoginTarget("cfg", "ding_new"); err != nil || target != "ding_new" || hint != "" || explicitHint {
+		t.Fatalf("new organization target = %q/%q/%v, %v", target, hint, explicitHint, err)
+	}
+	authLoadProfiles = func(string) (*authpkg.ProfilesConfig, error) {
+		return &authpkg.ProfilesConfig{Profiles: []authpkg.Profile{
+			{CorpID: "ding_ambiguous", UserID: "user_1"},
+			{CorpID: "ding_ambiguous", UserID: "user_2"},
+		}}, nil
+	}
+	if _, _, _, err := resolveAuthLoginTarget("cfg", "ding_ambiguous"); err == nil {
+		t.Fatal("ambiguous known organization must require an exact profile")
+	}
+}
+
+func TestCrossPlatformCoverageOAuthAndDeviceKeepFreshUnknownIdentityIsolatedFromExactHistory(t *testing.T) {
+	oldResolve := authResolveProfile
+	oldLoad := authLoadProfiles
+	oldDevice := authDeviceLogin
+	oldOAuth := authOAuthLogin
+	oldInteractive := authLoginInteractiveTerminal
+	t.Cleanup(func() {
+		authResolveProfile = oldResolve
+		authLoadProfiles = oldLoad
+		authDeviceLogin = oldDevice
+		authOAuthLogin = oldOAuth
+		authLoginInteractiveTerminal = oldInteractive
+	})
+
+	authResolveProfile = authpkg.ResolveProfile
+	authLoadProfiles = authpkg.LoadProfiles
+	authLoginInteractiveTerminal = func() bool { return false }
+	t.Cleanup(func() { authpkg.SetRuntimeProfile("") })
+
+	for _, flow := range []string{"oauth", "device"} {
+		t.Run(flow, func(t *testing.T) {
+			configDir := t.TempDir()
+			keychainDir := t.TempDir()
+			t.Setenv("DWS_CONFIG_DIR", configDir)
+			t.Setenv(keychain.DisableKeychainEnv, "1")
+			t.Setenv(keychain.StorageDirEnv, keychainDir)
+			// StorageDirEnv isolates file-backed keychains, while Windows uses
+			// DPAPI-protected HKCU values. Give every flow its own namespace so
+			// OAuth/device fixtures cannot leak into each other or later tests.
+			t.Setenv(keychain.TestNamespaceEnv, keychainDir)
+			t.Cleanup(func() {
+				if err := keychain.RemoveAuthTokenEntries(keychain.Service); err != nil {
+					t.Errorf("clean auth keychain fixture: %v", err)
+				}
+			})
+			authpkg.SetRuntimeProfile("")
+
+			const (
+				corpID        = "ding_same"
+				historicalUID = "user_a"
+				exactSelector = corpID + ":" + historicalUID
+			)
+			oldToken := &authpkg.TokenData{
+				AccessToken:  "old-user-a-access",
+				RefreshToken: "old-user-a-refresh",
+				ExpiresAt:    time.Now().Add(time.Hour),
+				RefreshExpAt: time.Now().Add(24 * time.Hour),
+				CorpID:       corpID,
+				CorpName:     "Same Corp",
+				UserID:       historicalUID,
+				UserName:     "Historical User A",
+			}
+			if err := authpkg.SaveTokenData(configDir, oldToken); err != nil {
+				t.Fatalf("persist historical exact identity: %v", err)
+			}
+
+			caller := &authCoverageCaller{err: errors.New("contact unavailable")}
+			var enriched *authpkg.TokenData
+			freshToken := func() *authpkg.TokenData {
+				return &authpkg.TokenData{
+					AccessToken:  "fresh-user-b-access-" + flow,
+					RefreshToken: "fresh-user-b-refresh-" + flow,
+					ExpiresAt:    time.Now().Add(time.Hour),
+					RefreshExpAt: time.Now().Add(24 * time.Hour),
+					CorpID:       corpID,
+				}
+			}
+			persistUnknown := func(ctx context.Context, identityEnricher func(context.Context, *authpkg.TokenData) error) (*authpkg.TokenData, error) {
+				if identityEnricher == nil {
+					return nil, errors.New("missing identity enricher")
+				}
+				data := freshToken()
+				if err := identityEnricher(ctx, data); err != nil {
+					return nil, err
+				}
+				enriched = data
+				if data.UserID != "" {
+					return nil, fmt.Errorf("historical profile supplied unproven userId %q", data.UserID)
+				}
+				if err := authpkg.SaveTokenData(configDir, data); err != nil {
+					return nil, err
+				}
+				return data, nil
+			}
+
+			flags := map[string]string{"profile": exactSelector}
+			switch flow {
+			case "device":
+				flags["device"] = "true"
+				authDeviceLogin = func(provider *authpkg.DeviceFlowProvider, ctx context.Context) (*authpkg.TokenData, error) {
+					return persistUnknown(ctx, provider.IdentityEnricher)
+				}
+			case "oauth":
+				authOAuthLogin = func(provider *authpkg.OAuthProvider, ctx context.Context, _ bool) (*authpkg.TokenData, error) {
+					if provider.TargetCorpID != corpID {
+						return nil, fmt.Errorf("OAuth target corp = %q", provider.TargetCorpID)
+					}
+					return persistUnknown(ctx, provider.IdentityEnricher)
+				}
+			}
+			if _, _, err := authCoverageRunLogin(t, caller, "table", true, flags); err != nil {
+				t.Fatalf("%s login with unresolved fresh identity: %v", flow, err)
+			}
+			if enriched == nil || enriched.UserID != "" ||
+				enriched.LegacyOrgScopedProfile != exactSelector ||
+				enriched.CorpName != "Same Corp" ||
+				enriched.UserName != "Historical User A" {
+				t.Fatalf("%s history hint became identity evidence: %#v", flow, enriched)
+			}
+
+			historical, err := authpkg.LoadTokenDataForProfile(configDir, exactSelector)
+			if err != nil {
+				t.Fatalf("load historical exact identity: %v", err)
+			}
+			if historical.AccessToken != oldToken.AccessToken || historical.UserID != historicalUID {
+				t.Fatalf("historical exact slot was overwritten: %#v", historical)
+			}
+
+			profiles, err := authpkg.LoadProfiles(configDir)
+			if err != nil {
+				t.Fatalf("load profiles: %v", err)
+			}
+			var unresolved *authpkg.Profile
+			for i := range profiles.Profiles {
+				profile := &profiles.Profiles[i]
+				if profile.CorpID == corpID && profile.UserID == "" {
+					unresolved = profile
+					break
+				}
+			}
+			if unresolved == nil {
+				t.Fatalf("fresh UID-less token did not create an unresolved profile: %#v", profiles.Profiles)
+			}
+			unresolvedSelector := authpkg.ProfileSelectionSelector(*unresolved, profiles)
+			if unresolvedSelector == "" || unresolvedSelector == exactSelector {
+				t.Fatalf("unresolved selector = %q", unresolvedSelector)
+			}
+			fresh, err := authpkg.LoadTokenDataForProfile(configDir, unresolvedSelector)
+			if err != nil {
+				t.Fatalf("load fresh unresolved identity: %v", err)
+			}
+			if fresh.AccessToken != "fresh-user-b-access-"+flow || fresh.UserID != "" {
+				t.Fatalf("fresh token was not isolated in unresolved org slot: %#v", fresh)
+			}
+		})
+	}
+}
+
+func TestCrossPlatformCoverageHistoricalIdentityPriorityAndBlankUserID(t *testing.T) {
+	oldLoad := authLoadProfiles
+	t.Cleanup(func() { authLoadProfiles = oldLoad })
+	authLoadProfiles = func(string) (*authpkg.ProfilesConfig, error) { return nil, nil }
+	if reused, err := enrichAuthLoginProfileFromHistory("cfg", &authpkg.TokenData{CorpID: "ding_same"}); reused || err != nil {
+		t.Fatalf("nil history registry = reused=%v err=%v", reused, err)
+	}
+
+	cfg := &authpkg.ProfilesConfig{
+		CurrentProfile: "ding_same:user_1",
+		OrgCurrentProfiles: map[string]string{
+			"ding_same": "ding_same:user_2",
+		},
+		Profiles: []authpkg.Profile{
+			{CorpID: "ding_same", CorpName: "Same Corp", UserID: "user_1", UserName: "First"},
+			{CorpID: "ding_same", CorpName: "Same Corp", UserID: "user_2", UserName: "Second"},
+		},
+	}
+	authLoadProfiles = func(string) (*authpkg.ProfilesConfig, error) { return cfg, nil }
+
+	explicitData := &authpkg.TokenData{CorpID: "ding_same"}
+	reused, err := enrichAuthLoginProfileFromHistory("cfg", explicitData, authLoginHistoryHint{Selector: "ding_same:user_1", Explicit: true})
+	if err != nil || !reused || explicitData.UserID != "" ||
+		explicitData.LegacyOrgScopedProfile != "ding_same:user_1" ||
+		explicitData.CorpName != "Same Corp" || explicitData.UserName != "First" {
+		t.Fatalf("explicit history selection = %#v, reused=%v err=%v", explicitData, reused, err)
+	}
+	mismatchedHintData := &authpkg.TokenData{CorpID: "ding_same"}
+	reused, err = enrichAuthLoginProfileFromHistory("cfg", mismatchedHintData, authLoginHistoryHint{Selector: "ding_other:user_9", Explicit: true})
+	if err != nil || reused || mismatchedHintData.UserID != "" {
+		t.Fatalf("cross-corp explicit hint reused another identity: %#v, reused=%v err=%v", mismatchedHintData, reused, err)
+	}
+	orgCurrentData := &authpkg.TokenData{CorpID: "ding_same"}
+	reused, err = enrichAuthLoginProfileFromHistory("cfg", orgCurrentData)
+	if err != nil || reused || orgCurrentData.UserID != "" {
+		t.Fatalf("implicit multi-account org-current was treated as identity proof: %#v, reused=%v err=%v", orgCurrentData, reused, err)
+	}
+
+	cfg.Profiles = []authpkg.Profile{cfg.Profiles[1]}
+	soleData := &authpkg.TokenData{CorpID: "ding_same"}
+	reused, err = enrichAuthLoginProfileFromHistory("cfg", soleData)
+	if err != nil || !reused || soleData.UserID != "" ||
+		soleData.CorpName != "Same Corp" || soleData.UserName != "Second" {
+		t.Fatalf("sole history selection = %#v, reused=%v err=%v", soleData, reused, err)
+	}
+
+	cfg.Profiles = []authpkg.Profile{
+		{CorpID: "ding_same", CorpName: "Same Corp", UserID: "user_1", UserName: "First"},
+		{CorpID: "ding_same", CorpName: "Same Corp", UserID: "user_2", UserName: "Second"},
+	}
+	cfg.OrgCurrentProfiles = nil
+	currentData := &authpkg.TokenData{CorpID: "ding_same"}
+	reused, err = enrichAuthLoginProfileFromHistory("cfg", currentData)
+	if err != nil || reused || currentData.UserID != "" {
+		t.Fatalf("implicit multi-account current was treated as identity proof: %#v, reused=%v err=%v", currentData, reused, err)
+	}
+
+	cfg.CurrentProfile = ""
+	ambiguousData := &authpkg.TokenData{CorpID: "ding_same"}
+	reused, err = enrichAuthLoginProfileFromHistory("cfg", ambiguousData)
+	if err != nil || reused || ambiguousData.UserID != "" {
+		t.Fatalf("ambiguous history selection = %#v, reused=%v err=%v", ambiguousData, reused, err)
+	}
+
+	cfg.Profiles = []authpkg.Profile{{
+		Name: "external-worker", CorpID: "ding_same", CorpName: "Legacy Corp", UserName: "Legacy Worker",
+	}}
+	blankData := &authpkg.TokenData{CorpID: "ding_same"}
+	reused, err = enrichAuthLoginProfileFromHistory("cfg", blankData, authLoginHistoryHint{Selector: "external-worker", Explicit: true})
+	if err != nil || !reused || blankData.UserID != "" || blankData.LegacyOrgScopedProfile != "external-worker" || blankData.CorpName != "Legacy Corp" || blankData.UserName != "Legacy Worker" {
+		t.Fatalf("blank-userId history selection = %#v, reused=%v err=%v", blankData, reused, err)
+	}
+	contactBlankData := &authpkg.TokenData{CorpID: "ding_same", AccessToken: "new-token"}
+	if err := enrichAuthLoginProfileFromContact(
+		context.Background(),
+		"cfg",
+		&authCoverageCaller{err: errors.New("contact unavailable")},
+		contactBlankData,
+		authLoginHistoryHint{Selector: "external-worker", Explicit: true},
+	); err != nil {
+		t.Fatalf("blank-userId history must keep contact best effort: %v", err)
+	}
+	if contactBlankData.LegacyOrgScopedProfile != "external-worker" {
+		t.Fatalf("blank-userId contact fallback did not authorize the historical organization slot: %#v", contactBlankData)
+	}
+
+	profiles := []*authpkg.Profile{
+		nil,
+		{Name: "duplicate", CorpID: "ding_same", UserID: "user_1"},
+		{Name: "duplicate", CorpID: "ding_same", UserID: "user_2"},
+	}
+	for _, tc := range []struct {
+		name     string
+		selector string
+		profiles []*authpkg.Profile
+		want     *authpkg.Profile
+	}{
+		{name: "empty selector", selector: "", profiles: profiles},
+		{name: "missing exact identity", selector: "ding_same:missing", profiles: profiles},
+		{name: "duplicate name", selector: "duplicate", profiles: profiles},
+		{name: "unmatched name", selector: "not-found", profiles: profiles},
+		{name: "sole organization selector", selector: "ding_same", profiles: profiles[1:2], want: profiles[1]},
+	} {
+		t.Run("selector "+tc.name, func(t *testing.T) {
+			if got := historicalProfileForSelector("ding_same", tc.selector, tc.profiles); got != tc.want {
+				t.Fatalf("historicalProfileForSelector(%q) = %#v, want %#v", tc.selector, got, tc.want)
+			}
+		})
+	}
+}
+
 func TestCrossPlatformCoverageAuthCoverageDefaultSeamClosures(t *testing.T) {
 	ctx, cancel := context.WithCancel(context.Background())
 	cancel()
@@ -748,7 +1337,7 @@ func TestCrossPlatformCoverageAuthCoveragePortableExchangeAndReset(t *testing.T)
 	if err := importCmd.RunE(badForce, nil); err == nil {
 		t.Fatal("invalid force flag should fail")
 	}
-	_, out, _ = authCoverageRoot(importCmd, "table", false)
+	_, _, _ = authCoverageRoot(importCmd, "table", false)
 	if err := importCmd.RunE(importCmd, nil); err == nil {
 		t.Fatal("missing input should fail")
 	}

internal/app/auth_command_test.go

@@ -195,6 +195,15 @@ func TestCrossPlatformCoverageAuthImportRejectsWindowsDPAPIBackend(t *testing.T)
 	}
 	t.Setenv("DWS_CONFIG_DIR", configDir)
 	t.Setenv(keychain.StorageDirEnv, keychainDir)
+	// Windows stores credentials in HKCU rather than StorageDirEnv. Use a
+	// fresh registry namespace so this zero-state assertion cannot inherit a
+	// token from an earlier test in the same package binary.
+	t.Setenv(keychain.TestNamespaceEnv, root)
+	t.Cleanup(func() {
+		if err := keychain.RemoveAuthTokenEntries(keychain.Service); err != nil {
+			t.Errorf("clean import guard keychain fixture: %v", err)
+		}
+	})
 
 	importCmd := NewRootCommand()
 	importCmd.SetOut(&bytes.Buffer{})

internal/app/auth_legacy_platform_coverage_test.go

@@ -0,0 +1,307 @@
+// Copyright 2026 Alibaba Group
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package app
+
+import (
+	"context"
+	"errors"
+	"io"
+	"strings"
+	"testing"
+	"time"
+
+	authpkg "github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/auth"
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/event/personal"
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/executor"
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/pat"
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/pkg/edition"
+)
+
+func TestCrossPlatformCoverageAuthLoginUsesStableBlankProfileForPostLoginAuthorization(t *testing.T) {
+	t.Setenv("DWS_CONFIG_DIR", t.TempDir())
+	oldOAuth := authOAuthLogin
+	oldLoadProfiles := authLoadProfiles
+	oldRecommend := authRunLoginRecommend
+	oldInteractive := authLoginInteractiveTerminal
+	oldResolve := authResolveProfile
+	t.Cleanup(func() {
+		authOAuthLogin = oldOAuth
+		authLoadProfiles = oldLoadProfiles
+		authRunLoginRecommend = oldRecommend
+		authLoginInteractiveTerminal = oldInteractive
+		authResolveProfile = oldResolve
+	})
+
+	const corpID = "corp_post_login_blank"
+	cfg := &authpkg.ProfilesConfig{Profiles: []authpkg.Profile{
+		{Name: "Fixture Organization", CorpID: corpID, CorpName: "Fixture Organization"},
+		{Name: "Exact Fixture", CorpID: corpID, CorpName: "Fixture Organization", UserID: "identity_exact"},
+	}}
+	wantSelector := authpkg.ProfileSelectionSelector(cfg.Profiles[0], cfg)
+	if wantSelector == "" || wantSelector == corpID {
+		t.Fatalf("blank selector = %q, want a stable account selector", wantSelector)
+	}
+	authResolveProfile = func(string, string) (*authpkg.Profile, error) {
+		return nil, errors.New("no implicit profile")
+	}
+	authLoadProfiles = func(string) (*authpkg.ProfilesConfig, error) { return cfg, nil }
+	authOAuthLogin = func(*authpkg.OAuthProvider, context.Context, bool) (*authpkg.TokenData, error) {
+		return &authpkg.TokenData{
+			AccessToken: "new-access",
+			ExpiresAt:   time.Now().Add(time.Hour),
+			CorpID:      corpID,
+		}, nil
+	}
+	authLoginInteractiveTerminal = func() bool { return false }
+	seenSelector := ""
+	authRunLoginRecommend = func(context.Context, edition.ToolCaller, io.Writer, pat.LoginRecommendOptions) error {
+		seenSelector = authpkg.RuntimeProfile()
+		return nil
+	}
+	if _, _, err := authCoverageRunLogin(t, nil, "table", true, map[string]string{"recommend": "true"}); err != nil {
+		t.Fatalf("blank-profile login error = %v", err)
+	}
+	if seenSelector != wantSelector {
+		t.Fatalf("post-login runtime selector = %q, want %q", seenSelector, wantSelector)
+	}
+}
+
+func TestCrossPlatformCoverageAuthStatusAndLogoutPreserveExactSelectors(t *testing.T) {
+	t.Run("status canonicalizes a known identity", func(t *testing.T) {
+		configDir := t.TempDir()
+		t.Setenv("DWS_CONFIG_DIR", configDir)
+		const exactSelector = "corp_status_fixture:identity_status_fixture"
+		if err := authpkg.SaveProfiles(configDir, &authpkg.ProfilesConfig{
+			Version: 2,
+			Profiles: []authpkg.Profile{{
+				Name:   "Status Fixture",
+				CorpID: "corp_status_fixture",
+				UserID: "identity_status_fixture",
+			}},
+		}); err != nil {
+			t.Fatalf("SaveProfiles() error = %v", err)
+		}
+
+		oldStatus := authOAuthStatus
+		t.Cleanup(func() { authOAuthStatus = oldStatus })
+		seenSelector := ""
+		authOAuthStatus = func(*authpkg.OAuthProvider) (*authpkg.TokenData, error) {
+			seenSelector = authpkg.RuntimeProfile()
+			return &authpkg.TokenData{
+				AccessToken: "access",
+				ExpiresAt:   time.Now().Add(time.Hour),
+				CorpID:      "corp_status_fixture",
+				UserID:      "identity_status_fixture",
+			}, nil
+		}
+		cmd := newAuthStatusCommand()
+		_, _, _ = authCoverageRoot(cmd, "table", false)
+		if err := cmd.Flags().Set("profile", " Status Fixture "); err != nil {
+			t.Fatal(err)
+		}
+		if err := cmd.RunE(cmd, nil); err != nil {
+			t.Fatalf("auth status error = %v", err)
+		}
+		if seenSelector != exactSelector {
+			t.Fatalf("status runtime selector = %q, want %q", seenSelector, exactSelector)
+		}
+	})
+
+	t.Run("logout keeps a blank local selector", func(t *testing.T) {
+		oldResolve := authResolveProfileDeletion
+		oldLoad := authLoadTokenForProfile
+		oldRevoke := authRevokeTokenForData
+		oldDelete := authDeleteProfileToken
+		t.Cleanup(func() {
+			authResolveProfileDeletion = oldResolve
+			authLoadTokenForProfile = oldLoad
+			authRevokeTokenForData = oldRevoke
+			authDeleteProfileToken = oldDelete
+		})
+
+		const selector = "legacy-external-worker"
+		authResolveProfileDeletion = func(string, string) (*authpkg.Profile, bool, error) {
+			return &authpkg.Profile{CorpID: "corp_logout_blank"}, true, nil
+		}
+		loadedSelector := ""
+		authLoadTokenForProfile = func(_ string, got string) (*authpkg.TokenData, error) {
+			loadedSelector = got
+			return &authpkg.TokenData{CorpID: "corp_logout_blank"}, nil
+		}
+		authRevokeTokenForData = func(context.Context, *authpkg.TokenData) error { return nil }
+		deletedSelector := ""
+		authDeleteProfileToken = func(_ string, got string) error {
+			deletedSelector = got
+			return nil
+		}
+
+		if err := logoutOneProfile(nil, context.Background(), "cfg", "  "+selector+"  "); err != nil {
+			t.Fatalf("logoutOneProfile() error = %v", err)
+		}
+		if loadedSelector != selector || deletedSelector != selector {
+			t.Fatalf("blank logout selectors = load %q delete %q, want %q", loadedSelector, deletedSelector, selector)
+		}
+	})
+}
+
+func TestCrossPlatformCoverageAuthHistorySelectorRemainingBranches(t *testing.T) {
+	if got := authLoginHistorySelector("cfg", nil); got != "" {
+		t.Fatalf("nil history selector = %q", got)
+	}
+
+	oldLoad := authLoadProfiles
+	t.Cleanup(func() { authLoadProfiles = oldLoad })
+	authLoadProfiles = func(string) (*authpkg.ProfilesConfig, error) {
+		return nil, errors.New("profiles unavailable")
+	}
+	profile := &authpkg.Profile{CorpID: "corp_history", UserID: "identity_history"}
+	if got := authLoginHistorySelector("cfg", profile); got != "corp_history:identity_history" {
+		t.Fatalf("history selector fallback = %q", got)
+	}
+
+	duplicateA := &authpkg.Profile{CorpID: "corp_history", UserID: "duplicate_identity"}
+	duplicateB := &authpkg.Profile{CorpID: "corp_history", UserID: "duplicate_identity"}
+	if got := historicalProfileForSelector(
+		"corp_history",
+		"corp_history:duplicate_identity",
+		[]*authpkg.Profile{duplicateA, duplicateB},
+	); got != nil {
+		t.Fatalf("duplicate stable identity selected %#v", got)
+	}
+
+	// Whitespace keeps the raw selector from matching the stable string while
+	// ParseIdentitySelector still resolves its components.
+	exactFallback := &authpkg.Profile{CorpID: "corp_history", UserID: "fallback_identity"}
+	if got := historicalProfileForSelector(
+		"corp_history",
+		"corp_history : fallback_identity",
+		[]*authpkg.Profile{exactFallback},
+	); got != exactFallback {
+		t.Fatalf("exact history fallback = %#v, want %#v", got, exactFallback)
+	}
+}
+
+func TestCrossPlatformCoverageProfileSwitchLegacyBlankAndNormalizedIdentityPointers(t *testing.T) {
+	t.Run("one legacy blank name", func(t *testing.T) {
+		profiles := []authpkg.Profile{
+			{Name: "Fixture Organization", CorpID: "corp_profile_fixture", CorpName: "Fixture Organization"},
+			{Name: "Exact Fixture", CorpID: "corp_profile_fixture", CorpName: "Fixture Organization", UserID: "identity_exact"},
+		}
+		cfg := &authpkg.ProfilesConfig{Profiles: profiles}
+		if got := profileSwitchProfileIndex(profiles, "Fixture Organization", cfg); got != 0 {
+			t.Fatalf("legacy blank profile index = %d, want 0", got)
+		}
+	})
+
+	t.Run("duplicate legacy names fall through to blank-name compatibility", func(t *testing.T) {
+		profiles := []authpkg.Profile{
+			{Name: "duplicate-legacy", CorpID: "corp_profile_fixture"},
+			{Name: "duplicate-legacy", CorpID: "corp_profile_fixture"},
+		}
+		cfg := &authpkg.ProfilesConfig{Profiles: profiles}
+		if got := profileSwitchProfileIndex(profiles, "duplicate-legacy", cfg); got != 0 {
+			t.Fatalf("duplicate legacy fallback index = %d, want 0", got)
+		}
+	})
+
+	t.Run("normalized exact identity", func(t *testing.T) {
+		profiles := []authpkg.Profile{{CorpID: "corp_profile_fixture", UserID: "identity_exact"}}
+		cfg := &authpkg.ProfilesConfig{Profiles: profiles}
+		if got := profileSwitchProfileIndex(profiles, "corp_profile_fixture : identity_exact", cfg); got != 0 {
+			t.Fatalf("normalized exact profile index = %d, want 0", got)
+		}
+		if got := profileSwitchProfileIndex(profiles, "corp_profile_fixture : missing", cfg); got != -1 {
+			t.Fatalf("missing normalized exact profile index = %d, want -1", got)
+		}
+	})
+}
+
+func TestCrossPlatformCoverageRuntimeRunnerPreservesBlankSelectorInSingleAndMultiRuns(t *testing.T) {
+	exact := authLogoutTestToken("corp_runner_blank")
+	exact.UserID = "identity_exact_runner"
+	other := authLogoutTestToken("corp_runner_other")
+	configDir := setupAuthLogoutProfiles(t, exact, other)
+	blank := authLogoutTestToken("corp_runner_blank")
+	blank.AccessToken = "access-unresolved-runner"
+	blank.RefreshToken = "refresh-unresolved-runner"
+	blank.UserID = ""
+	blank.UserName = ""
+	if err := authpkg.SaveTokenData(configDir, blank); err != nil {
+		t.Fatalf("SaveTokenData(blank) error = %v", err)
+	}
+	cfg, err := authpkg.LoadProfiles(configDir)
+	if err != nil {
+		t.Fatalf("LoadProfiles() error = %v", err)
+	}
+	blankSelector := ""
+	for _, profile := range cfg.Profiles {
+		if profile.CorpID == blank.CorpID && profile.UserID == "" {
+			blankSelector = authpkg.ProfileSelectionSelector(profile, cfg)
+			break
+		}
+	}
+	if blankSelector == "" || blankSelector == blank.CorpID {
+		t.Fatalf("blank runner selector = %q, want exact local selector", blankSelector)
+	}
+
+	runner := &runtimeRunner{fallback: multiProfileFallbackRunner{}}
+	invocation := executor.Invocation{
+		Kind:             "helper_invocation",
+		CanonicalProduct: "contact",
+		Tool:             "get_current_user_profile",
+	}
+	authpkg.SetRuntimeProfile(blankSelector)
+	result, err := runner.Run(context.Background(), invocation)
+	if err != nil {
+		t.Fatalf("single blank Run() error = %v", err)
+	}
+	content := result.Response["content"].(map[string]any)
+	if got := content["runtimeProfile"]; got != blankSelector {
+		t.Fatalf("single blank runtime profile = %#v, want %q", got, blankSelector)
+	}
+	if got := authpkg.RuntimeProfile(); got != blankSelector {
+		t.Fatalf("single blank runtime restoration = %q, want %q", got, blankSelector)
+	}
+
+	authpkg.SetRuntimeProfile(blankSelector + ",corp_runner_other")
+	result, err = runner.Run(context.Background(), invocation)
+	if err != nil {
+		t.Fatalf("multi blank Run() error = %v", err)
+	}
+	entries := result.Response["content"].(map[string]any)["profiles"].([]any)
+	if len(entries) != 2 {
+		t.Fatalf("multi blank profiles = %#v, want two", entries)
+	}
+	first := entries[0].(map[string]any)
+	if first["selector"] != blankSelector || first["profile"] != blankSelector || first["userId"] != "" {
+		t.Fatalf("multi blank first entry = %#v", first)
+	}
+}
+
+func TestCrossPlatformCoveragePersonalBusSelectorCanonicalFallback(t *testing.T) {
+	authpkg.SetRuntimeProfile("")
+	t.Cleanup(func() { authpkg.SetRuntimeProfile("") })
+	identity := personal.Identity{
+		CorpID:   "corp_event_fallback",
+		UserID:   "identity_event_fallback",
+		SourceID: "open",
+	}
+	if got := personalBusProfileSelector(t.TempDir(), identity); got != "corp_event_fallback:identity_event_fallback" {
+		t.Fatalf("personal bus fallback selector = %q", got)
+	}
+	args := personalBusSpawnArgs(identity, "", "", "   ")
+	if got := strings.Join(args, " "); !strings.Contains(got, "--profile corp_event_fallback:identity_event_fallback") {
+		t.Fatalf("personal bus default profile args = %q", got)
+	}
+}

internal/app/auth_login_persistence_boundary_test.go

@@ -0,0 +1,107 @@
+// Copyright 2026 Alibaba Group
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+// http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package app
+
+import (
+	"strings"
+	"testing"
+	"time"
+
+	authpkg "github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/auth"
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/keychain"
+)
+
+func TestPATFreshAuthorizationSaveUsesLoginIsolationBoundary(t *testing.T) {
+	configDir := t.TempDir()
+	t.Setenv(keychain.DisableKeychainEnv, "1")
+	t.Setenv(keychain.StorageDirEnv, t.TempDir())
+	const (
+		corpID = "corp_pat_login_boundary"
+		userID = "exact-user"
+	)
+	cfg := &authpkg.ProfilesConfig{
+		Version: 2,
+		Profiles: []authpkg.Profile{
+			{Name: "External Account", CorpID: corpID, CorpName: "PAT Boundary Organization"},
+			{Name: "Exact Account", CorpID: corpID, CorpName: "PAT Boundary Organization", UserID: userID},
+		},
+	}
+	blankSelector := authpkg.ProfileSelectionSelector(cfg.Profiles[0], cfg)
+	cfg.CurrentProfile = blankSelector
+	cfg.PrimaryProfile = blankSelector
+	if err := authpkg.SaveProfiles(configDir, cfg); err != nil {
+		t.Fatalf("SaveProfiles() error = %v", err)
+	}
+	blank := &authpkg.TokenData{AccessToken: "existing-unresolved", CorpID: corpID, CorpName: "PAT Boundary Organization"}
+	exact := &authpkg.TokenData{AccessToken: "existing-exact", CorpID: corpID, CorpName: "PAT Boundary Organization", UserID: userID}
+	if err := authpkg.SaveTokenDataKeychainForCorpID(corpID, blank); err != nil {
+		t.Fatalf("save unresolved token: %v", err)
+	}
+	if err := authpkg.SaveTokenDataKeychainForIdentity(corpID, userID, exact); err != nil {
+		t.Fatalf("save exact token: %v", err)
+	}
+	previousRuntimeProfile := authpkg.RuntimeProfile()
+	authpkg.SetRuntimeProfile("")
+	t.Cleanup(func() { authpkg.SetRuntimeProfile(previousRuntimeProfile) })
+
+	fresh := &authpkg.TokenData{AccessToken: "pat-fresh-unknown", CorpID: corpID, CorpName: "PAT Boundary Organization"}
+	err := patSaveTokenData(configDir, fresh)
+	if err == nil || !strings.Contains(err.Error(), "fresh UID-less token") {
+		t.Fatalf("patSaveTokenData() error = %v, want unresolved-sibling protection", err)
+	}
+	persisted, loadErr := authpkg.LoadTokenDataKeychainForCorpID(corpID)
+	if loadErr != nil || persisted.AccessToken != blank.AccessToken || persisted.UserID != "" {
+		t.Fatalf("PAT save changed unresolved sibling: token=%#v err=%v", persisted, loadErr)
+	}
+}
+
+func TestManualLoginSaveRepairsHalfMigratedGlobalBeforeOverwrite(t *testing.T) {
+	configDir := t.TempDir()
+	t.Setenv(keychain.DisableKeychainEnv, "1")
+	t.Setenv(keychain.StorageDirEnv, t.TempDir())
+	const (
+		corpID = "corp_manual_login_boundary"
+		userID = "legacy-user"
+	)
+	selector := corpID + ":" + userID
+	if err := authpkg.SaveProfiles(configDir, &authpkg.ProfilesConfig{
+		Version:        2,
+		CurrentProfile: selector,
+		Profiles: []authpkg.Profile{{
+			Name: "Legacy Exact Account", CorpID: corpID, CorpName: "Manual Boundary Organization", UserID: userID,
+		}},
+	}); err != nil {
+		t.Fatalf("SaveProfiles() error = %v", err)
+	}
+	legacy := &authpkg.TokenData{AccessToken: "only-legacy-copy", CorpID: corpID, CorpName: "Manual Boundary Organization"}
+	if err := authpkg.SaveTokenDataKeychain(legacy); err != nil {
+		t.Fatalf("save half-migrated global: %v", err)
+	}
+	manual := &authpkg.TokenData{AccessToken: "manual-default", ExpiresAt: time.Now().Add(time.Hour)}
+	if err := authSaveTokenData(configDir, manual); err != nil {
+		t.Fatalf("authSaveTokenData(manual) error = %v", err)
+	}
+	org, err := authpkg.LoadTokenDataKeychainForCorpID(corpID)
+	if err != nil || org.AccessToken != legacy.AccessToken || org.UserID != "" {
+		t.Fatalf("organization repair = %#v, %v", org, err)
+	}
+	identity, err := authpkg.LoadTokenDataKeychainForIdentity(corpID, userID)
+	if err != nil || identity.AccessToken != legacy.AccessToken || identity.UserID != userID {
+		t.Fatalf("identity repair = %#v, %v", identity, err)
+	}
+	global, err := authpkg.LoadTokenDataKeychain()
+	if err != nil || global.AccessToken != manual.AccessToken || global.CorpID != "" {
+		t.Fatalf("manual global = %#v, %v", global, err)
+	}
+}

internal/app/blank_profile_selector_regression_test.go

@@ -0,0 +1,144 @@
+// Copyright 2026 Alibaba Group
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package app
+
+import (
+	"testing"
+
+	authpkg "github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/auth"
+)
+
+func blankProfileSelectorAppFixture(blankName, corpName string) *authpkg.ProfilesConfig {
+	const (
+		corpID      = "corp_selector_fixture"
+		exactUserID = "identity_exact_fixture"
+	)
+	exactSelector := corpID + ":" + exactUserID
+	cfg := &authpkg.ProfilesConfig{
+		Version:         2,
+		PrimaryProfile:  exactSelector,
+		PreviousProfile: exactSelector,
+		OrgCurrentProfiles: map[string]string{
+			corpID: exactSelector,
+		},
+		Profiles: []authpkg.Profile{
+			{
+				Name:     "Exact Fixture Account",
+				CorpID:   corpID,
+				CorpName: corpName,
+				UserID:   exactUserID,
+				UserName: "Exact Fixture Account",
+				Status:   authpkg.ProfileStatusActive,
+			},
+			{
+				Name:     blankName,
+				CorpID:   corpID,
+				CorpName: corpName,
+				Status:   authpkg.ProfileStatusActive,
+			},
+		},
+	}
+	cfg.CurrentProfile = authpkg.ProfileSelectionSelector(cfg.Profiles[1], cfg)
+	return cfg
+}
+
+func captureProfileListSelectors(t *testing.T, cfg *authpkg.ProfilesConfig) ([]string, []profileView) {
+	t.Helper()
+	originalLoadToken := profileLoadTokenData
+	selectors := make([]string, 0, len(cfg.Profiles))
+	profileLoadTokenData = func(_ string, selector string) (*authpkg.TokenData, error) {
+		selectors = append(selectors, selector)
+		return nil, authpkg.ErrTokenDataNotFound
+	}
+	t.Cleanup(func() { profileLoadTokenData = originalLoadToken })
+	views := profileViews("unused-config-dir", cfg)
+	return selectors, views
+}
+
+func TestCrossPlatformCoverageBlankProfileNameMatchingCorpNameRoundTripsThroughListAndTUI(t *testing.T) {
+	cfg := blankProfileSelectorAppFixture("Fixture Organization", "Fixture Organization")
+	blank := cfg.Profiles[1]
+	blankSelector := authpkg.ProfileSelectionSelector(blank, cfg)
+
+	if blankSelector == blank.Name || blankSelector == blank.CorpID {
+		t.Fatalf("unsafe blank selector = %q, want reserved exact selector", blankSelector)
+	}
+	if got := profileCLISelector(blank, cfg); got != blankSelector {
+		t.Errorf("profileCLISelector(blank) = %q, want %q", got, blankSelector)
+	}
+	if got := profileSwitchProfileIndex(cfg.Profiles, cfg.CurrentProfile, cfg); got != 1 {
+		t.Errorf("profileSwitchProfileIndex(blank current) = %d, want 1", got)
+	}
+	model := newProfileSwitchTUIModel(cfg, cfg.CurrentProfile)
+	if model.selected != 1 {
+		t.Errorf("TUI selected index = %d, want blank profile index 1", model.selected)
+	}
+	if got := model.selectedCorpID(); got != blankSelector {
+		t.Errorf("TUI selected selector = %q, want %q", got, blankSelector)
+	}
+
+	selectors, views := captureProfileListSelectors(t, cfg)
+	if len(selectors) != 2 || selectors[0] != cfg.PreviousProfile || selectors[1] != blankSelector {
+		t.Errorf("profile list token selectors = %#v, want exact then %q", selectors, blankSelector)
+	}
+	if len(views) != 2 {
+		t.Fatalf("profile list views = %#v, want two entries", views)
+	}
+	if views[0].IsCurrent {
+		t.Error("exact account should not be marked current when blank local selector is current")
+	}
+	if views[1].Profile != blankSelector || !views[1].IsCurrent {
+		t.Errorf("blank list view = %#v, want local selector marked current", views[1])
+	}
+}
+
+func TestCrossPlatformCoverageBlankProfileNameContainingColonWinsOverIdentityParsingInListAndTUI(t *testing.T) {
+	cfg := blankProfileSelectorAppFixture("legacy:outsourced", "Fixture Organization")
+	blank := cfg.Profiles[1]
+	blankSelector := authpkg.ProfileSelectionSelector(blank, cfg)
+
+	if blankSelector == blank.Name {
+		t.Fatalf("colon-containing name leaked as selector %q", blankSelector)
+	}
+	if _, _, parsedAsIdentity := authpkg.ParseIdentitySelector(blankSelector); parsedAsIdentity {
+		t.Fatalf("stable blank selector %q was parsed as an identity", blankSelector)
+	}
+	if got := profileCLISelector(blank, cfg); got != blankSelector {
+		t.Errorf("profileCLISelector(colon blank) = %q, want %q", got, blankSelector)
+	}
+	if got := profileSwitchProfileIndex(cfg.Profiles, cfg.CurrentProfile, cfg); got != 1 {
+		t.Errorf("profileSwitchProfileIndex(colon blank current) = %d, want 1", got)
+	}
+	model := newProfileSwitchTUIModel(cfg, cfg.CurrentProfile)
+	if model.selected != 1 {
+		t.Errorf("TUI selected index = %d, want colon-name blank profile index 1", model.selected)
+	}
+	if got := model.selectedCorpID(); got != blankSelector {
+		t.Errorf("TUI selected selector = %q, want %q", got, blankSelector)
+	}
+
+	selectors, views := captureProfileListSelectors(t, cfg)
+	if len(selectors) != 2 || selectors[0] != cfg.PreviousProfile || selectors[1] != blankSelector {
+		t.Errorf("profile list token selectors = %#v, want exact then %q", selectors, blankSelector)
+	}
+	if len(views) != 2 {
+		t.Fatalf("profile list views = %#v, want two entries", views)
+	}
+	if views[0].IsCurrent {
+		t.Error("exact account should not be marked current when colon-name blank selector is current")
+	}
+	if views[1].Profile != blankSelector || !views[1].IsCurrent {
+		t.Errorf("colon-name blank list view = %#v, want local selector marked current", views[1])
+	}
+}

internal/app/event_personal_command.go

@@ -232,6 +232,7 @@ func runPersonalEventConsume(c *cobra.Command, opts personalConsumeOptions) erro
 	editionName := editionNameOrDefault()
 	workDir := eventWorkDir(configDir, editionName, dwsevent.SourceKindPersonalStream, identityHash)
 	ipcEndpoint := defaultIPCEndpoint(workDir, editionName, dwsevent.SourceKindPersonalStream, identityHash)
+	spawnProfileSelector := personalBusProfileSelector(configDir, identity)
 
 	routes, err := consume.ParseRoutes(opts.Common.RoutesRaw)
 	if err != nil {
@@ -247,7 +248,7 @@ func runPersonalEventConsume(c *cobra.Command, opts personalConsumeOptions) erro
 			WorkDir:        workDir,
 			IPCEndpoint:    ipcEndpoint,
 			ClientID:       identity.ClientID,
-			SpawnExtraArgs: personalBusSpawnArgs(identity, opts.StreamTicketMode, personalEventStreamTicketURL(opts.StreamTicketURL, configDir)),
+			SpawnExtraArgs: personalBusSpawnArgs(identity, opts.StreamTicketMode, personalEventStreamTicketURL(opts.StreamTicketURL, configDir), spawnProfileSelector),
 			Compact:        opts.Common.Compact,
 			MaxEvents:      opts.Common.MaxEvents,
 			Duration:       opts.Common.Duration,
@@ -304,7 +305,7 @@ func runPersonalEventConsume(c *cobra.Command, opts personalConsumeOptions) erro
 		WorkDir:          workDir,
 		IPCEndpoint:      ipcEndpoint,
 		ClientID:         identity.ClientID,
-		SpawnExtraArgs:   personalBusSpawnArgs(identity, opts.StreamTicketMode, opts.StreamTicketURL),
+		SpawnExtraArgs:   personalBusSpawnArgs(identity, opts.StreamTicketMode, opts.StreamTicketURL, spawnProfileSelector),
 		Compact:          opts.Common.Compact,
 		MaxEvents:        opts.Common.MaxEvents,
 		Duration:         opts.Common.Duration,
@@ -869,7 +870,44 @@ func newPersonalStreamSource(ctx context.Context, opts personalStreamSourceOptio
 	})
 }
 
-func personalBusSpawnArgs(identity personal.Identity, ticketMode, ticketURL string) []string {
+func personalBusProfileSelector(configDir string, identity personal.Identity) string {
+	// The parent already resolved and loaded this selector. Preserve it before
+	// consulting identity metadata: personal event discovery can fill an empty
+	// token userId from runtime defaults, and that inferred value must not turn a
+	// historical unresolved account into a different exact same-corp account in
+	// the detached child.
+	if selector := strings.TrimSpace(authpkg.RuntimeProfile()); selector != "" {
+		return selector
+	}
+	if cfg, err := authpkg.LoadProfiles(configDir); err == nil && cfg != nil {
+		// With no explicit process-local override, LoadTokenData selected the
+		// persisted current profile. Prefer that selection over the enriched
+		// identity: $currentUserId may describe an exact same-corp account even
+		// though the token came from the historical unresolved profile.
+		currentSelector := strings.TrimSpace(cfg.CurrentProfile)
+		for i := range cfg.Profiles {
+			profile := cfg.Profiles[i]
+			selector := authpkg.ProfileSelectionSelector(profile, cfg)
+			if selector == currentSelector &&
+				(strings.TrimSpace(identity.CorpID) == "" || strings.TrimSpace(profile.CorpID) == strings.TrimSpace(identity.CorpID)) {
+				return selector
+			}
+		}
+		for i := range cfg.Profiles {
+			profile := cfg.Profiles[i]
+			if strings.TrimSpace(profile.CorpID) == strings.TrimSpace(identity.CorpID) &&
+				strings.TrimSpace(profile.UserID) == strings.TrimSpace(identity.UserID) {
+				return authpkg.ProfileSelectionSelector(profile, cfg)
+			}
+		}
+	}
+	return authpkg.ProfileSelector(authpkg.Profile{
+		CorpID: identity.CorpID,
+		UserID: identity.UserID,
+	})
+}
+
+func personalBusSpawnArgs(identity personal.Identity, ticketMode, ticketURL string, profileSelectors ...string) []string {
 	args := []string{
 		"--source-kind", string(dwsevent.SourceKindPersonalStream),
 		"--stream-source-id", identity.SourceID,
@@ -878,10 +916,11 @@ func personalBusSpawnArgs(identity personal.Identity, ticketMode, ticketURL stri
 	// credentials as the parent, including when one organization has multiple
 	// logged-in users.
 	if cid := strings.TrimSpace(identity.CorpID); cid != "" {
-		args = append(args, "--profile", authpkg.ProfileSelector(authpkg.Profile{
-			CorpID: identity.CorpID,
-			UserID: identity.UserID,
-		}))
+		profileSelector := authpkg.ProfileSelector(authpkg.Profile{CorpID: identity.CorpID, UserID: identity.UserID})
+		if len(profileSelectors) > 0 && strings.TrimSpace(profileSelectors[0]) != "" {
+			profileSelector = strings.TrimSpace(profileSelectors[0])
+		}
+		args = append(args, "--profile", profileSelector)
 	}
 	if strings.TrimSpace(ticketMode) != "" {
 		args = append(args, "--stream-ticket-mode", ticketMode)

internal/app/event_profile_selector_compat_test.go

@@ -0,0 +1,121 @@
+// Copyright 2026 Alibaba Group
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package app
+
+import (
+	"testing"
+
+	authpkg "github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/auth"
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/event/personal"
+)
+
+func TestPersonalBusProfileSelectorUsesDefaultBlankCurrentBeforeRuntimeEnrichedIdentity(t *testing.T) {
+	configDir, cfg, blankSelector, exactSelector := seedPersonalBusProfileSelectorConfig(t)
+	cfg.CurrentProfile = blankSelector
+	cfg.OrgCurrentProfiles = map[string]string{cfg.Profiles[0].CorpID: exactSelector}
+	if err := authpkg.SaveProfiles(configDir, cfg); err != nil {
+		t.Fatalf("SaveProfiles() error = %v", err)
+	}
+	authpkg.SetRuntimeProfile("")
+	t.Cleanup(func() { authpkg.SetRuntimeProfile("") })
+
+	identityAfterRuntimeEnrichment := personal.Identity{
+		CorpID: cfg.Profiles[0].CorpID,
+		UserID: cfg.Profiles[1].UserID,
+	}
+	if got := personalBusProfileSelector(configDir, identityAfterRuntimeEnrichment); got != blankSelector {
+		t.Fatalf("personalBusProfileSelector() = %q, want default blank selector %q", got, blankSelector)
+	}
+}
+
+func TestPersonalBusProfileSelectorUsesDefaultExactCurrent(t *testing.T) {
+	configDir, cfg, _, exactSelector := seedPersonalBusProfileSelectorConfig(t)
+	cfg.CurrentProfile = exactSelector
+	cfg.OrgCurrentProfiles = map[string]string{cfg.Profiles[0].CorpID: exactSelector}
+	if err := authpkg.SaveProfiles(configDir, cfg); err != nil {
+		t.Fatalf("SaveProfiles() error = %v", err)
+	}
+	authpkg.SetRuntimeProfile("")
+	t.Cleanup(func() { authpkg.SetRuntimeProfile("") })
+
+	identity := personal.Identity{CorpID: cfg.Profiles[1].CorpID, UserID: cfg.Profiles[1].UserID}
+	if got := personalBusProfileSelector(configDir, identity); got != exactSelector {
+		t.Fatalf("personalBusProfileSelector() = %q, want default exact selector %q", got, exactSelector)
+	}
+}
+
+func TestPersonalBusProfileSelectorPrefersExplicitRuntimeSelector(t *testing.T) {
+	configDir, cfg, blankSelector, _ := seedPersonalBusProfileSelectorConfig(t)
+	cfg.CurrentProfile = blankSelector
+	if err := authpkg.SaveProfiles(configDir, cfg); err != nil {
+		t.Fatalf("SaveProfiles() error = %v", err)
+	}
+
+	const explicitSelector = "corp_explicit:user_explicit"
+	authpkg.SetRuntimeProfile(explicitSelector)
+	t.Cleanup(func() { authpkg.SetRuntimeProfile("") })
+
+	identity := personal.Identity{CorpID: cfg.Profiles[1].CorpID, UserID: cfg.Profiles[1].UserID}
+	if got := personalBusProfileSelector(configDir, identity); got != explicitSelector {
+		t.Fatalf("personalBusProfileSelector() = %q, want explicit selector %q", got, explicitSelector)
+	}
+}
+
+func TestCrossPlatformCoveragePersonalBusProfileSelectorFallsBackToMatchingIdentity(t *testing.T) {
+	configDir, cfg, _, exactSelector := seedPersonalBusProfileSelectorConfig(t)
+	cfg.Profiles = append(cfg.Profiles, authpkg.Profile{
+		Name:     "Other Current",
+		CorpID:   "corp_event_other_fixture",
+		CorpName: "Other Fixture Organization",
+		UserID:   "identity_event_other_fixture",
+	})
+	cfg.CurrentProfile = authpkg.ProfileSelectionSelector(cfg.Profiles[2], cfg)
+	if err := authpkg.SaveProfiles(configDir, cfg); err != nil {
+		t.Fatalf("SaveProfiles() error = %v", err)
+	}
+	authpkg.SetRuntimeProfile("")
+	t.Cleanup(func() { authpkg.SetRuntimeProfile("") })
+
+	identity := personal.Identity{CorpID: cfg.Profiles[1].CorpID, UserID: cfg.Profiles[1].UserID}
+	if got := personalBusProfileSelector(configDir, identity); got != exactSelector {
+		t.Fatalf("personalBusProfileSelector() = %q, want identity fallback %q", got, exactSelector)
+	}
+}
+
+func seedPersonalBusProfileSelectorConfig(t *testing.T) (string, *authpkg.ProfilesConfig, string, string) {
+	t.Helper()
+	configDir := t.TempDir()
+	cfg := &authpkg.ProfilesConfig{
+		Version: 2,
+		Profiles: []authpkg.Profile{
+			{
+				Name:     "External Fixture",
+				CorpID:   "corp_event_current_fixture",
+				CorpName: "Fixture Organization",
+			},
+			{
+				Name:     "Exact Fixture",
+				CorpID:   "corp_event_current_fixture",
+				CorpName: "Fixture Organization",
+				UserID:   "identity_runtime_enriched_fixture",
+			},
+		},
+	}
+	blankSelector := authpkg.ProfileSelectionSelector(cfg.Profiles[0], cfg)
+	exactSelector := authpkg.ProfileSelectionSelector(cfg.Profiles[1], cfg)
+	if blankSelector == "" || blankSelector == cfg.Profiles[0].CorpID {
+		t.Fatalf("blank selector = %q, want stable account selector", blankSelector)
+	}
+	return configDir, cfg, blankSelector, exactSelector
+}

internal/app/event_registration_test.go

@@ -13,14 +13,18 @@ import (
 func TestEventCommandRemainsVisibleAsBuiltInPublicGroup(t *testing.T) {
 	root := &cobra.Command{Use: "dws"}
 	event := newEventCommand()
+	markdown := &cobra.Command{Use: "markdown"}
 	unregistered := &cobra.Command{Use: "unregistered", Run: func(*cobra.Command, []string) {}}
-	root.AddCommand(event, unregistered)
+	root.AddCommand(event, markdown, unregistered)
 
 	hideNonDirectRuntimeCommands(root)
 
 	if event.Hidden {
 		t.Fatal("built-in event command was hidden by the direct-runtime visibility filter")
 	}
+	if markdown.Hidden {
+		t.Fatal("locally routed markdown command was hidden by the direct-runtime visibility filter")
+	}
 	if !unregistered.Hidden {
 		t.Fatal("control command outside the built-in/direct-runtime sets remained visible")
 	}

internal/app/event_stdin_gate_test.go

@@ -17,7 +17,9 @@ import (
 	"testing"
 	"time"
 
+	authpkg "github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/auth"
 	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/event/personal"
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/keychain"
 )
 
 // A bounded run never arms the stdin-EOF watcher, regardless of stdin
@@ -63,3 +65,100 @@ func TestPersonalBusSpawnArgs_ForwardsProfile(t *testing.T) {
 		}
 	}
 }
+
+func TestCrossPlatformCoveragePersonalBusSpawnArgsPreservesReservedBlankProfile(t *testing.T) {
+	configDir := t.TempDir()
+	t.Setenv(keychain.DisableKeychainEnv, "1")
+	t.Setenv(keychain.StorageDirEnv, t.TempDir())
+	cfg := &authpkg.ProfilesConfig{
+		Version: 2,
+		OrgCurrentProfiles: map[string]string{
+			"corp_event_fixture": "corp_event_fixture:identity_exact_fixture",
+		},
+		Profiles: []authpkg.Profile{
+			{
+				Name:     "Fixture Organization",
+				CorpID:   "corp_event_fixture",
+				CorpName: "Fixture Organization",
+			},
+			{
+				Name:     "Exact Fixture Account",
+				CorpID:   "corp_event_fixture",
+				CorpName: "Fixture Organization",
+				UserID:   "identity_exact_fixture",
+			},
+		},
+	}
+	blankSelector := authpkg.ProfileSelectionSelector(cfg.Profiles[0], cfg)
+	cfg.PrimaryProfile = blankSelector
+	cfg.CurrentProfile = blankSelector
+	if err := authpkg.SaveProfiles(configDir, cfg); err != nil {
+		t.Fatalf("SaveProfiles() error = %v", err)
+	}
+	blankToken := &authpkg.TokenData{
+		AccessToken: "parent-blank-token",
+		CorpID:      "corp_event_fixture",
+		CorpName:    "Fixture Organization",
+	}
+	exactToken := &authpkg.TokenData{
+		AccessToken: "other-exact-token",
+		CorpID:      "corp_event_fixture",
+		CorpName:    "Fixture Organization",
+		UserID:      "identity_exact_fixture",
+	}
+	if err := authpkg.SaveTokenDataKeychainForCorpID(blankToken.CorpID, blankToken); err != nil {
+		t.Fatalf("save parent blank token: %v", err)
+	}
+	if err := authpkg.SaveTokenDataKeychainForIdentity(exactToken.CorpID, exactToken.UserID, exactToken); err != nil {
+		t.Fatalf("save other exact token: %v", err)
+	}
+
+	// Runtime identity enrichment points at the exact sibling, but the parent
+	// already loaded the persisted blank current profile.
+	identity := personal.Identity{
+		CorpID:   "corp_event_fixture",
+		UserID:   "identity_exact_fixture",
+		SourceID: "open",
+	}
+	selector := personalBusProfileSelector(configDir, identity)
+	want := blankSelector
+	if selector != want || selector == identity.CorpID {
+		t.Fatalf("personalBusProfileSelector(blank) = %q, want reserved %q", selector, want)
+	}
+	args := personalBusSpawnArgs(identity, "", "", selector)
+	forwardedSelector := ""
+	for i := 0; i+1 < len(args); i++ {
+		if args[i] == "--profile" && args[i+1] == want {
+			forwardedSelector = args[i+1]
+			break
+		}
+	}
+	if forwardedSelector == "" {
+		t.Fatalf("spawn args did not preserve reserved blank selector: %v", args)
+	}
+	parentToken, err := authpkg.LoadTokenDataForProfile(configDir, selector)
+	if err != nil {
+		t.Fatalf("load parent token: %v", err)
+	}
+	authpkg.SetRuntimeProfile(forwardedSelector)
+	t.Cleanup(func() { authpkg.SetRuntimeProfile("") })
+	childToken, err := authpkg.LoadTokenData(configDir)
+	if err != nil {
+		t.Fatalf("load detached child token: %v", err)
+	}
+	if parentToken.AccessToken != blankToken.AccessToken ||
+		childToken.AccessToken != parentToken.AccessToken ||
+		childToken.UserID != "" {
+		t.Fatalf("parent/child token drift: parent=%#v child=%#v", parentToken, childToken)
+	}
+
+	authpkg.SetRuntimeProfile(want)
+	inferredExact := personal.Identity{
+		CorpID:   "corp_event_fixture",
+		UserID:   "identity_exact_fixture",
+		SourceID: "open",
+	}
+	if got := personalBusProfileSelector(configDir, inferredExact); got != want {
+		t.Fatalf("runtime blank selector changed after inferred userId: got %q, want %q", got, want)
+	}
+}

internal/app/mcp_url_command.go

@@ -0,0 +1,108 @@
+// Copyright 2026 Alibaba Group
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package app
+
+import (
+	"encoding/json"
+	"fmt"
+	"strings"
+
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/cli"
+	apperrors "github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/errors"
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/output"
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/pkg/edition"
+	"github.com/spf13/cobra"
+)
+
+const (
+	mcpMetaServerID = "mcp-meta"
+	mcpMetaURLTool  = "get_mcp_server_url"
+)
+
+func newMCPURLGroup(caller edition.ToolCaller) *cobra.Command {
+	group := &cobra.Command{
+		Use:               "url",
+		Short:             "管理 MCP 服务连接地址",
+		Args:              cobra.NoArgs,
+		DisableAutoGenTag: true,
+		RunE: func(cmd *cobra.Command, _ []string) error {
+			return cmd.Help()
+		},
+	}
+	group.AddCommand(newMCPURLGetCommand(caller))
+	return group
+}
+
+func newMCPURLGetCommand(caller edition.ToolCaller) *cobra.Command {
+	cmd := &cobra.Command{
+		Use:   "get <mcpId>",
+		Short: "按 mcpId 获取 MCP 的 Streamable HTTP 服务地址",
+		Long: "输入 MCP 市场 mcpId，返回以当前用户和组织身份访问该 MCP 的 " +
+			"Streamable HTTP 服务地址。\n\n" +
+			"安全提示：返回的 mcpURL 和 mcpJSON 可能包含身份凭据，仅限个人使用，" +
+			"请勿分享到群聊、文档、邮件、代码仓库或日志。",
+		Example: "  dws mcp url get 2480\n" +
+			"  dws mcp url get 2480 --format json",
+		Args:              cobra.ExactArgs(1),
+		DisableAutoGenTag: true,
+		RunE: func(cmd *cobra.Command, args []string) error {
+			if caller == nil {
+				return fmt.Errorf("MCP tool caller is not configured")
+			}
+			mcpID := strings.TrimSpace(args[0])
+			if mcpID == "" {
+				return fmt.Errorf("mcpId 不能为空")
+			}
+
+			result, err := caller.CallTool(cmd.Context(), mcpMetaServerID, mcpMetaURLTool, map[string]any{
+				"mcpId": mcpID,
+			})
+			if err != nil {
+				return fmt.Errorf("获取 MCP 服务地址: %w", err)
+			}
+			return writeMCPURLResult(cmd, result)
+		},
+	}
+	cli.AnnotateRuntimePositionals(cmd, cli.RuntimeSchemaPositional{
+		Name:        "mcp_id",
+		Type:        "string",
+		Description: "钉钉 MCP 市场中的 mcpId",
+		Required:    true,
+		Index:       0,
+	})
+	return cmd
+}
+
+func writeMCPURLResult(cmd *cobra.Command, result *edition.ToolResult) error {
+	if result == nil {
+		return fmt.Errorf("MCP 元服务返回空结果")
+	}
+	// get_mcp_server_url returns one JSON document in its first non-empty text
+	// block. Other block types and trailing blocks are intentionally ignored.
+	for _, block := range result.Content {
+		if block.Type != "text" || strings.TrimSpace(block.Text) == "" {
+			continue
+		}
+		if err := apperrors.ClassifyMCPResponseText(block.Text); err != nil {
+			return err
+		}
+
+		var payload any
+		if err := json.Unmarshal([]byte(block.Text), &payload); err != nil {
+			return fmt.Errorf("MCP 元服务返回了无效 JSON: %w", err)
+		}
+		return output.WriteCommandPayload(cmd, payload, output.FormatJSON)
+	}
+	return fmt.Errorf("MCP 元服务返回空结果")
+}

internal/app/mcp_url_command_test.go

@@ -0,0 +1,194 @@
+// Copyright 2026 Alibaba Group
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package app
+
+import (
+	"bytes"
+	"context"
+	"encoding/json"
+	"errors"
+	"strings"
+	"testing"
+
+	apperrors "github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/errors"
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/pkg/edition"
+	"github.com/spf13/cobra"
+)
+
+type mcpURLTestCaller struct {
+	productID string
+	toolName  string
+	args      map[string]any
+	result    *edition.ToolResult
+	err       error
+}
+
+func (c *mcpURLTestCaller) CallTool(_ context.Context, productID, toolName string, args map[string]any) (*edition.ToolResult, error) {
+	c.productID = productID
+	c.toolName = toolName
+	c.args = args
+	return c.result, c.err
+}
+
+func (*mcpURLTestCaller) Format() string { return "json" }
+func (*mcpURLTestCaller) DryRun() bool   { return false }
+func (*mcpURLTestCaller) Fields() string { return "" }
+func (*mcpURLTestCaller) JQ() string     { return "" }
+
+func executeMCPURLCommand(t *testing.T, caller edition.ToolCaller, args ...string) (string, error) {
+	t.Helper()
+	root := &cobra.Command{Use: "mcp", SilenceErrors: true, SilenceUsage: true}
+	root.AddCommand(newMCPURLGroup(caller))
+	var out bytes.Buffer
+	root.SetOut(&out)
+	root.SetErr(&out)
+	root.SetArgs(args)
+	err := root.ExecuteContext(t.Context())
+	return out.String(), err
+}
+
+func TestMCPURLGetCallsMetaServerAndPreservesResponse(t *testing.T) {
+	const response = `{"result":{"mcpURL":"https://example.test/mcp?key=one&token=two","mcpJSON":{"transport":"streamable-http"},"name":"Example"}}`
+	caller := &mcpURLTestCaller{
+		result: &edition.ToolResult{Content: []edition.ContentBlock{{Type: "text", Text: response}}},
+	}
+
+	out, err := executeMCPURLCommand(t, caller, "url", "get", " 10043 ")
+	if err != nil {
+		t.Fatalf("execute mcp url get: %v", err)
+	}
+	if caller.productID != mcpMetaServerID {
+		t.Fatalf("productID = %q, want %q", caller.productID, mcpMetaServerID)
+	}
+	if caller.toolName != mcpMetaURLTool {
+		t.Fatalf("toolName = %q, want %q", caller.toolName, mcpMetaURLTool)
+	}
+	if got := caller.args["mcpId"]; got != "10043" {
+		t.Fatalf("mcpId = %#v, want %q", got, "10043")
+	}
+
+	var payload map[string]any
+	if err := json.Unmarshal([]byte(out), &payload); err != nil {
+		t.Fatalf("output is not JSON: %v\n%s", err, out)
+	}
+	result, ok := payload["result"].(map[string]any)
+	if !ok {
+		t.Fatalf("output result = %#v", payload["result"])
+	}
+	if got := result["mcpURL"]; got != "https://example.test/mcp?key=one&token=two" {
+		t.Fatalf("result.mcpURL = %#v", got)
+	}
+}
+
+func TestMCPURLGetRejectsBlankID(t *testing.T) {
+	_, err := executeMCPURLCommand(t, &mcpURLTestCaller{}, "url", "get", "   ")
+	if err == nil || !strings.Contains(err.Error(), "mcpId 不能为空") {
+		t.Fatalf("error = %v, want blank mcpId error", err)
+	}
+}
+
+func TestMCPURLGroupShowsHelp(t *testing.T) {
+	out, err := executeMCPURLCommand(t, nil, "url")
+	if err != nil {
+		t.Fatalf("execute mcp url: %v", err)
+	}
+	if !strings.Contains(out, "get") {
+		t.Fatalf("help output does not list get command:\n%s", out)
+	}
+}
+
+func TestMCPURLGetRejectsMissingCaller(t *testing.T) {
+	_, err := executeMCPURLCommand(t, nil, "url", "get", "10043")
+	if err == nil || !strings.Contains(err.Error(), "caller is not configured") {
+		t.Fatalf("error = %v, want missing caller error", err)
+	}
+}
+
+func TestMCPURLGetPropagatesCallError(t *testing.T) {
+	caller := &mcpURLTestCaller{err: errors.New("permission denied")}
+	_, err := executeMCPURLCommand(t, caller, "url", "get", "10043")
+	if err == nil || !strings.Contains(err.Error(), "permission denied") {
+		t.Fatalf("error = %v, want call error", err)
+	}
+}
+
+func TestMCPURLGetRejectsInvalidJSON(t *testing.T) {
+	caller := &mcpURLTestCaller{
+		result: &edition.ToolResult{Content: []edition.ContentBlock{{Type: "text", Text: "not-json"}}},
+	}
+	_, err := executeMCPURLCommand(t, caller, "url", "get", "10043")
+	if err == nil || !strings.Contains(err.Error(), "无效 JSON") {
+		t.Fatalf("error = %v, want invalid JSON error", err)
+	}
+}
+
+func TestMCPURLGetRejectsEmptyResults(t *testing.T) {
+	tests := []struct {
+		name   string
+		result *edition.ToolResult
+	}{
+		{name: "nil result"},
+		{
+			name: "no usable text content",
+			result: &edition.ToolResult{Content: []edition.ContentBlock{
+				{Type: "image", Text: "ignored"},
+				{Type: "text", Text: "   "},
+			}},
+		},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			caller := &mcpURLTestCaller{result: tt.result}
+			_, err := executeMCPURLCommand(t, caller, "url", "get", "10043")
+			if err == nil || !strings.Contains(err.Error(), "返回空结果") {
+				t.Fatalf("error = %v, want empty result error", err)
+			}
+		})
+	}
+}
+
+func TestMCPURLGetClassifiesBusinessError(t *testing.T) {
+	caller := &mcpURLTestCaller{
+		result: &edition.ToolResult{Content: []edition.ContentBlock{{
+			Type: "text",
+			Text: `{"success":false,"errorMsg":"搜索内容不能为空"}`,
+		}}},
+	}
+	_, err := executeMCPURLCommand(t, caller, "url", "get", "10043")
+	if err == nil {
+		t.Fatal("expected classified business error")
+	}
+	var typed *apperrors.Error
+	if !errors.As(err, &typed) || typed.Reason != "business_error" {
+		t.Fatalf("error = %#v, want classified business error", err)
+	}
+}
+
+func TestRootRegistersMCPURLGet(t *testing.T) {
+	root := NewRootCommand(t.Context())
+	mcp, _, err := root.Find([]string{"mcp"})
+	if err != nil {
+		t.Fatalf("find mcp: %v", err)
+	}
+	if mcp.Hidden {
+		t.Fatal("mcp command must be public when it contains reviewed public helpers")
+	}
+	cmd, _, err := root.Find([]string{"mcp", "url", "get"})
+	if err != nil {
+		t.Fatalf("find mcp url get: %v", err)
+	}
+	if got := cmd.CommandPath(); got != "dws mcp url get" {
+		t.Fatalf("command path = %q, want %q", got, "dws mcp url get")
+	}
+}

internal/app/multi_profile_runner_test.go

@@ -107,6 +107,45 @@ func TestRuntimeRunnerDeduplicatesByResolvedIdentityInSameCorp(t *testing.T) {
 	}
 }
 
+func TestCrossPlatformCoverageRuntimeRunnerDeduplicatesReservedAndOrganizationAliasesForBlankProfile(t *testing.T) {
+	exact := authLogoutTestToken("corp_blank_alias")
+	exact.UserID = "identity_exact_alias"
+	configDir := setupAuthLogoutProfiles(t, exact)
+	blank := authLogoutTestToken("corp_blank_alias")
+	blank.AccessToken = "access-unresolved-alias"
+	blank.RefreshToken = "refresh-unresolved-alias"
+	blank.UserID = ""
+	blank.UserName = ""
+	if err := authpkg.SaveTokenData(configDir, blank); err != nil {
+		t.Fatalf("SaveTokenData(blank) error = %v", err)
+	}
+	cfg, err := authpkg.LoadProfiles(configDir)
+	if err != nil {
+		t.Fatalf("LoadProfiles() error = %v", err)
+	}
+	var reserved string
+	for _, profile := range cfg.Profiles {
+		if profile.CorpID == blank.CorpID && profile.UserID == "" {
+			reserved = authpkg.ProfileSelectionSelector(profile, cfg)
+			break
+		}
+	}
+	if reserved == "" || reserved == blank.CorpID {
+		t.Fatalf("blank selector = %q, want reserved selector", reserved)
+	}
+
+	selections, multi, err := resolveMultiProfileSelections(configDir, reserved+","+blank.CorpID)
+	if err != nil {
+		t.Fatalf("resolveMultiProfileSelections() error = %v", err)
+	}
+	if !multi || len(selections) != 1 {
+		t.Fatalf("blank aliases = multi %v selections %#v, want one identity", multi, selections)
+	}
+	if selections[0].Selector != reserved || selections[0].Profile.UserID != "" {
+		t.Fatalf("blank selection = %#v, want first reserved alias preserved", selections[0])
+	}
+}
+
 func TestRuntimeRunnerKeepsSingleProfileBehavior(t *testing.T) {
 	setupAuthLogoutProfiles(t, authLogoutTestToken("corp_a"), authLogoutTestToken("corp_b"))
 	authpkg.SetRuntimeProfile("corp_a")

internal/app/pat_auth_retry.go

@@ -61,7 +61,7 @@ var (
 	patPollDeviceFlowWithInterval = pollPatDeviceFlowWithInterval
 	patSaveAppConfig              = authpkg.SaveAppConfig
 	patExchangeCodeForToken       = authpkg.ExchangeCodeForToken
-	patSaveTokenData              = authpkg.SaveTokenData
+	patSaveTokenData              = authpkg.SaveLoginTokenData
 	patSleep                      = time.Sleep
 	patPollHTTPDo                 = (*http.Client).Do
 	patPollNewRequest             = http.NewRequestWithContext

internal/app/profile_command.go

@@ -266,7 +266,7 @@ func selectProfileSwitchProfile(cmd *cobra.Command, configDir string) (string, e
 	}
 	choice := strings.TrimSpace(cfg.CurrentProfile)
 	if choice == "" {
-		choice = authpkg.ProfileSelector(cfg.Profiles[0])
+		choice = authpkg.ProfileSelectionSelector(cfg.Profiles[0], cfg)
 	}
 	return profileSwitchTUIRunner(cmd, cfg, choice)
 }
@@ -428,11 +428,36 @@ func (m profileSwitchTUIModel) selectedCorpID() string {
 	if m.selected < 0 || m.selected >= len(m.profiles) {
 		return ""
 	}
-	return authpkg.ProfileSelector(m.profiles[m.selected])
+	selected := m.profiles[m.selected]
+	return authpkg.ProfileSelectionSelector(selected, &authpkg.ProfilesConfig{Profiles: m.profiles})
 }
 
 func profileSwitchProfileIndex(profiles []authpkg.Profile, selector string, cfg *authpkg.ProfilesConfig) int {
 	selector = strings.TrimSpace(selector)
+	for i, profile := range profiles {
+		if authpkg.ProfileSelectionSelector(profile, cfg) == selector {
+			return i
+		}
+	}
+	// Accept an old current/previous pointer long enough for the migration path
+	// to canonicalize it. Only an unresolved profile in a multi-account
+	// organization qualifies, so ordinary exact account names cannot capture an
+	// identity selector that contains ':'.
+	legacyBlank := -1
+	for i, profile := range profiles {
+		if strings.TrimSpace(profile.UserID) != "" || strings.TrimSpace(profile.Name) != selector ||
+			profileCountForCorp(cfg, profile.CorpID) <= 1 {
+			continue
+		}
+		if legacyBlank >= 0 {
+			legacyBlank = -1
+			break
+		}
+		legacyBlank = i
+	}
+	if legacyBlank >= 0 {
+		return legacyBlank
+	}
 	if corpID, userID, exact := authpkg.ParseIdentitySelector(selector); exact {
 		for i, p := range profiles {
 			if strings.TrimSpace(p.CorpID) == corpID && strings.TrimSpace(p.UserID) == userID {
@@ -443,6 +468,9 @@ func profileSwitchProfileIndex(profiles []authpkg.Profile, selector string, cfg
 	}
 	fallback := -1
 	for i, p := range profiles {
+		if strings.TrimSpace(p.UserID) == "" && strings.TrimSpace(p.Name) == selector {
+			return i
+		}
 		if strings.TrimSpace(p.CorpID) == selector {
 			if fallback < 0 {
 				fallback = i
@@ -486,7 +514,7 @@ func profileSwitchProfileCells(p authpkg.Profile, cfg *authpkg.ProfilesConfig) (
 }
 
 func profileSwitchProfileStatus(p authpkg.Profile, cfg *authpkg.ProfilesConfig) string {
-	if cfg != nil && profileSelectorSelectsProfile(cfg.CurrentProfile, p, profileIsOrgCurrent(p, cfg), profileCountForCorp(cfg, p.CorpID) <= 1) {
+	if cfg != nil && profileSelectorSelectsProfile(cfg.CurrentProfile, p, cfg, profileIsOrgCurrent(p, cfg), profileCountForCorp(cfg, p.CorpID) <= 1) {
 		return "当前组织"
 	}
 	return ""
@@ -636,13 +664,14 @@ func writeProfileListTable(w io.Writer, configDir string, cfg *authpkg.ProfilesC
 	}
 	fmt.Fprintf(w, "%-3s %-28s %-34s %-10s %s\n", "CUR", "ORG_NAME", "CORP_ID", "STATUS", "USER")
 	for _, p := range cfg.Profiles {
+		selector := profileCLISelector(p, cfg)
 		view := profileViewFromProfile(
 			p,
 			cfg,
 			cfg.PrimaryProfile,
 			cfg.CurrentProfile,
 			profileCountForCorp(cfg, p.CorpID) == 1,
-			loadProfileTokenState(configDir, p),
+			loadProfileTokenState(configDir, p, selector),
 		)
 		current := ""
 		if view.IsCurrent {
@@ -698,13 +727,14 @@ func profileViews(configDir string, cfg *authpkg.ProfilesConfig) []profileView {
 	}
 	views := make([]profileView, 0, len(cfg.Profiles))
 	for _, p := range cfg.Profiles {
+		selector := profileCLISelector(p, cfg)
 		views = append(views, profileViewFromProfile(
 			p,
 			cfg,
 			cfg.PrimaryProfile,
 			cfg.CurrentProfile,
 			profileCountForCorp(cfg, p.CorpID) == 1,
-			loadProfileTokenState(configDir, p),
+			loadProfileTokenState(configDir, p, selector),
 		))
 	}
 	return views
@@ -719,7 +749,7 @@ func profileViewFromProfile(
 ) profileView {
 	isOrgCurrent := profileIsOrgCurrent(p, cfg)
 	view := profileView{
-		Profile:           authpkg.ProfileSelector(p),
+		Profile:           profileCLISelector(p, cfg),
 		CorpID:            p.CorpID,
 		CorpName:          profileOrgName(p),
 		UserID:            p.UserID,
@@ -731,8 +761,8 @@ func profileViewFromProfile(
 		RefreshExpAt:      p.RefreshExpAt,
 		LastLoginAt:       p.LastLoginAt,
 		LastUsedAt:        p.LastUsedAt,
-		IsPrimary:         profileSelectorSelectsProfile(primaryProfile, p, isOrgCurrent, onlyAccountInOrg),
-		IsCurrent:         profileSelectorSelectsProfile(currentProfile, p, isOrgCurrent, onlyAccountInOrg),
+		IsPrimary:         profileSelectorSelectsProfile(primaryProfile, p, cfg, isOrgCurrent, onlyAccountInOrg),
+		IsCurrent:         profileSelectorSelectsProfile(currentProfile, p, cfg, isOrgCurrent, onlyAccountInOrg),
 		IsOrgCurrent:      isOrgCurrent,
 	}
 	if tokenState != nil {
@@ -743,8 +773,12 @@ func profileViewFromProfile(
 	return view
 }
 
-func loadProfileTokenState(configDir string, profile authpkg.Profile) *profileTokenState {
-	data, err := profileLoadTokenData(configDir, authpkg.ProfileSelector(profile))
+func loadProfileTokenState(configDir string, profile authpkg.Profile, selectors ...string) *profileTokenState {
+	selector := authpkg.ProfileSelector(profile)
+	if len(selectors) > 0 && strings.TrimSpace(selectors[0]) != "" {
+		selector = strings.TrimSpace(selectors[0])
+	}
+	data, err := profileLoadTokenData(configDir, selector)
 	if errors.Is(err, authpkg.ErrTokenDataNotFound) || (err == nil && data == nil) {
 		return &profileTokenState{Status: authpkg.ProfileStatusRevoked}
 	}
@@ -762,6 +796,10 @@ func loadProfileTokenState(configDir string, profile authpkg.Profile) *profileTo
 	}
 }
 
+func profileCLISelector(profile authpkg.Profile, cfg *authpkg.ProfilesConfig) string {
+	return authpkg.ProfileSelectionSelector(profile, cfg)
+}
+
 func profileTokenTime(value time.Time) string {
 	if value.IsZero() {
 		return ""
@@ -769,8 +807,11 @@ func profileTokenTime(value time.Time) string {
 	return value.Format(time.RFC3339)
 }
 
-func profileSelectorSelectsProfile(selector string, profile authpkg.Profile, isOrgCurrent, onlyAccountInOrg bool) bool {
+func profileSelectorSelectsProfile(selector string, profile authpkg.Profile, cfg *authpkg.ProfilesConfig, isOrgCurrent, onlyAccountInOrg bool) bool {
 	selector = strings.TrimSpace(selector)
+	if selector == authpkg.ProfileSelectionSelector(profile, cfg) {
+		return true
+	}
 	if corpID, userID, exact := authpkg.ParseIdentitySelector(selector); exact {
 		return corpID == strings.TrimSpace(profile.CorpID) && userID == strings.TrimSpace(profile.UserID)
 	}

internal/app/root.go

@@ -385,11 +385,16 @@ func newRootCommandWithEngine(rootCtx context.Context, engine *pipeline.Engine,
 
 	schemaCmd := newSchemaCommand(loader)
 	mcpCmd := newMCPCommand(rootCtx, loader, runner, engine)
-	mcpCmd.Hidden = true
+	// The legacy dynamic MCP surface remains disabled, but reviewed static MCP
+	// helpers registered below are part of the public CLI and Schema surface.
+	mcpCmd.Hidden = false
+	mcpCmd.Short = "管理 MCP 服务连接信息"
+	mcpCmd.Long = "管理经过审核并纳入 Schema 的 MCP 服务连接辅助能力。"
 	// Wrap the caller so every MCP tool call's shape is recorded to the local
 	// usage log (privacy-preserving; see internal/shortcut/usage). Powers
 	// `dws shortcut stats` and future high-frequency shortcut distillation.
 	patCaller := newRecordingToolCaller(newToolCallerAdapter(runner, flags))
+	mcpCmd.AddCommand(newMCPURLGroup(patCaller))
 
 	utilityCommands := []*cobra.Command{
 		newAuthCommand(patCaller),
@@ -619,6 +624,7 @@ func hideNonDirectRuntimeCommands(root *cobra.Command) {
 		"profile":    true,
 		"version":    true,
 		"help":       true,
+		"markdown":   true,
 		"recovery":   true,
 		"schema":     true,
 		"mcp":        true,

internal/app/root_help_test.go

@@ -15,11 +15,14 @@ package app
 
 import (
 	"bytes"
+	stderrors "errors"
 	"io"
 	"os"
+	"path/filepath"
 	"strings"
 	"testing"
 
+	apperrors "github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/errors"
 	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/pkg/edition"
 	"github.com/spf13/cobra"
 )
@@ -75,7 +78,14 @@ func TestRootKeepsMainBranchChatCompatibilityCommands(t *testing.T) {
 		}
 	}
 
-	mediaUpload := mustFindCommand(t, root, "chat", "media", "upload")
+	mediaGroup := mustFindCommand(t, root, "chat", "media")
+	if mediaGroup.Deprecated == "" || mediaGroup.Hidden || !mediaGroup.Runnable() {
+		t.Fatalf("chat media compatibility contract: deprecated=%q hidden=%v runnable=%v", mediaGroup.Deprecated, mediaGroup.Hidden, mediaGroup.Runnable())
+	}
+	mediaUpload := mustFindCommand(t, mediaGroup, "upload")
+	if mediaUpload.Deprecated == "" || mediaUpload.Hidden || !mediaUpload.Runnable() {
+		t.Fatalf("chat media upload compatibility contract: deprecated=%q hidden=%v runnable=%v", mediaUpload.Deprecated, mediaUpload.Hidden, mediaUpload.Runnable())
+	}
 	for _, flag := range []string{"file", "type"} {
 		if mediaUpload.Flags().Lookup(flag) == nil {
 			t.Fatalf("chat media upload missing --%s", flag)
@@ -88,6 +98,113 @@ func TestRootKeepsMainBranchChatCompatibilityCommands(t *testing.T) {
 	mustFindCommand(t, root, "conference", "meeting", "reserve")
 }
 
+func TestChatHelpAndSchemaHideRetiredMediaUpload(t *testing.T) {
+	for _, args := range [][]string{
+		{"chat", "--help"},
+		{"chat", "media", "--help"},
+	} {
+		root := NewRootCommand()
+		var output bytes.Buffer
+		root.SetOut(&output)
+		root.SetErr(&output)
+		root.SetArgs(args)
+		if err := root.Execute(); err != nil {
+			t.Fatalf("dws %s: %v\n%s", strings.Join(args, " "), err, output.String())
+		}
+		for _, line := range strings.Split(output.String(), "\n") {
+			fields := strings.Fields(line)
+			if len(fields) > 0 && (fields[0] == "media" || fields[0] == "upload") {
+				t.Fatalf("dws %s exposes retired command in Help line %q:\n%s", strings.Join(args, " "), line, output.String())
+			}
+		}
+	}
+
+	root := NewRootCommand()
+	var output bytes.Buffer
+	root.SetOut(&output)
+	root.SetErr(&output)
+	root.SetArgs([]string{"schema", "--cli-path", "chat media upload", "--format", "json"})
+	err := root.Execute()
+	if err == nil {
+		t.Fatalf("retired chat media upload remains queryable from Schema:\n%s", output.String())
+	}
+	if !strings.Contains(err.Error(), "unknown runtime schema path") {
+		t.Fatalf("retired chat media upload Schema error = %v, want unknown path", err)
+	}
+}
+
+func TestRootChatMediaUploadWithoutAppCredentialsReturnsMigrationValidation(t *testing.T) {
+	for _, key := range []string{"DWS_CLIENT_ID", "DWS_CLIENT_SECRET"} {
+		value, existed := os.LookupEnv(key)
+		if err := os.Unsetenv(key); err != nil {
+			t.Fatalf("unset %s: %v", key, err)
+		}
+		t.Cleanup(func() {
+			if existed {
+				_ = os.Setenv(key, value)
+				return
+			}
+			_ = os.Unsetenv(key)
+		})
+		if _, exists := os.LookupEnv(key); exists {
+			t.Fatalf("%s is still set", key)
+		}
+	}
+	t.Setenv("DWS_CONFIG_DIR", filepath.Join(t.TempDir(), "config"))
+
+	filePath := filepath.Join(t.TempDir(), "image.png")
+	if err := os.WriteFile(filePath, []byte("image"), 0o600); err != nil {
+		t.Fatalf("write image fixture: %v", err)
+	}
+	commandArgs := []string{
+		"chat", "media", "upload",
+		"--file", filePath,
+		"--type", "image",
+	}
+	previousArgs := os.Args
+	os.Args = append([]string{"dws"}, commandArgs...)
+	t.Cleanup(func() { os.Args = previousArgs })
+
+	root := NewRootCommand()
+	var output bytes.Buffer
+	root.SetOut(&output)
+	root.SetErr(&output)
+	root.SetArgs(commandArgs)
+	err := root.Execute()
+	if err == nil {
+		t.Fatalf("chat media upload succeeded without app credentials:\n%s", output.String())
+	}
+
+	var typed *apperrors.Error
+	if !stderrors.As(err, &typed) {
+		t.Fatalf("chat media upload error type = %T, want *errors.Error: %v", err, err)
+	}
+	if typed.Category != apperrors.CategoryValidation {
+		t.Fatalf("chat media upload category = %q, want %q", typed.Category, apperrors.CategoryValidation)
+	}
+	if exitCode := apperrors.ExitCode(err); exitCode != 3 {
+		t.Fatalf("chat media upload exit code = %d, want 3", exitCode)
+	}
+
+	got := output.String() + "\n" + err.Error()
+	for _, want := range []string{"已下线", "chat message send --msg-type file --file-path"} {
+		if !strings.Contains(got, want) {
+			t.Fatalf("chat media upload migration output missing %q:\n%s", want, got)
+		}
+	}
+	for _, forbidden := range []string{
+		"DWS_CLIENT_ID",
+		"DWS_CLIENT_SECRET",
+		"缺少应用凭证",
+		"AppSecret",
+		"clientSecret",
+	} {
+		if strings.Contains(got, forbidden) {
+			t.Fatalf("chat media upload returned credential error %q:\n%s", forbidden, got)
+		}
+	}
+}
+
 func TestRootKeepsContactWukongCompatibilityCommands(t *testing.T) {
 	root := NewRootCommand()
 	label := mustFindCommand(t, root, "contact", "label")

internal/app/runner.go

@@ -210,7 +210,14 @@ func (r *runtimeRunner) Run(ctx context.Context, invocation executor.Invocation)
 		if profile == nil {
 			return executor.Result{}, apperrors.NewValidation(fmt.Sprintf("profile %q not found", rawProfile))
 		}
-		authpkg.SetRuntimeProfile(authpkg.ProfileSelector(*profile))
+		resolvedSelector := authpkg.ProfileSelector(*profile)
+		if strings.TrimSpace(profile.UserID) == "" {
+			// Preserve a unique local-name selector for an unresolved account.
+			// Reducing it to corpId can select a different exact account through
+			// the organization's current-account pointer.
+			resolvedSelector = rawProfile
+		}
+		authpkg.SetRuntimeProfile(resolvedSelector)
 		defer authpkg.SetRuntimeProfile(rawProfile)
 	}
 
@@ -351,6 +358,9 @@ func (r *runtimeRunner) runMultiProfile(ctx context.Context, invocation executor
 
 	for _, selection := range selections {
 		resolvedSelector := authpkg.ProfileSelector(selection.Profile)
+		if strings.TrimSpace(selection.Profile.UserID) == "" {
+			resolvedSelector = selection.Selector
+		}
 		authpkg.SetRuntimeProfile(resolvedSelector)
 		result, err := r.runSingle(ctx, cloneInvocation(invocation), false)
 

internal/app/schema_shortcut_contract_test.go

@@ -0,0 +1,358 @@
+// Copyright 2026 Alibaba Group
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+
+package app
+
+import (
+	"bytes"
+	"encoding/json"
+	"fmt"
+	"strings"
+	"testing"
+
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/shortcut"
+)
+
+const publicShortcutSchemaCount = 210
+
+func TestEmbeddedSchemaPublishesEveryPublicShortcutContract(t *testing.T) {
+	tools := embeddedSchemaAllToolsForHelpFlagTest(t, NewRootCommand())
+	public := make([]shortcut.Shortcut, 0, publicShortcutSchemaCount)
+	for _, candidate := range shortcut.All() {
+		if candidate.UserDefined || !shortcut.InPublicCatalog(candidate.Service, candidate.Command) {
+			continue
+		}
+		public = append(public, candidate)
+	}
+	if got := len(public); got != publicShortcutSchemaCount {
+		t.Fatalf("public built-in shortcuts = %d, want %d", got, publicShortcutSchemaCount)
+	}
+
+	deliveredShortcuts := 0
+	for canonical := range tools {
+		if strings.Contains(canonical, ".shortcut_") {
+			deliveredShortcuts++
+		}
+	}
+	if deliveredShortcuts != publicShortcutSchemaCount {
+		t.Fatalf("embedded schema --all shortcut tools = %d, want %d", deliveredShortcuts, publicShortcutSchemaCount)
+	}
+
+	for _, declared := range public {
+		declared := declared
+		t.Run(declared.Service+"/"+strings.TrimPrefix(declared.Command, "+"), func(t *testing.T) {
+			canonical := shortcutSchemaCanonical(declared)
+			tool := tools[canonical]
+			if tool == nil {
+				t.Fatalf("embedded schema --all is missing %s (%s %s)", canonical, declared.Service, declared.Command)
+			}
+			assertEmbeddedShortcutIdentityAndSelection(t, tool, declared, canonical)
+			assertEmbeddedShortcutSafetyAndInterface(t, tool, declared, canonical)
+			assertEmbeddedShortcutParameters(t, tool, declared, canonical)
+			assertEmbeddedShortcutConstraints(t, tool, declared, canonical)
+		})
+	}
+}
+
+func TestEmbeddedShortcutProgressiveQueriesReturnCompleteContracts(t *testing.T) {
+	leaf := executeShortcutSchemaQuery(t, "--cli-path", "chat +messages-read-status")
+	if got, want := schemaContractString(leaf["canonical_path"]), "chat.shortcut_messages_read_status"; got != want {
+		t.Fatalf("shortcut leaf canonical_path = %q, want %q", got, want)
+	}
+	if got, want := schemaContractString(leaf["confirmation"]), "not_required"; got != want {
+		t.Fatalf("shortcut leaf confirmation = %q, want %q", got, want)
+	}
+	conversationID := schemaContractMap(leaf["parameters"])["conversation-id"]
+	if required, _ := conversationID["required"].(bool); !required {
+		t.Fatal("public --conversation-id must become required after hidden compatibility aliases are removed from Schema")
+	}
+	if got := leaf["constraints"]; got != nil {
+		t.Fatalf("shortcut leaf constraints = %#v, want omitted after hidden compatibility aliases collapse", got)
+	}
+
+	constrainedLeaf := executeShortcutSchemaQuery(t, "--cli-path", "calendar +freebusy")
+	wantConstraints := map[string]any{
+		"require_one_of": [][]string{{"users", "rooms"}},
+	}
+	if got := constrainedLeaf["constraints"]; !schemaContractJSONEqual(got, wantConstraints) {
+		t.Fatalf("shortcut leaf constraints = %#v, want %#v", got, wantConstraints)
+	}
+
+	product := executeShortcutSchemaQuery(t, "chat")
+	productPayload, _ := product["product"].(map[string]any)
+	if got, want := int(product["count"].(float64)), 120; got != want {
+		t.Fatalf("schema chat count = %d, want %d", got, want)
+	}
+	summaries := schemaContractObjectSlice(productPayload["tools"])
+	shortcutCount := 0
+	for _, summary := range summaries {
+		if strings.HasPrefix(schemaContractString(summary["canonical_path"]), "chat.shortcut_") {
+			shortcutCount++
+		}
+	}
+	if shortcutCount != 42 {
+		t.Fatalf("schema chat shortcut summaries = %d, want 42", shortcutCount)
+	}
+}
+
+func executeShortcutSchemaQuery(t testing.TB, args ...string) map[string]any {
+	t.Helper()
+	root := NewRootCommand()
+	var stdout, stderr bytes.Buffer
+	root.SetOut(&stdout)
+	root.SetErr(&stderr)
+	root.SetArgs(append([]string{"schema"}, append(args, "--format", "json")...))
+	if err := root.Execute(); err != nil {
+		t.Fatalf("execute dws schema %q: %v; stderr=%s", strings.Join(args, " "), err, stderr.String())
+	}
+	var payload map[string]any
+	if err := json.Unmarshal(stdout.Bytes(), &payload); err != nil {
+		t.Fatalf("decode dws schema %q: %v", strings.Join(args, " "), err)
+	}
+	return payload
+}
+
+func shortcutSchemaCanonical(declared shortcut.Shortcut) string {
+	name := strings.ReplaceAll(strings.TrimPrefix(declared.Command, "+"), "-", "_")
+	return declared.Service + ".shortcut_" + name
+}
+
+func assertEmbeddedShortcutIdentityAndSelection(
+	t testing.TB,
+	tool map[string]any,
+	declared shortcut.Shortcut,
+	canonical string,
+) {
+	t.Helper()
+	if got, want := schemaContractString(tool["canonical_path"]), canonical; got != want {
+		t.Errorf("canonical_path = %q, want %q", got, want)
+	}
+	if got, want := schemaContractString(tool["primary_cli_path"]), declared.Service+" "+declared.Command; got != want {
+		t.Errorf("%s primary_cli_path = %q, want %q", canonical, got, want)
+	}
+	if got, want := schemaContractString(tool["agent_summary"]), declared.Description; got != want {
+		t.Errorf("%s agent_summary = %q, want %q", canonical, got, want)
+	}
+	if got, want := schemaContractStringSlice(tool["use_when"]), []string{declared.Intent}; !schemaContractJSONEqual(got, want) {
+		t.Errorf("%s use_when = %#v, want %#v", canonical, got, want)
+	}
+	if len(schemaContractStringSlice(tool["avoid_when"])) == 0 {
+		t.Errorf("%s has no reviewed avoid_when", canonical)
+	}
+	examples := schemaContractStringSlice(tool["examples"])
+	if len(examples) == 0 || len(examples) > 2 {
+		t.Errorf("%s examples = %d, want 1..2", canonical, len(examples))
+	}
+	for _, example := range examples {
+		if strings.Contains(example, "--yes") {
+			t.Errorf("%s stores unsafe example %q", canonical, example)
+		}
+		if !strings.HasPrefix(example, "dws "+declared.Service+" "+declared.Command) {
+			t.Errorf("%s example does not use its primary path: %q", canonical, example)
+		}
+	}
+}
+
+func assertEmbeddedShortcutSafetyAndInterface(
+	t testing.TB,
+	tool map[string]any,
+	declared shortcut.Shortcut,
+	canonical string,
+) {
+	t.Helper()
+	risk := declared.Risk
+	if risk == "" {
+		risk = shortcut.RiskRead
+	}
+	wantEffect, wantRisk, wantConfirmation, wantIdempotency := "read", "low", "not_required", "idempotent"
+	switch risk {
+	case shortcut.RiskWrite:
+		wantEffect, wantRisk, wantConfirmation, wantIdempotency = "write", "medium", "user_required", "unknown"
+	case shortcut.RiskHighWrite:
+		wantEffect, wantRisk, wantConfirmation, wantIdempotency = "destructive", "high", "user_required", "unknown"
+	}
+	for field, want := range map[string]string{
+		"effect":         wantEffect,
+		"risk":           wantRisk,
+		"confirmation":   wantConfirmation,
+		"idempotency":    wantIdempotency,
+		"interface_mode": "composite",
+		"availability":   "available",
+	} {
+		if got := schemaContractString(tool[field]); got != want {
+			t.Errorf("%s %s = %q, want %q", canonical, field, got, want)
+		}
+	}
+	if strings.TrimSpace(schemaContractString(tool["interface_reason"])) == "" {
+		t.Errorf("%s has no reviewed composite interface reason", canonical)
+	}
+}
+
+func assertEmbeddedShortcutParameters(
+	t testing.TB,
+	tool map[string]any,
+	declared shortcut.Shortcut,
+	canonical string,
+) {
+	t.Helper()
+	parameters := schemaContractMap(tool["parameters"])
+	publicFlags := make([]shortcut.Flag, 0, len(declared.Flags))
+	for _, flag := range declared.Flags {
+		if !flag.Hidden {
+			publicFlags = append(publicFlags, flag)
+		}
+	}
+	if got, want := len(parameters), len(publicFlags); got != want {
+		t.Errorf("%s parameters = %d, want %d", canonical, got, want)
+	}
+	for _, flag := range publicFlags {
+		parameter := parameters[flag.Name]
+		if parameter == nil {
+			t.Errorf("%s is missing parameter --%s", canonical, flag.Name)
+			continue
+		}
+		flagType := flag.Type
+		if flagType == "" {
+			flagType = shortcut.FlagString
+		}
+		wantType := map[shortcut.FlagType]string{
+			shortcut.FlagString:      "string",
+			shortcut.FlagBool:        "boolean",
+			shortcut.FlagInt:         "integer",
+			shortcut.FlagStringSlice: "array",
+		}[flagType]
+		if got := schemaContractString(parameter["type"]); got != wantType {
+			t.Errorf("%s --%s type = %q, want %q", canonical, flag.Name, got, wantType)
+		}
+		if got, _ := parameter["required"].(bool); got != shortcutSchemaRequired(declared, flag.Name) {
+			t.Errorf("%s --%s required = %t, want %t", canonical, flag.Name, got, shortcutSchemaRequired(declared, flag.Name))
+		}
+		if got, want := schemaContractString(parameter["default"]), shortcutSchemaDefault(flag); got != want {
+			t.Errorf("%s --%s default = %q, want %q", canonical, flag.Name, got, want)
+		}
+		gotEnum := schemaContractStringSlice(parameter["enum"])
+		if len(flag.Enum) == 0 {
+			if len(gotEnum) != 0 {
+				t.Errorf("%s --%s enum = %#v, want empty", canonical, flag.Name, gotEnum)
+			}
+		} else if !schemaContractJSONEqual(gotEnum, flag.Enum) {
+			t.Errorf("%s --%s enum = %#v, want %#v", canonical, flag.Name, gotEnum, flag.Enum)
+		}
+	}
+}
+
+func shortcutSchemaDefault(flag shortcut.Flag) string {
+	value := strings.TrimSpace(flag.Default)
+	switch flag.Type {
+	case shortcut.FlagBool:
+		if value != "true" {
+			return ""
+		}
+	case shortcut.FlagInt:
+		if value == "0" {
+			return ""
+		}
+	case shortcut.FlagStringSlice:
+		if value != "" {
+			return "[" + value + "]"
+		}
+	}
+	return value
+}
+
+func shortcutSchemaRequired(declared shortcut.Shortcut, flagName string) bool {
+	for _, flag := range declared.Flags {
+		if flag.Name == flagName && flag.Required {
+			return true
+		}
+	}
+	public := make(map[string]bool, len(declared.Flags))
+	for _, flag := range declared.Flags {
+		if !flag.Hidden {
+			public[flag.Name] = true
+		}
+	}
+	for _, constraint := range declared.Constraints {
+		if constraint.Kind != shortcut.ConstraintAtLeastOne && constraint.Kind != shortcut.ConstraintExactlyOne {
+			continue
+		}
+		visible := make([]string, 0, len(constraint.Flags))
+		for _, constrained := range constraint.Flags {
+			if public[constrained] {
+				visible = append(visible, constrained)
+			}
+		}
+		if len(visible) == 1 && visible[0] == flagName {
+			return true
+		}
+	}
+	return false
+}
+
+func assertEmbeddedShortcutConstraints(
+	t testing.TB,
+	tool map[string]any,
+	declared shortcut.Shortcut,
+	canonical string,
+) {
+	t.Helper()
+	public := make(map[string]bool, len(declared.Flags))
+	for _, flag := range declared.Flags {
+		if !flag.Hidden {
+			public[flag.Name] = true
+		}
+	}
+	want := map[string][][]string{}
+	for _, constraint := range declared.Constraints {
+		flags := make([]string, 0, len(constraint.Flags))
+		for _, flagName := range constraint.Flags {
+			if public[flagName] {
+				flags = append(flags, flagName)
+			}
+		}
+		switch constraint.Kind {
+		case shortcut.ConstraintAtLeastOne:
+			if len(flags) > 1 {
+				want["require_one_of"] = append(want["require_one_of"], flags)
+			}
+		case shortcut.ConstraintExactlyOne:
+			if len(flags) > 1 {
+				want["require_one_of"] = append(want["require_one_of"], flags)
+				want["mutually_exclusive"] = append(want["mutually_exclusive"], flags)
+			}
+		case shortcut.ConstraintMutuallyExclusive:
+			if len(flags) > 1 {
+				want["mutually_exclusive"] = append(want["mutually_exclusive"], flags)
+			}
+		case shortcut.ConstraintCustom:
+			for _, flagName := range flags {
+				description := schemaContractString(schemaContractMap(tool["parameters"])[flagName]["description"])
+				for _, requiredText := range []string{"原文不能为空", "不能重复"} {
+					if !strings.Contains(description, requiredText) {
+						t.Errorf("%s --%s description does not publish custom constraint %q: %q", canonical, flagName, requiredText, description)
+					}
+				}
+			}
+		default:
+			t.Errorf("%s has unsupported declared shortcut constraint %q", canonical, constraint.Kind)
+		}
+	}
+	if len(want) == 0 {
+		if got := tool["constraints"]; got != nil {
+			t.Errorf("%s constraints = %#v, want omitted", canonical, got)
+		}
+		return
+	}
+	if got := tool["constraints"]; !schemaContractJSONEqual(got, want) {
+		t.Errorf("%s constraints = %s, want %s", canonical, mustShortcutJSON(got), mustShortcutJSON(want))
+	}
+}
+
+func mustShortcutJSON(value any) string {
+	encoded, err := json.Marshal(value)
+	if err != nil {
+		return fmt.Sprintf("%#v", value)
+	}
+	return string(encoded)
+}

internal/auth/access_token_snapshot_test.go

@@ -3,6 +3,10 @@ package auth
 import (
 	"context"
 	"errors"
+	"io"
+	"log/slog"
+	"net/http"
+	"net/url"
 	"testing"
 	"time"
 )
@@ -22,15 +26,92 @@ func TestCrossPlatformCoverageOAuthProviderTokenSnapshotPreservesLoadFailure(t *
 	}
 }
 
-func TestCrossPlatformCoverageOAuthProviderLoginPreservesLoadFailure(t *testing.T) {
-	oldLoad := oauthLoadToken
-	want := errors.New("keychain permission denied")
-	oauthLoadToken = func(string) (*TokenData, error) { return nil, want }
-	t.Cleanup(func() { oauthLoadToken = oldLoad })
+func TestCrossPlatformCoverageOAuthProviderLoginReauthorizesAfterLoadFailureAndRejectsUnreadableTarget(t *testing.T) {
+	cleanupKeychain(t)
+	setLoginPreflightCredentials(t)
 
-	_, err := NewOAuthProvider(t.TempDir(), nil).Login(context.Background(), false)
-	if !errors.Is(err, want) {
-		t.Fatalf("error = %v, want cause %v", err, want)
+	oldLoad := oauthLoadToken
+	oldOpenBrowser := oauthOpenBrowser
+	oldExchange := oauthExchange
+	oldCheckStatus := oauthCheckStatus
+	oldSave := oauthSaveToken
+	oldKeychainGet := authKeychainGet
+	oldLoginTimeout := oauthLoginTimeout
+	t.Cleanup(func() {
+		oauthLoadToken = oldLoad
+		oauthOpenBrowser = oldOpenBrowser
+		oauthExchange = oldExchange
+		oauthCheckStatus = oldCheckStatus
+		oauthSaveToken = oldSave
+		authKeychainGet = oldKeychainGet
+		oauthLoginTimeout = oldLoginTimeout
+	})
+	oauthLoginTimeout = 2 * time.Second
+
+	loadErr := errors.New("keychain permission denied")
+	targetErr := errors.New("target token ciphertext is unreadable")
+	oauthLoadToken = func(string) (*TokenData, error) { return nil, loadErr }
+
+	browserCalls := 0
+	oauthOpenBrowser = func(authURL string) error {
+		browserCalls++
+		parsed, err := url.Parse(authURL)
+		if err != nil {
+			return err
+		}
+		callbackURL := parsed.Query().Get("redirect_uri") + "?code=reauthorize"
+		response, err := (&http.Client{Timeout: 5 * time.Second}).Get(callbackURL)
+		if err != nil {
+			return err
+		}
+		_, _ = io.Copy(io.Discard, response.Body)
+		return response.Body.Close()
+	}
+
+	exchangeCalls := 0
+	oauthExchange = func(*OAuthProvider, context.Context, string) (*TokenData, error) {
+		exchangeCalls++
+		return &TokenData{
+			AccessToken: "new-access",
+			CorpID:      "corp-target",
+			UserID:      "user-target",
+		}, nil
+	}
+	oauthCheckStatus = func(*OAuthProvider, context.Context, string) (*CLIAuthStatus, error) {
+		return &CLIAuthStatus{Success: true, Result: &CLIAuthResult{CLIAuthEnabled: true}}, nil
+	}
+
+	targetReads := 0
+	authKeychainGet = func(_ string, account string) (string, error) {
+		if account == TokenAccountForIdentity("corp-target", "user-target") {
+			targetReads++
+			return "", targetErr
+		}
+		return "", nil
+	}
+	saveCalls := 0
+	oauthSaveToken = func(string, *TokenData) error {
+		saveCalls++
+		return nil
+	}
+
+	provider := NewOAuthProvider(t.TempDir(), slog.New(slog.NewTextHandler(io.Discard, nil)))
+	provider.Output = io.Discard
+	_, err := provider.Login(context.Background(), false)
+	if !errors.Is(err, targetErr) {
+		t.Fatalf("Login() error = %v, want target cause %v", err, targetErr)
+	}
+	if errors.Is(err, loadErr) {
+		t.Fatalf("Login() returned stale load failure instead of reauthorizing: %v", err)
+	}
+	if browserCalls != 1 || exchangeCalls != 1 {
+		t.Fatalf("authorization calls = browser:%d exchange:%d, want 1 each", browserCalls, exchangeCalls)
+	}
+	if targetReads != 1 {
+		t.Fatalf("target slot reads = %d, want 1", targetReads)
+	}
+	if saveCalls != 0 {
+		t.Fatalf("SaveTokenData calls = %d, want 0", saveCalls)
 	}
 }
 

internal/auth/auth_legacy_platform_coverage_test.go

@@ -0,0 +1,297 @@
+// Copyright 2026 Alibaba Group
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package auth
+
+import (
+	"errors"
+	"strings"
+	"testing"
+
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/keychain"
+)
+
+// The native coverage jobs intentionally execute only TestCrossPlatformCoverage
+// entry points. Keep the compatibility assertions below as independently named
+// regression tests for the stable make target, and exercise the same functions
+// here as isolated subtests so their cleanup hooks run between cases.
+func TestCrossPlatformCoverageAuthLegacyCompatibilityRegressions(t *testing.T) {
+	t.Setenv(keychain.DisableKeychainEnv, "1")
+	t.Setenv(keychain.StorageDirEnv, t.TempDir())
+	tests := []struct {
+		name string
+		run  func(*testing.T)
+	}{
+		{"prepare unique global owner", TestPrepareLoginPersistenceRepairsOnlySafeGlobalOwner},
+		{"token load isolation matrix", TestTokenLoadIsolationMatrix},
+		{"reject future profiles before remote work", TestPersistingLoginFlowsRejectFutureProfilesBeforeRemoteWork},
+		{"fresh UID-less isolation", TestFreshUIDLessExactLoginCannotOverwriteExistingUnresolvedProfile},
+		{"reject mismatched exact switch", TestSetCurrentProfileRejectsMismatchedExactIdentitySlotBesideBlankProfile},
+		{"reject unreadable previous identity", TestUsePreviousProfileRejectsUnreadableExactIdentityBesideBlankProfile},
+		{"reauthorization guidance without profile", TestLegacyRefreshReauthorizationGuidanceWithoutProfileStillExplainsLogin},
+		{"repair v3 unresolved profile", TestPrepareLoginPersistenceV3UnresolvedProfileRepair},
+		{"device ignores unrelated unreadable profile", TestDeviceLoginIgnoresUnreadableUnrelatedProfile},
+		{"reject unreadable global before login", TestPrepareLoginPersistenceUnreadableGlobalFailsClosedBeforeRemote},
+		{"device validates resolved target", TestDeviceFlowChecksResolvedTargetBeforeSave},
+		{"accept recoverable credential material", TestPrepareLoginPersistenceRequiresCredentialMaterialButNotValidity},
+		{"auth code validates resolved target", TestExchangeAuthCodeChecksResolvedTargetBeforeSave},
+		{"standalone exchange prepares and marks fresh", TestExchangeCodeForTokenPreparesBeforeRemoteAndMarksFresh},
+	}
+	for _, test := range tests {
+		t.Run(test.name, test.run)
+	}
+}
+
+func TestCrossPlatformCoverageHalfMigratedGlobalRepairRemainingEdges(t *testing.T) {
+	t.Setenv(keychain.DisableKeychainEnv, "1")
+	t.Setenv(keychain.StorageDirEnv, t.TempDir())
+
+	t.Run("nil global token", func(t *testing.T) {
+		isolateHalfMigratedRepairHooks(t)
+		profilesLoadLegacy = func() (*TokenData, error) { return nil, nil }
+		cfg := &ProfilesConfig{Version: profilesVersion, Profiles: []Profile{{CorpID: "corp_nil_global"}}}
+		if err := repairHalfMigratedGlobalTokenLocked(cfg); err != nil {
+			t.Fatalf("repairHalfMigratedGlobalTokenLocked() error = %v", err)
+		}
+	})
+
+	t.Run("global token without organization", func(t *testing.T) {
+		isolateHalfMigratedRepairHooks(t)
+		profilesLoadLegacy = func() (*TokenData, error) {
+			return &TokenData{AccessToken: "legacy"}, nil
+		}
+		cfg := &ProfilesConfig{Version: profilesVersion, Profiles: []Profile{{CorpID: "corp_other"}}}
+		if err := repairHalfMigratedGlobalTokenLocked(cfg); err != nil {
+			t.Fatalf("repairHalfMigratedGlobalTokenLocked() error = %v", err)
+		}
+	})
+
+	t.Run("orphan global organization", func(t *testing.T) {
+		isolateHalfMigratedRepairHooks(t)
+		profilesLoadLegacy = func() (*TokenData, error) {
+			return &TokenData{AccessToken: "legacy", CorpID: "corp_orphan"}, nil
+		}
+		cfg := &ProfilesConfig{Version: profilesVersion, Profiles: []Profile{{CorpID: "corp_other"}}}
+		if err := repairHalfMigratedGlobalTokenLocked(cfg); err != nil {
+			t.Fatalf("repairHalfMigratedGlobalTokenLocked() error = %v", err)
+		}
+	})
+
+	t.Run("identity repair write failure", func(t *testing.T) {
+		isolateHalfMigratedRepairHooks(t)
+		failure := errors.New("identity repair write failure")
+		const corpID, userID = "corp_identity_repair", "user_identity_repair"
+		profilesLoadLegacy = func() (*TokenData, error) {
+			return &TokenData{AccessToken: "legacy", CorpID: corpID}, nil
+		}
+		profilesLoadCorp = func(string) (*TokenData, error) {
+			return &TokenData{AccessToken: "organization", CorpID: corpID}, nil
+		}
+		profilesSaveIdentity = func(string, string, *TokenData) error { return failure }
+		cfg := &ProfilesConfig{Version: profilesVersion, Profiles: []Profile{{CorpID: corpID, UserID: userID}}}
+		if err := repairHalfMigratedGlobalTokenLocked(cfg); !errors.Is(err, failure) {
+			t.Fatalf("repairHalfMigratedGlobalTokenLocked() error = %v, want %v", err, failure)
+		}
+	})
+
+	t.Run("organization repair write failure", func(t *testing.T) {
+		isolateHalfMigratedRepairHooks(t)
+		failure := errors.New("organization repair write failure")
+		const corpID = "corp_organization_repair"
+		profilesLoadLegacy = func() (*TokenData, error) {
+			return &TokenData{AccessToken: "legacy", CorpID: corpID}, nil
+		}
+		profilesSaveCorp = func(string, *TokenData) error { return failure }
+		cfg := &ProfilesConfig{Version: profilesVersion, Profiles: []Profile{{CorpID: corpID}}}
+		if err := repairHalfMigratedGlobalTokenLocked(cfg); !errors.Is(err, failure) {
+			t.Fatalf("repairHalfMigratedGlobalTokenLocked() error = %v, want %v", err, failure)
+		}
+	})
+
+	t.Run("nil profile is not canonical", func(t *testing.T) {
+		if loginProfileHasUsableCanonicalToken(nil, nil, nil) {
+			t.Fatal("nil profile was treated as a usable canonical token")
+		}
+	})
+
+	t.Run("global write preflight reports unreadable slot", func(t *testing.T) {
+		oldGet := authKeychainGet
+		failure := errors.New("global ciphertext is unreadable")
+		authKeychainGet = func(_, account string) (string, error) {
+			if account == keychain.AccountToken {
+				return "", failure
+			}
+			return "", nil
+		}
+		t.Cleanup(func() { authKeychainGet = oldGet })
+
+		err := preflightTokenWritePersistence(t.TempDir(), &TokenData{AccessToken: "fresh"})
+		if !errors.Is(err, failure) {
+			t.Fatalf("preflightTokenWritePersistence() error = %v, want %v", err, failure)
+		}
+	})
+}
+
+func TestCrossPlatformCoverageLegacyProfileGuardRemainingEdges(t *testing.T) {
+	t.Run("empty v3 registry normalizes and persists", func(t *testing.T) {
+		oldLoad := profilesLoad
+		oldSave := profilesSave
+		cfg := &ProfilesConfig{Version: profilesUnresolvedSelectorVersion}
+		profilesLoad = func(string) (*ProfilesConfig, error) { return cfg, nil }
+		saves := 0
+		profilesSave = func(string, *ProfilesConfig) error {
+			saves++
+			return nil
+		}
+		t.Cleanup(func() {
+			profilesLoad = oldLoad
+			profilesSave = oldSave
+		})
+
+		if err := ensureProfilesMigrationLocked(t.TempDir()); err != nil {
+			t.Fatalf("ensureProfilesMigrationLocked() error = %v", err)
+		}
+		if cfg.Version != profilesVersion || saves != 1 {
+			t.Fatalf("normalized registry = version %d saves %d", cfg.Version, saves)
+		}
+	})
+
+	if normalizeProfilesVersionForSelectors(nil) {
+		t.Fatal("nil profile registry reported a version change")
+	}
+	future := &ProfilesConfig{Version: profilesMaxVersion + 1}
+	if normalizeProfilesVersionForSelectors(future) {
+		t.Fatal("future profile registry reported a version change")
+	}
+	if profilesConfigContainsUnresolvedSelector(nil) {
+		t.Fatal("nil profile registry contained an unresolved selector")
+	}
+	reserved := unresolvedProfileSelector("corp_org_current_guard")
+	if !profilesConfigContainsUnresolvedSelector(&ProfilesConfig{
+		OrgCurrentProfiles: map[string]string{"corp_org_current_guard": reserved},
+	}) {
+		t.Fatal("reserved organization-current selector was not detected")
+	}
+	if selectorConflictsWithOrganizationGrammar(nil, "corp") {
+		t.Fatal("nil profile registry reported an organization-selector conflict")
+	}
+	if err := validateIdentityOnlyProfileToken(Profile{}); !errors.Is(err, ErrTokenDataNotFound) {
+		t.Fatalf("validateIdentityOnlyProfileToken(blank) error = %v", err)
+	}
+
+	oldLoadIdentity := profilesLoadIdentity
+	profilesLoadIdentity = func(string, string) (*TokenData, error) { return nil, nil }
+	t.Cleanup(func() { profilesLoadIdentity = oldLoadIdentity })
+	if err := validateIdentityOnlyProfileToken(Profile{CorpID: "corp_nil_identity", UserID: "user_nil_identity"}); !errors.Is(err, ErrTokenDataNotFound) {
+		t.Fatalf("validateIdentityOnlyProfileToken(nil token) error = %v", err)
+	}
+}
+
+func TestCrossPlatformCoverageTokenPersistenceRemainingEdges(t *testing.T) {
+	t.Setenv(keychain.DisableKeychainEnv, "1")
+	t.Setenv(keychain.StorageDirEnv, t.TempDir())
+
+	plan := planTokenPersistenceWrites(&ProfilesConfig{}, nil, "")
+	if !plan.WriteGlobal || plan.CorpID != "" {
+		t.Fatalf("nil-token write plan = %#v", plan)
+	}
+	if err := SaveLoginTokenData(t.TempDir(), nil); err == nil {
+		t.Fatal("SaveLoginTokenData(nil) succeeded")
+	}
+
+	futureDir := t.TempDir()
+	writeFutureProfilesForLoginPreflight(t, futureDir)
+	err := SaveLoginTokenData(futureDir, &TokenData{AccessToken: "fresh"})
+	if err == nil || !strings.Contains(err.Error(), "newer than supported") {
+		t.Fatalf("SaveLoginTokenData(future schema) error = %v", err)
+	}
+
+	t.Run("nil identity token", func(t *testing.T) {
+		isolateTokenProfileLoadHooks(t)
+		tokenLoadKeychainIdentity = func(string, string) (*TokenData, error) { return nil, nil }
+		_, err := tokenLoadProfileIdentity(Profile{CorpID: "corp_nil_identity", UserID: "user_nil_identity"})
+		if !errors.Is(err, ErrTokenDataNotFound) {
+			t.Fatalf("tokenLoadProfileIdentity() error = %v", err)
+		}
+	})
+
+	t.Run("nil organization fallback", func(t *testing.T) {
+		isolateTokenProfileLoadHooks(t)
+		tokenLoadKeychainForCorpID = func(string) (*TokenData, error) { return nil, nil }
+		_, err := tokenLoadProfileIdentity(Profile{CorpID: "corp_nil_org", UserID: "user_nil_org"})
+		if !errors.Is(err, ErrTokenDataNotFound) {
+			t.Fatalf("tokenLoadProfileIdentity() error = %v", err)
+		}
+	})
+
+	t.Run("organization fallback belongs to another organization", func(t *testing.T) {
+		isolateTokenProfileLoadHooks(t)
+		tokenLoadKeychainForCorpID = func(string) (*TokenData, error) {
+			return &TokenData{AccessToken: "wrong", CorpID: "corp_other", UserID: "user_wrong_org"}, nil
+		}
+		_, err := tokenLoadProfileIdentity(Profile{CorpID: "corp_expected", UserID: "user_wrong_org"})
+		if err == nil || !strings.Contains(err.Error(), "contains token for corpId") {
+			t.Fatalf("tokenLoadProfileIdentity() error = %v", err)
+		}
+	})
+
+	t.Run("identity repair write failure", func(t *testing.T) {
+		isolateTokenProfileLoadHooks(t)
+		failure := errors.New("identity repair write failure")
+		const corpID, userID = "corp_identity_save", "user_identity_save"
+		tokenLoadKeychainForCorpID = func(string) (*TokenData, error) {
+			return &TokenData{AccessToken: "organization", CorpID: corpID, UserID: userID}, nil
+		}
+		tokenSaveKeychainForIdentity = func(string, string, *TokenData) error { return failure }
+		_, err := tokenLoadProfileIdentity(Profile{CorpID: corpID, UserID: userID})
+		if !errors.Is(err, failure) {
+			t.Fatalf("tokenLoadProfileIdentity() error = %v, want %v", err, failure)
+		}
+	})
+}
+
+func isolateHalfMigratedRepairHooks(t *testing.T) {
+	t.Helper()
+	oldLoadLegacy := profilesLoadLegacy
+	oldLoadCorp := profilesLoadCorp
+	oldLoadIdentity := profilesLoadIdentity
+	oldSaveCorp := profilesSaveCorp
+	oldSaveIdentity := profilesSaveIdentity
+	t.Cleanup(func() {
+		profilesLoadLegacy = oldLoadLegacy
+		profilesLoadCorp = oldLoadCorp
+		profilesLoadIdentity = oldLoadIdentity
+		profilesSaveCorp = oldSaveCorp
+		profilesSaveIdentity = oldSaveIdentity
+	})
+	profilesLoadLegacy = func() (*TokenData, error) { return nil, ErrTokenDataNotFound }
+	profilesLoadCorp = func(string) (*TokenData, error) { return nil, ErrTokenDataNotFound }
+	profilesLoadIdentity = func(string, string) (*TokenData, error) { return nil, ErrTokenDataNotFound }
+	profilesSaveCorp = func(string, *TokenData) error { return nil }
+	profilesSaveIdentity = func(string, string, *TokenData) error { return nil }
+}
+
+func isolateTokenProfileLoadHooks(t *testing.T) {
+	t.Helper()
+	oldLoadIdentity := tokenLoadKeychainIdentity
+	oldLoadCorp := tokenLoadKeychainForCorpID
+	oldSaveIdentity := tokenSaveKeychainForIdentity
+	t.Cleanup(func() {
+		tokenLoadKeychainIdentity = oldLoadIdentity
+		tokenLoadKeychainForCorpID = oldLoadCorp
+		tokenSaveKeychainForIdentity = oldSaveIdentity
+	})
+	tokenLoadKeychainIdentity = func(string, string) (*TokenData, error) { return nil, ErrTokenDataNotFound }
+	tokenLoadKeychainForCorpID = func(string) (*TokenData, error) { return nil, ErrTokenDataNotFound }
+	tokenSaveKeychainForIdentity = func(string, string, *TokenData) error { return nil }
+}

internal/auth/blank_profile_selector_regression_test.go

@@ -0,0 +1,357 @@
+// Copyright 2026 Alibaba Group
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package auth
+
+import (
+	"encoding/json"
+	"os"
+	"strings"
+	"testing"
+)
+
+type blankProfileSelectorFixture struct {
+	configDir     string
+	corpID        string
+	blankName     string
+	blankSelector string
+	exactUserID   string
+	exactSelector string
+	blankToken    *TokenData
+	exactToken    *TokenData
+}
+
+func seedBlankProfileSelectorFixture(
+	t *testing.T,
+	blankName string,
+	corpName string,
+	blankCurrent bool,
+) blankProfileSelectorFixture {
+	t.Helper()
+	cleanupKeychain(t)
+
+	configDir := t.TempDir()
+	corpID := "corp_selector_fixture"
+	exactUserID := "identity_exact_fixture"
+	exactSelector := profileSelector(corpID, exactUserID)
+
+	blankToken := testToken("at_unresolved_fixture", corpID, corpName)
+	blankToken.UserID = ""
+	blankToken.UserName = ""
+	exactToken := testToken("at_exact_fixture", corpID, corpName)
+	exactToken.UserID = exactUserID
+	exactToken.UserName = "Exact Fixture Account"
+
+	if err := SaveTokenDataKeychainForCorpID(corpID, blankToken); err != nil {
+		t.Fatalf("SaveTokenDataKeychainForCorpID(blank) error = %v", err)
+	}
+	if err := SaveTokenDataKeychainForIdentity(corpID, exactUserID, exactToken); err != nil {
+		t.Fatalf("SaveTokenDataKeychainForIdentity(exact) error = %v", err)
+	}
+	if err := SaveTokenDataKeychain(exactToken); err != nil {
+		t.Fatalf("SaveTokenDataKeychain(exact mirror) error = %v", err)
+	}
+	if err := WriteTokenMarker(configDir); err != nil {
+		t.Fatalf("WriteTokenMarker() error = %v", err)
+	}
+
+	cfg := &ProfilesConfig{
+		Version:        profilesVersion,
+		PrimaryProfile: exactSelector,
+		OrgCurrentProfiles: map[string]string{
+			corpID: exactSelector,
+		},
+		Profiles: []Profile{
+			{
+				Name:     blankName,
+				CorpID:   corpID,
+				CorpName: corpName,
+				Status:   ProfileStatusActive,
+			},
+			{
+				Name:     "Exact Fixture Account",
+				CorpID:   corpID,
+				CorpName: corpName,
+				UserID:   exactUserID,
+				UserName: "Exact Fixture Account",
+				Status:   ProfileStatusActive,
+			},
+		},
+	}
+	blankSelector := ProfileSelectionSelector(cfg.Profiles[0], cfg)
+	persistedBlankPointer := strings.TrimSpace(blankName)
+	if selectorConflictsWithOrganizationGrammar(cfg, persistedBlankPointer) {
+		// An ambiguous local name has always been captured by CorpId/CorpName
+		// grammar in the public resolver. New writers must use the reserved
+		// selector to preserve exact blank-profile intent without changing that
+		// precedence.
+		persistedBlankPointer = blankSelector
+		if _, reserved := parseUnresolvedProfileSelector(persistedBlankPointer); reserved {
+			cfg.Version = profilesUnresolvedSelectorVersion
+		}
+	}
+	cfg.CurrentProfile = exactSelector
+	cfg.PreviousProfile = persistedBlankPointer
+	if blankCurrent {
+		cfg.CurrentProfile = persistedBlankPointer
+		cfg.PreviousProfile = exactSelector
+	}
+	data, err := json.MarshalIndent(cfg, "", "  ")
+	if err != nil {
+		t.Fatalf("json.MarshalIndent(profiles) error = %v", err)
+	}
+	data = append(data, '\n')
+	if err := os.WriteFile(ProfilesPath(configDir), data, 0o600); err != nil {
+		t.Fatalf("os.WriteFile(profiles.json) error = %v", err)
+	}
+
+	return blankProfileSelectorFixture{
+		configDir:     configDir,
+		corpID:        corpID,
+		blankName:     blankName,
+		blankSelector: blankSelector,
+		exactUserID:   exactUserID,
+		exactSelector: exactSelector,
+		blankToken:    blankToken,
+		exactToken:    exactToken,
+	}
+}
+
+func TestCrossPlatformCoverageBlankProfileNameMatchingCorpNameResolvesCurrentProfileExactly(t *testing.T) {
+	fixture := seedBlankProfileSelectorFixture(t, "Fixture Organization", "Fixture Organization", true)
+
+	selected, exact, err := ResolveProfileWithScope(fixture.configDir, "")
+	if err != nil {
+		t.Fatalf("ResolveProfileWithScope(current) error = %v", err)
+	}
+	if selected == nil || selected.CorpID != fixture.corpID || selected.UserID != "" {
+		t.Fatalf("resolved current profile = %#v, want unresolved profile", selected)
+	}
+	if !exact {
+		t.Fatal("current local-name selector should resolve one exact unresolved profile")
+	}
+	cfg, err := LoadProfiles(fixture.configDir)
+	if err != nil {
+		t.Fatalf("LoadProfiles() error = %v", err)
+	}
+	if cfg.CurrentProfile != fixture.blankSelector {
+		t.Fatalf("current profile = %q, want stable unresolved selector %q", cfg.CurrentProfile, fixture.blankSelector)
+	}
+}
+
+func TestCrossPlatformCoverageBlankProfileNameMatchingCorpNameLoadsOrganizationToken(t *testing.T) {
+	fixture := seedBlankProfileSelectorFixture(t, "Fixture Organization", "Fixture Organization", false)
+
+	if fixture.blankSelector == fixture.blankName || fixture.blankSelector == fixture.corpID {
+		t.Fatalf("unsafe blank selector = %q, want reserved exact selector", fixture.blankSelector)
+	}
+	loaded, err := LoadTokenDataForProfile(fixture.configDir, fixture.blankSelector)
+	if err != nil {
+		t.Fatalf("LoadTokenDataForProfile(blank local name) error = %v", err)
+	}
+	if loaded.UserID != "" || loaded.AccessToken != fixture.blankToken.AccessToken {
+		t.Fatalf("loaded token = %#v, want unresolved organization token", loaded)
+	}
+}
+
+func TestCrossPlatformCoverageBlankProfileNameMatchingCorpNameRoundTripsPreviousProfile(t *testing.T) {
+	fixture := seedBlankProfileSelectorFixture(t, "Fixture Organization", "Fixture Organization", false)
+
+	selected, err := UsePreviousProfile(fixture.configDir)
+	if err != nil {
+		t.Fatalf("UsePreviousProfile() error = %v", err)
+	}
+	if selected == nil || selected.CorpID != fixture.corpID || selected.UserID != "" {
+		t.Fatalf("selected previous profile = %#v, want unresolved profile", selected)
+	}
+
+	cfg, err := LoadProfiles(fixture.configDir)
+	if err != nil {
+		t.Fatalf("LoadProfiles() error = %v", err)
+	}
+	if cfg.CurrentProfile != fixture.blankSelector || cfg.PreviousProfile != fixture.exactSelector {
+		t.Fatalf(
+			"profile pointers = current %q previous %q, want %q and %q",
+			cfg.CurrentProfile,
+			cfg.PreviousProfile,
+			fixture.blankSelector,
+			fixture.exactSelector,
+		)
+	}
+}
+
+func TestCrossPlatformCoverageBlankProfileNameMatchingCorpNameDeletesOnlyUnresolvedProfile(t *testing.T) {
+	fixture := seedBlankProfileSelectorFixture(t, "Fixture Organization", "Fixture Organization", false)
+
+	if err := DeleteTokenDataForProfile(fixture.configDir, fixture.blankSelector); err != nil {
+		t.Fatalf("DeleteTokenDataForProfile(blank local name) error = %v", err)
+	}
+
+	cfg, err := LoadProfiles(fixture.configDir)
+	if err != nil {
+		t.Fatalf("LoadProfiles() error = %v", err)
+	}
+	if len(cfg.Profiles) != 1 || cfg.Profiles[0].CorpID != fixture.corpID || cfg.Profiles[0].UserID != fixture.exactUserID {
+		t.Fatalf("profiles after blank deletion = %#v, want only exact account", cfg.Profiles)
+	}
+	loaded, err := LoadTokenDataForProfile(fixture.configDir, fixture.exactSelector)
+	if err != nil {
+		t.Fatalf("LoadTokenDataForProfile(exact after blank deletion) error = %v", err)
+	}
+	if loaded.UserID != fixture.exactUserID || loaded.AccessToken != fixture.exactToken.AccessToken {
+		t.Fatalf("exact token after blank deletion = %#v, want exact account preserved", loaded)
+	}
+}
+
+func TestCrossPlatformCoverageBlankProfileNameContainingColonWinsOverIdentitySyntax(t *testing.T) {
+	fixture := seedBlankProfileSelectorFixture(t, "legacy:outsourced", "Fixture Organization", true)
+
+	if fixture.blankSelector == fixture.blankName {
+		t.Fatalf("colon-containing name leaked as selector %q", fixture.blankSelector)
+	}
+	if _, _, parsedAsIdentity := ParseIdentitySelector(fixture.blankSelector); parsedAsIdentity {
+		t.Fatalf("stable blank selector %q was parsed as an identity", fixture.blankSelector)
+	}
+	selected, exact, err := ResolveProfileWithScope(fixture.configDir, "")
+	if err != nil {
+		t.Fatalf("ResolveProfileWithScope(colon local name) error = %v", err)
+	}
+	if selected == nil || selected.CorpID != fixture.corpID || selected.UserID != "" {
+		t.Fatalf("resolved colon-name profile = %#v, want unresolved profile", selected)
+	}
+	if !exact {
+		t.Fatal("colon-containing local name should resolve one exact unresolved profile")
+	}
+
+	cfg, err := LoadProfiles(fixture.configDir)
+	if err != nil {
+		t.Fatalf("LoadProfiles() error = %v", err)
+	}
+	if cfg.CurrentProfile != fixture.blankSelector {
+		t.Fatalf("current profile = %q, want migrated colon-name selector %q", cfg.CurrentProfile, fixture.blankSelector)
+	}
+
+	loaded, err := LoadTokenDataForProfile(fixture.configDir, fixture.blankSelector)
+	if err != nil {
+		t.Fatalf("LoadTokenDataForProfile(colon local name) error = %v", err)
+	}
+	if loaded.UserID != "" || loaded.AccessToken != fixture.blankToken.AccessToken {
+		t.Fatalf("loaded colon-name token = %#v, want unresolved organization token", loaded)
+	}
+}
+
+func TestCrossPlatformCoverageRealExactSelectorWinsOverMatchingBlankLegacyName(t *testing.T) {
+	const exactSelector = "corp_selector_fixture:identity_exact_fixture"
+	fixture := seedBlankProfileSelectorFixture(t, exactSelector, "Fixture Organization", true)
+
+	selected, exact, err := ResolveProfileWithScope(fixture.configDir, "")
+	if err != nil {
+		t.Fatalf("ResolveProfileWithScope(current exact collision) error = %v", err)
+	}
+	if selected == nil || selected.UserID != fixture.exactUserID || !exact {
+		t.Fatalf("resolved current collision = %#v exact=%v, want real exact identity", selected, exact)
+	}
+	cfg, err := LoadProfiles(fixture.configDir)
+	if err != nil {
+		t.Fatalf("LoadProfiles() error = %v", err)
+	}
+	if cfg.CurrentProfile != exactSelector {
+		t.Fatalf("current collision selector = %q, want real exact %q", cfg.CurrentProfile, exactSelector)
+	}
+
+	blank, err := LoadTokenDataForProfile(fixture.configDir, fixture.blankSelector)
+	if err != nil {
+		t.Fatalf("LoadTokenDataForProfile(reserved blank collision) error = %v", err)
+	}
+	if blank.UserID != "" || blank.AccessToken != fixture.blankToken.AccessToken {
+		t.Fatalf("reserved blank collision token = %#v", blank)
+	}
+	if err := DeleteTokenDataForProfile(fixture.configDir, fixture.blankSelector); err != nil {
+		t.Fatalf("DeleteTokenDataForProfile(reserved blank collision) error = %v", err)
+	}
+	exactToken, err := LoadTokenDataForProfile(fixture.configDir, exactSelector)
+	if err != nil || exactToken.UserID != fixture.exactUserID {
+		t.Fatalf("exact identity after blank collision delete = %#v, %v", exactToken, err)
+	}
+}
+
+func TestCrossPlatformCoverageUnrelatedProfileDeletionPreservesBlankOrganizationCurrentMapping(t *testing.T) {
+	fixture := seedBlankProfileSelectorFixture(t, "Fixture Organization", "Fixture Organization", true)
+	cfg, err := LoadProfiles(fixture.configDir)
+	if err != nil {
+		t.Fatalf("LoadProfiles() error = %v", err)
+	}
+	cfg.Profiles = append(cfg.Profiles, Profile{
+		Name:   "Unrelated Account",
+		CorpID: "corp_unrelated_fixture",
+		UserID: "identity_unrelated_fixture",
+		Status: ProfileStatusActive,
+	})
+	if err := SaveProfiles(fixture.configDir, cfg); err != nil {
+		t.Fatalf("SaveProfiles(unrelated) error = %v", err)
+	}
+	if _, err := RemoveProfile(fixture.configDir, "corp_unrelated_fixture:identity_unrelated_fixture"); err != nil {
+		t.Fatalf("RemoveProfile(unrelated) error = %v", err)
+	}
+
+	cfg, err = LoadProfiles(fixture.configDir)
+	if err != nil {
+		t.Fatalf("LoadProfiles(after unrelated delete) error = %v", err)
+	}
+	if cfg.CurrentProfile != fixture.blankSelector {
+		t.Fatalf("blank current after unrelated delete = %q, want %q", cfg.CurrentProfile, fixture.blankSelector)
+	}
+	if got := cfg.OrgCurrentProfiles[fixture.corpID]; got != fixture.exactSelector {
+		t.Fatalf("organization current after unrelated delete = %q, want %q", got, fixture.exactSelector)
+	}
+	selected, err := ResolveProfile(fixture.configDir, fixture.corpID)
+	if err != nil || selected.UserID != fixture.exactUserID {
+		t.Fatalf("organization selector after unrelated delete = %#v, %v", selected, err)
+	}
+}
+
+func TestCrossPlatformCoverageSameCorpNonCurrentDeletionPreservesExactOrganizationCurrent(t *testing.T) {
+	fixture := seedBlankProfileSelectorFixture(t, "Fixture Organization", "Fixture Organization", true)
+	cfg, err := LoadProfiles(fixture.configDir)
+	if err != nil {
+		t.Fatalf("LoadProfiles() error = %v", err)
+	}
+	cfg.Profiles = append(cfg.Profiles, Profile{
+		Name:     "Another Exact Account",
+		CorpID:   fixture.corpID,
+		CorpName: "Fixture Organization",
+		UserID:   "identity_noncurrent_fixture",
+		Status:   ProfileStatusActive,
+	})
+	if err := SaveProfiles(fixture.configDir, cfg); err != nil {
+		t.Fatalf("SaveProfiles(non-current exact) error = %v", err)
+	}
+	if _, err := RemoveProfile(fixture.configDir, fixture.corpID+":identity_noncurrent_fixture"); err != nil {
+		t.Fatalf("RemoveProfile(non-current exact) error = %v", err)
+	}
+
+	cfg, err = LoadProfiles(fixture.configDir)
+	if err != nil {
+		t.Fatalf("LoadProfiles(after same-corp delete) error = %v", err)
+	}
+	if cfg.CurrentProfile != fixture.blankSelector {
+		t.Fatalf("blank current after same-corp delete = %q, want %q", cfg.CurrentProfile, fixture.blankSelector)
+	}
+	if got := cfg.OrgCurrentProfiles[fixture.corpID]; got != fixture.exactSelector {
+		t.Fatalf("organization current after same-corp delete = %q, want %q", got, fixture.exactSelector)
+	}
+	selected, err := ResolveProfile(fixture.configDir, fixture.corpID)
+	if err != nil || selected.UserID != fixture.exactUserID {
+		t.Fatalf("organization selector after same-corp delete = %#v, %v", selected, err)
+	}
+}

internal/auth/coverage_edges_test.go

@@ -3159,7 +3159,13 @@ func TestCrossPlatformCoverageMultiAccountSelectorAndIdentityLoadEdges(t *testin
 	if got, err := loadTokenForProfileIdentity(profile); err != nil || got.AccessToken != "mirror" {
 		t.Fatalf("identity repair = %#v %v", got, err)
 	}
-	if _, err := loadTokenForProfileIdentity(Profile{CorpID: "corp-a"}); err != nil {
-		t.Fatalf("organization-only token load = %v", err)
+	if _, err := loadTokenForProfileIdentity(Profile{CorpID: "corp-a"}); err == nil {
+		t.Fatal("unresolved profile loaded organization token owned by an exact identity")
+	}
+	profilesLoadCorp = func(string) (*TokenData, error) {
+		return &TokenData{CorpID: "corp-a", AccessToken: "organization"}, nil
+	}
+	if got, err := loadTokenForProfileIdentity(Profile{CorpID: "corp-a"}); err != nil || got.AccessToken != "organization" {
+		t.Fatalf("organization-only token load = %#v %v", got, err)
 	}
 }

internal/auth/device_flow.go

@@ -195,7 +195,7 @@ func (p *DeviceFlowProvider) resetCredentialState() {
 }
 
 func (p *DeviceFlowProvider) Login(ctx context.Context) (*TokenData, error) {
-	if err := preflightTokenPersistence(p.configDir); err != nil {
+	if err := prepareLoginPersistence(p.configDir); err != nil {
 		return nil, fmt.Errorf("%s: %w", i18n.T("本地登录态无法安全更新"), err)
 	}
 
@@ -357,6 +357,9 @@ func (p *DeviceFlowProvider) loginOnce(ctx context.Context, attempt int) (*Token
 	if err := oauthProvider.prepareLoginToken(ctx, tokenData); err != nil {
 		return nil, fmt.Errorf("%s: %w", i18n.T("保存 token 失败"), err)
 	}
+	if err := preflightTokenWritePersistence(p.configDir, tokenData); err != nil {
+		return nil, fmt.Errorf("%s: %w", i18n.T("本地登录态无法安全更新"), err)
+	}
 	if err := deviceSaveToken(p.configDir, tokenData); err != nil {
 		return nil, fmt.Errorf("%s: %w", i18n.T("保存 token 失败"), err)
 	}

internal/auth/keychain_store.go

@@ -138,11 +138,177 @@ func loadTokenDataKeychainAccount(account string) (*TokenData, error) {
 	return &data, nil
 }
 
-// preflightTokenPersistence verifies that every registered token slot can be
-// read before an OAuth login or exchange can target any profile.
-// A missing slot is safe (first login or a legacy fallback); any other error
-// stops the remote operation when existing ciphertext is already known to be
-// unreadable and therefore unsafe to update.
+// prepareLoginPersistence rejects profile registries written by a newer client
+// and protects the legacy global mirror before a new authorization flow
+// performs remote work. Version-1 registries keep using the existing full
+// migration in saveTokenDataLocked before any compatibility mirror is
+// overwritten.
+//
+// For v2/v3, only the organization referenced by the readable global mirror is
+// inspected. A uniquely matching half-migrated profile is repaired from that
+// mirror under the profiles lock. Missing or damaged slots in unrelated
+// organizations and orphan inventory are deliberately not scanned.
+func prepareLoginPersistence(configDir string) error {
+	if h := edition.Get(); h.SaveToken != nil {
+		return nil
+	}
+	return withProfilesLock(configDir, func() error {
+		cfg, err := profilesLoad(configDir)
+		if err != nil {
+			return fmt.Errorf("load token profiles: %w", err)
+		}
+		if err := ensureProfilesWritable(cfg); err != nil {
+			return err
+		}
+		return repairHalfMigratedGlobalTokenLocked(cfg)
+	})
+}
+
+// repairHalfMigratedGlobalTokenLocked preserves the only readable copy left by
+// an interrupted v1.0.53 migration. The caller must hold the profiles lock.
+func repairHalfMigratedGlobalTokenLocked(cfg *ProfilesConfig) error {
+	if cfg == nil || cfg.Version < profilesVersion || len(cfg.Profiles) == 0 {
+		return nil
+	}
+
+	global, err := profilesLoadLegacy()
+	if errors.Is(err, ErrTokenDataNotFound) {
+		return nil
+	}
+	if err != nil {
+		return fmt.Errorf(
+			"legacy token slot %q is unreadable; refusing to overwrite a potentially unique old login: %w",
+			keychain.AccountToken,
+			err,
+		)
+	}
+	if global == nil {
+		return nil
+	}
+	corpID := strings.TrimSpace(global.CorpID)
+	if corpID == "" {
+		return nil
+	}
+	profiles := profilesForCorpID(cfg, corpID)
+	if len(profiles) == 0 {
+		// A readable global token for an unregistered organization is an orphan,
+		// not a profile credential that this registry still promises to retain.
+		return nil
+	}
+
+	orgToken, orgErr := profilesLoadCorp(corpID)
+	allCanonical := true
+	for _, profile := range profiles {
+		if !loginProfileHasUsableCanonicalToken(profile, orgToken, orgErr) {
+			allCanonical = false
+			break
+		}
+	}
+	if allCanonical {
+		return nil
+	}
+
+	profile := uniqueV2GlobalRepairProfile(cfg, corpID)
+	if profile == nil {
+		return fmt.Errorf(
+			"legacy token slot %q may be the only recoverable login for one of %d accounts in organization %q; refusing to overwrite it until each account has a usable identity slot",
+			keychain.AccountToken,
+			len(profiles),
+			corpID,
+		)
+	}
+	userID := strings.TrimSpace(profile.UserID)
+	if userID != "" &&
+		orgErr == nil &&
+		loginTokenHasCredentialMaterial(orgToken) &&
+		legacyTokenMatchesV2RepairProfile(orgToken, profile) {
+		if err := repairLoginIdentityToken(profile, orgToken); err != nil {
+			return err
+		}
+		return nil
+	}
+	if !legacyTokenMatchesV2RepairProfile(global, profile) {
+		return fmt.Errorf(
+			"legacy token slot %q does not safely match the only profile in organization %q; refusing to overwrite a potentially unique old login",
+			keychain.AccountToken,
+			corpID,
+		)
+	}
+	if !loginTokenHasCredentialMaterial(global) {
+		return fmt.Errorf(
+			"legacy token slot %q has no recoverable credential material for organization %q; refusing to overwrite a potentially unique old login",
+			keychain.AccountToken,
+			corpID,
+		)
+	}
+
+	// The matching global token is the only recoverable copy. Overwrite a
+	// damaged organization slot as well as filling a missing one.
+	if err := profilesSaveCorp(corpID, global); err != nil {
+		return fmt.Errorf("repair organization token slot %q: %w", TokenAccountForCorpID(corpID), err)
+	}
+	if userID == "" {
+		return nil
+	}
+	return repairLoginIdentityToken(profile, global)
+}
+
+func loginProfileHasUsableCanonicalToken(
+	profile *Profile,
+	orgToken *TokenData,
+	orgErr error,
+) bool {
+	if profile == nil {
+		return false
+	}
+	corpID := strings.TrimSpace(profile.CorpID)
+	userID := strings.TrimSpace(profile.UserID)
+	if userID == "" {
+		return orgErr == nil &&
+			loginTokenHasCredentialMaterial(orgToken) &&
+			strings.TrimSpace(orgToken.CorpID) == corpID &&
+			strings.TrimSpace(orgToken.UserID) == ""
+	}
+	identity, err := profilesLoadIdentity(corpID, userID)
+	if err == nil &&
+		loginTokenHasCredentialMaterial(identity) &&
+		strings.TrimSpace(identity.CorpID) == corpID &&
+		strings.TrimSpace(identity.UserID) == userID {
+		return true
+	}
+	return false
+}
+
+func loginTokenHasCredentialMaterial(data *TokenData) bool {
+	return data != nil &&
+		(strings.TrimSpace(data.AccessToken) != "" ||
+			strings.TrimSpace(data.RefreshToken) != "" ||
+			strings.TrimSpace(data.PersistentCode) != "")
+}
+
+func repairLoginIdentityToken(profile *Profile, source *TokenData) error {
+	corpID := strings.TrimSpace(profile.CorpID)
+	userID := strings.TrimSpace(profile.UserID)
+	identityToken := source
+	if strings.TrimSpace(source.UserID) == "" {
+		enriched := *source
+		enriched.UserID = userID
+		if strings.TrimSpace(enriched.UserName) == "" {
+			enriched.UserName = strings.TrimSpace(profile.UserName)
+		}
+		identityToken = &enriched
+	}
+	if err := profilesSaveIdentity(corpID, userID, identityToken); err != nil {
+		return fmt.Errorf("repair identity token slot %q: %w", TokenAccountForIdentity(corpID, userID), err)
+	}
+	return nil
+}
+
+// preflightTokenPersistence verifies every persisted token slot, including
+// unregistered/orphan ciphertext. Keep this full-inventory validator for
+// migration, export and explicit storage diagnostics; login must use the
+// schema-only and target-only preflights instead so an unrelated damaged
+// account cannot block reauthorization.
 func preflightTokenPersistence(configDir string) error {
 	if h := edition.Get(); h.SaveToken != nil {
 		return nil
@@ -191,10 +357,11 @@ func preflightTokenPersistence(configDir string) error {
 	return nil
 }
 
-// preflightTokenRefreshPersistence checks only the slots a refresh can write.
-// An unrelated broken profile must not prevent the current profile from using
-// its still-valid credentials.
-func preflightTokenRefreshPersistence(configDir string, data *TokenData) error {
+// preflightTokenWritePersistence checks only the slots SaveTokenData can write
+// for data under the current runtime selector. It is shared by login and
+// refresh so both paths stay aligned with the same identity/org/global mirror
+// isolation rules.
+func preflightTokenWritePersistence(configDir string, data *TokenData) error {
 	if h := edition.Get(); h.SaveToken != nil {
 		return nil
 	}
@@ -206,33 +373,46 @@ func preflightTokenRefreshPersistence(configDir string, data *TokenData) error {
 		return err
 	}
 
-	if _, err := LoadTokenDataKeychain(); err != nil && !errors.Is(err, ErrTokenDataNotFound) {
-		return fmt.Errorf("legacy token slot %q is unreadable: %w", keychain.AccountToken, err)
+	plan := planTokenPersistenceWrites(cfg, data, RuntimeProfile())
+	if err := validateTokenPersistenceWritePlan(cfg, data, plan); err != nil {
+		return err
 	}
-	if data == nil || strings.TrimSpace(data.CorpID) == "" {
-		return nil
-	}
-	corpID := strings.TrimSpace(data.CorpID)
-	userID := strings.TrimSpace(data.UserID)
-	if userID != "" {
-		if _, err := LoadTokenDataKeychainForIdentity(corpID, userID); err != nil && !errors.Is(err, ErrTokenDataNotFound) {
-			return fmt.Errorf("identity token slot %q is unreadable: %w", TokenAccountForIdentity(corpID, userID), err)
+	if plan.WriteGlobal {
+		if _, err := LoadTokenDataKeychain(); err != nil && !errors.Is(err, ErrTokenDataNotFound) {
+			return fmt.Errorf("legacy token slot %q is unreadable: %w", keychain.AccountToken, err)
 		}
 	}
-	checkOrganizationMirror := true
-	if _, _, exact := ParseIdentitySelector(RuntimeProfile()); exact {
-		checkOrganizationMirror =
-			exactProfileSelectorForCorp(cfg, corpID, cfg.OrgCurrentProfiles[corpID]) ==
-				profileSelector(corpID, userID)
+	if plan.CorpID == "" {
+		return nil
 	}
-	if checkOrganizationMirror {
-		if _, err := LoadTokenDataKeychainForCorpID(corpID); err != nil && !errors.Is(err, ErrTokenDataNotFound) {
-			return fmt.Errorf("profile token slot %q is unreadable: %w", TokenAccountForCorpID(corpID), err)
+	if plan.WriteIdentity {
+		if _, err := LoadTokenDataKeychainForIdentity(plan.CorpID, plan.UserID); err != nil && !errors.Is(err, ErrTokenDataNotFound) {
+			return fmt.Errorf(
+				"identity token slot %q is unreadable: %w",
+				TokenAccountForIdentity(plan.CorpID, plan.UserID),
+				err,
+			)
+		}
+	}
+	if plan.WriteOrganization {
+		if _, err := LoadTokenDataKeychainForCorpID(plan.CorpID); err != nil && !errors.Is(err, ErrTokenDataNotFound) {
+			return fmt.Errorf(
+				"profile token slot %q is unreadable: %w",
+				TokenAccountForCorpID(plan.CorpID),
+				err,
+			)
 		}
 	}
 	return nil
 }
 
+// preflightTokenRefreshPersistence checks only the slots a refresh can write.
+// An unrelated broken profile must not prevent the current profile from using
+// its still-valid credentials.
+func preflightTokenRefreshPersistence(configDir string, data *TokenData) error {
+	return preflightTokenWritePersistence(configDir, data)
+}
+
 // DeleteTokenDataKeychain removes TokenData from the platform keychain.
 func DeleteTokenDataKeychain() error {
 	return authKeychainRemove(keychain.Service, keychain.AccountToken)

internal/auth/legacy_compat_coverage_gaps_test.go

@@ -0,0 +1,288 @@
+// Copyright 2026 Alibaba Group
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+// http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package auth
+
+import (
+	"errors"
+	"testing"
+)
+
+func TestCrossPlatformCoverageLegacySelectorCompatibilityEdges(t *testing.T) {
+	upgradeDir := t.TempDir()
+	upgradeCfg := &ProfilesConfig{
+		Version: profilesVersion,
+		Profiles: []Profile{{
+			Name:   "Legacy Organization",
+			CorpID: "corp_upgrade_fixture",
+		}},
+	}
+	if err := upsertProfileFromToken(upgradeDir, upgradeCfg, &TokenData{
+		CorpID:   "corp_upgrade_fixture",
+		UserID:   "identity_upgrade_fixture",
+		UserName: "Upgraded Account",
+	}, false); err != nil {
+		t.Fatalf("upsertProfileFromToken(upgrade legacy profile) error = %v", err)
+	}
+	if len(upgradeCfg.Profiles) != 1 || upgradeCfg.Profiles[0].UserID != "identity_upgrade_fixture" {
+		t.Fatalf("upgraded profiles = %#v", upgradeCfg.Profiles)
+	}
+
+	renameDir := t.TempDir()
+	renameCfg := &ProfilesConfig{
+		Version: profilesVersion,
+		Profiles: []Profile{
+			{Name: "duplicate", CorpID: "corp_rename_fixture"},
+			{Name: "duplicate", CorpID: "corp_rename_fixture", UserID: "identity_exact_fixture"},
+		},
+	}
+	if err := upsertProfileFromToken(renameDir, renameCfg, &TokenData{
+		CorpID:   "corp_rename_fixture",
+		CorpName: "Renamed Organization",
+	}, false); err != nil {
+		t.Fatalf("upsertProfileFromToken(rename blank profile) error = %v", err)
+	}
+	if renameCfg.Profiles[0].Name != "Renamed Organization" {
+		t.Fatalf("blank profile name = %q, want conflict-free organization name", renameCfg.Profiles[0].Name)
+	}
+
+	blank := Profile{CorpID: "corp_selector_fixture"}
+	if got := storedProfileSelector(nil, nil); got != "" {
+		t.Fatalf("storedProfileSelector(nil profile) = %q", got)
+	}
+	if got := storedProfileSelector(nil, &blank); got != blank.CorpID {
+		t.Fatalf("storedProfileSelector(nil config) = %q", got)
+	}
+	if localProfileSelectorIsSafe(nil, &blank, "local") ||
+		localProfileSelectorIsSafe(&ProfilesConfig{}, nil, "local") {
+		t.Fatal("nil selector inputs were treated as safe")
+	}
+	if got := unresolvedProfileSelector("   "); got != "" {
+		t.Fatalf("unresolvedProfileSelector(blank) = %q", got)
+	}
+	if _, ok := parseUnresolvedProfileSelector(unresolvedProfileSelectorPrefix + "!"); ok {
+		t.Fatal("invalid base64 legacy selector parsed successfully")
+	}
+	if _, ok := parseUnresolvedProfileSelector(unresolvedProfileSelectorPrefix + "IA"); ok {
+		t.Fatal("blank decoded legacy selector parsed successfully")
+	}
+
+	previousRuntime := RuntimeProfile()
+	SetRuntimeProfile("")
+	t.Cleanup(func() { SetRuntimeProfile(previousRuntime) })
+	if got := StableTokenProfileSelector(t.TempDir(), nil); got != "" {
+		t.Fatalf("StableTokenProfileSelector(nil) = %q", got)
+	}
+
+	ambiguousDir := t.TempDir()
+	ambiguousCfg := &ProfilesConfig{
+		Version:        profilesVersion,
+		CurrentProfile: "corp_ambiguous_fixture",
+		Profiles: []Profile{
+			{Name: "First", CorpID: "corp_ambiguous_fixture", UserID: "identity_first_fixture"},
+			{Name: "Second", CorpID: "corp_ambiguous_fixture", UserID: "identity_second_fixture"},
+		},
+	}
+	if err := SaveProfiles(ambiguousDir, ambiguousCfg); err != nil {
+		t.Fatalf("SaveProfiles(ambiguous current) error = %v", err)
+	}
+	ambiguousToken := &TokenData{CorpID: "corp_ambiguous_fixture", UserID: "identity_first_fixture"}
+	if got, want := StableTokenProfileSelector(ambiguousDir, ambiguousToken), "corp_ambiguous_fixture:identity_first_fixture"; got != want {
+		t.Fatalf("StableTokenProfileSelector(ambiguous organization) = %q, want %q", got, want)
+	}
+
+	reserved := unresolvedProfileSelector("corp_missing_fixture")
+	emptyCfg := &ProfilesConfig{}
+	if _, _, err := resolveProfileSelection("", emptyCfg, reserved); err == nil {
+		t.Fatal("missing reserved profile selection succeeded")
+	}
+	if _, _, err := resolveProfileDeletionSelection(emptyCfg, reserved); err == nil {
+		t.Fatal("missing reserved profile deletion succeeded")
+	}
+	if got := canonicalStoredSelector(emptyCfg, reserved); got != "" {
+		t.Fatalf("canonical missing reserved selector = %q", got)
+	}
+	if !selectorTargetsCorp(reserved, "corp_missing_fixture") {
+		t.Fatal("reserved selector did not target its organization")
+	}
+
+	localCfg := &ProfilesConfig{Profiles: []Profile{{
+		Name:   "local-profile-fixture",
+		CorpID: "corp_local_fixture",
+		UserID: "identity_local_fixture",
+	}}}
+	if got, want := canonicalStoredSelector(localCfg, "local-profile-fixture"), "corp_local_fixture:identity_local_fixture"; got != want {
+		t.Fatalf("canonical local selector = %q, want %q", got, want)
+	}
+	if unresolvedProfileForCorp(nil, "corp") != nil || unresolvedProfileForLocalName(nil, "local") != nil {
+		t.Fatal("nil profile registry returned an unresolved profile")
+	}
+	if unresolvedProfileForLocalName(localCfg, "   ") != nil {
+		t.Fatal("blank local name returned an unresolved profile")
+	}
+	duplicateCfg := &ProfilesConfig{Profiles: []Profile{
+		{Name: "duplicate-blank", CorpID: "corp_duplicate_one"},
+		{Name: "Exact One", CorpID: "corp_duplicate_one", UserID: "identity_one"},
+		{Name: "duplicate-blank", CorpID: "corp_duplicate_two"},
+		{Name: "Exact Two", CorpID: "corp_duplicate_two", UserID: "identity_two"},
+	}}
+	if unresolvedProfileForLocalName(duplicateCfg, "duplicate-blank") != nil {
+		t.Fatal("duplicate unresolved local name selected an arbitrary profile")
+	}
+
+	oldLoadCorp := tokenLoadKeychainForCorpID
+	t.Cleanup(func() { tokenLoadKeychainForCorpID = oldLoadCorp })
+	tokenLoadKeychainForCorpID = func(string) (*TokenData, error) { return nil, nil }
+	if _, err := tokenLoadProfileIdentity(Profile{CorpID: "corp_nil_token_fixture"}); !errors.Is(err, ErrTokenDataNotFound) {
+		t.Fatalf("nil organization token error = %v", err)
+	}
+}
+
+func TestCrossPlatformCoverageLegacyProfileLifecycleErrorEdges(t *testing.T) {
+	oldEnsure := profilesEnsureMigration
+	oldLoad := profilesLoad
+	oldSave := profilesSave
+	oldLoadCorp := profilesLoadCorp
+	oldLoadLegacy := profilesLoadLegacy
+	oldLoadIdentity := profilesLoadIdentity
+	oldSaveCorp := profilesSaveCorp
+	oldDeleteCorp := profilesDeleteCorp
+	oldDeleteLegacy := profilesDeleteLegacy
+	oldDeleteMarker := profilesDeleteMarker
+	t.Cleanup(func() {
+		profilesEnsureMigration = oldEnsure
+		profilesLoad = oldLoad
+		profilesSave = oldSave
+		profilesLoadCorp = oldLoadCorp
+		profilesLoadLegacy = oldLoadLegacy
+		profilesLoadIdentity = oldLoadIdentity
+		profilesSaveCorp = oldSaveCorp
+		profilesDeleteCorp = oldDeleteCorp
+		profilesDeleteLegacy = oldDeleteLegacy
+		profilesDeleteMarker = oldDeleteMarker
+	})
+
+	identityFailure := errors.New("identity load failure")
+	profilesEnsureMigration = func(string) error { return nil }
+	profilesSave = func(string, *ProfilesConfig) error { return nil }
+	profilesLoadCorp = func(string) (*TokenData, error) { return nil, ErrTokenDataNotFound }
+	profilesLoadLegacy = func() (*TokenData, error) { return nil, ErrTokenDataNotFound }
+	profilesLoadIdentity = func(string, string) (*TokenData, error) { return nil, identityFailure }
+	profilesSaveCorp = func(string, *TokenData) error { return nil }
+	profilesDeleteCorp = func(string) error { return nil }
+	profilesDeleteLegacy = func() error { return nil }
+	profilesDeleteMarker = func(string) error { return nil }
+
+	setCurrentCfg := &ProfilesConfig{
+		Version:        profilesVersion,
+		CurrentProfile: "corp_set_fixture:identity_set_fixture",
+		OrgCurrentProfiles: map[string]string{
+			"corp_set_fixture": "corp_set_fixture:identity_set_fixture",
+		},
+		Profiles: []Profile{{
+			Name:   "Set Account",
+			CorpID: "corp_set_fixture",
+			UserID: "identity_set_fixture",
+		}},
+	}
+	profilesLoad = func(string) (*ProfilesConfig, error) { return setCurrentCfg, nil }
+	if _, err := setCurrentProfileLocked(t.TempDir(), "corp_set_fixture:identity_set_fixture"); !errors.Is(err, identityFailure) {
+		t.Fatalf("setCurrentProfileLocked sync error = %v", err)
+	}
+
+	usePreviousCfg := &ProfilesConfig{
+		Version:         profilesVersion,
+		CurrentProfile:  "corp_previous_fixture:identity_current_fixture",
+		PreviousProfile: "corp_previous_fixture:identity_previous_fixture",
+		OrgCurrentProfiles: map[string]string{
+			"corp_previous_fixture": "corp_previous_fixture:identity_current_fixture",
+		},
+		Profiles: []Profile{
+			{Name: "Current", CorpID: "corp_previous_fixture", UserID: "identity_current_fixture"},
+			{Name: "Previous", CorpID: "corp_previous_fixture", UserID: "identity_previous_fixture"},
+		},
+	}
+	profilesLoad = func(string) (*ProfilesConfig, error) { return usePreviousCfg, nil }
+	if _, err := usePreviousProfileLocked(t.TempDir()); !errors.Is(err, identityFailure) {
+		t.Fatalf("usePreviousProfileLocked sync error = %v", err)
+	}
+
+	snapshotFailure := errors.New("organization snapshot failure")
+	profilesLoadCorp = func(string) (*TokenData, error) { return nil, snapshotFailure }
+	if _, err := snapshotProfileSelectionMirrors(t.TempDir(), "corp_snapshot_fixture", true); !errors.Is(err, snapshotFailure) {
+		t.Fatalf("snapshot organization error = %v", err)
+	}
+	profilesLoadCorp = func(string) (*TokenData, error) { return nil, ErrTokenDataNotFound }
+
+	operationFailure := errors.New("selection operation failure")
+	organizationRestoreFailure := errors.New("organization restore failure")
+	profilesSaveCorp = func(string, *TokenData) error { return organizationRestoreFailure }
+	withOrganization := profileSelectionMirrorSnapshot{
+		organization: tokenSlotSnapshot{known: true, exists: true, token: &TokenData{CorpID: "corp_rollback_fixture"}},
+		marker:       tokenMarkerSnapshot{known: true},
+	}
+	if err := rollbackProfileSelection(t.TempDir(), &ProfilesConfig{}, "corp_rollback_fixture", withOrganization, operationFailure); !errors.Is(err, operationFailure) || !errors.Is(err, organizationRestoreFailure) {
+		t.Fatalf("rollback organization save error = %v", err)
+	}
+
+	organizationDeleteFailure := errors.New("organization delete failure")
+	profilesSaveCorp = func(string, *TokenData) error { return nil }
+	profilesDeleteCorp = func(string) error { return organizationDeleteFailure }
+	withoutOrganization := profileSelectionMirrorSnapshot{
+		organization: tokenSlotSnapshot{known: true},
+		marker:       tokenMarkerSnapshot{known: true},
+	}
+	if err := rollbackProfileSelection(t.TempDir(), &ProfilesConfig{}, "corp_rollback_fixture", withoutOrganization, operationFailure); !errors.Is(err, operationFailure) || !errors.Is(err, organizationDeleteFailure) {
+		t.Fatalf("rollback organization delete error = %v", err)
+	}
+	profilesDeleteCorp = func(string) error { return nil }
+
+	remainingCfg := &ProfilesConfig{
+		Version:        profilesVersion,
+		CurrentProfile: "corp_removed_fixture:identity_removed_fixture",
+		OrgCurrentProfiles: map[string]string{
+			"corp_removed_fixture": "corp_removed_fixture:identity_removed_fixture",
+		},
+		Profiles: []Profile{
+			{Name: "Removed", CorpID: "corp_removed_fixture", UserID: "identity_removed_fixture"},
+			{Name: "Remaining", CorpID: "corp_remaining_fixture", UserID: "identity_remaining_fixture"},
+		},
+	}
+	profilesLoad = func(string) (*ProfilesConfig, error) { return remainingCfg, nil }
+	if _, err := removeProfileLocked(t.TempDir(), "corp_removed_fixture:identity_removed_fixture"); err != nil {
+		t.Fatalf("removeProfileLocked(single fallback) error = %v", err)
+	}
+	if remainingCfg.CurrentProfile != "corp_remaining_fixture:identity_remaining_fixture" {
+		t.Fatalf("fallback current profile = %q", remainingCfg.CurrentProfile)
+	}
+
+	blankCfg := &ProfilesConfig{
+		Version:         profilesVersion,
+		CurrentProfile:  "corp_blank_fixture:identity_exact_fixture",
+		PreviousProfile: "legacy-blank-fixture",
+		OrgCurrentProfiles: map[string]string{
+			"corp_blank_fixture": "corp_blank_fixture:identity_exact_fixture",
+		},
+		Profiles: []Profile{
+			{Name: "legacy-blank-fixture", CorpID: "corp_blank_fixture"},
+			{Name: "Exact", CorpID: "corp_blank_fixture", UserID: "identity_exact_fixture"},
+		},
+	}
+	profilesLoad = func(string) (*ProfilesConfig, error) { return blankCfg, nil }
+	if _, err := removeProfileLocked(t.TempDir(), "corp_blank_fixture:identity_exact_fixture"); err != nil {
+		t.Fatalf("removeProfileLocked(blank fallback) error = %v", err)
+	}
+	if blankCfg.CurrentProfile != "corp_blank_fixture" || blankCfg.OrgCurrentProfiles["corp_blank_fixture"] != "" {
+		t.Fatalf("blank fallback selection = current %q org %q", blankCfg.CurrentProfile, blankCfg.OrgCurrentProfiles["corp_blank_fixture"])
+	}
+}

internal/auth/legacy_version_migration_test.go

@@ -0,0 +1,879 @@
+// Copyright 2026 Alibaba Group
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package auth
+
+import (
+	"encoding/json"
+	"errors"
+	"os"
+	"strings"
+	"testing"
+
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/keychain"
+)
+
+// The fixture builders spell out the exact v1 profiles and token JSON keys.
+// They deliberately avoid the current TokenData and ProfilesConfig serializers
+// so that historical field omission and account names stay part of the upgrade
+// contract exercised by these tests.
+func TestCrossPlatformCoverageV1044GlobalSlotWithoutProfilesMigrates(t *testing.T) {
+	for _, tc := range []struct {
+		name   string
+		userID string
+	}{
+		{name: "known user", userID: "legacy-user-v1044"},
+		{name: "unresolved external worker"},
+	} {
+		t.Run(tc.name, func(t *testing.T) {
+			cleanupHistoricalKeychain(t)
+			configDir := t.TempDir()
+			corpID := "ding_v1044_" + strings.ReplaceAll(tc.name, " ", "_")
+			seedHistoricalTokenSlot(t, keychain.AccountToken, historicalTokenJSON(
+				t, "global-v1044-"+tc.name, corpID, "V1044 Org", tc.userID, "",
+			))
+
+			loaded, err := LoadTokenData(configDir)
+			if err != nil {
+				t.Fatalf("LoadTokenData() error = %v", err)
+			}
+			if loaded.CorpID != corpID || loaded.UserID != tc.userID {
+				t.Fatalf("migrated v1.0.44 token = %#v", loaded)
+			}
+			if !TokenDataExistsKeychainForCorpID(corpID) {
+				t.Fatal("v1.0.44 global token was not copied to its organization slot")
+			}
+			if tc.userID != "" && !TokenDataExistsKeychainForIdentity(corpID, tc.userID) {
+				t.Fatal("v1.0.44 known identity slot was not created")
+			}
+		})
+	}
+}
+
+func TestCrossPlatformCoverageV1050AndV1051GlobalSlotWithV1ProfilesMigratesIdentity(t *testing.T) {
+	tests := []struct {
+		name        string
+		corpID      string
+		corpName    string
+		userID      string
+		userName    string
+		tokenHasUID bool
+	}{
+		{
+			name:        "v1.0.50 token and profile both carry userId",
+			corpID:      "ding_v1050",
+			corpName:    "V1050 Org",
+			userID:      "legacy-user-v1050",
+			userName:    "V1050 User",
+			tokenHasUID: true,
+		},
+		{
+			name:        "v1.0.51 profile supplies omitted token userId",
+			corpID:      "ding_v1051",
+			corpName:    "V1051 Org",
+			userID:      "legacy-user-v1051",
+			userName:    "V1051 User",
+			tokenHasUID: false,
+		},
+	}
+
+	for _, tc := range tests {
+		t.Run(tc.name, func(t *testing.T) {
+			cleanupHistoricalKeychain(t)
+			configDir := t.TempDir()
+			writeHistoricalV1Profiles(t, configDir, []historicalV1Profile{{
+				name:     tc.corpName,
+				corpID:   tc.corpID,
+				corpName: tc.corpName,
+				userID:   tc.userID,
+				userName: tc.userName,
+				clientID: "ding-client-" + tc.corpID,
+			}}, tc.corpID, "", tc.corpID)
+
+			tokenUserID, tokenUserName := "", ""
+			if tc.tokenHasUID {
+				tokenUserID, tokenUserName = tc.userID, tc.userName
+			}
+			seedHistoricalTokenSlot(t, keychain.AccountToken, historicalTokenJSON(
+				t,
+				"global-"+tc.corpID,
+				tc.corpID,
+				tc.corpName,
+				tokenUserID,
+				tokenUserName,
+			))
+
+			// An ordinary first read is the upgrade trigger. A recoverable global
+			// token must populate both the organization and exact-identity slots
+			// even when a version-1 profiles registry already exists.
+			if _, err := LoadTokenData(configDir); err != nil {
+				t.Fatalf("LoadTokenData() error = %v", err)
+			}
+			migrated, err := LoadTokenDataKeychainForIdentity(tc.corpID, tc.userID)
+			if err != nil {
+				t.Fatalf("LoadTokenDataKeychainForIdentity(%q, %q) error = %v", tc.corpID, tc.userID, err)
+			}
+			if migrated.CorpID != tc.corpID || migrated.UserID != tc.userID {
+				t.Fatalf("migrated identity token = %#v", migrated)
+			}
+			if migrated.AccessToken != "global-"+tc.corpID ||
+				migrated.RefreshToken != "refresh-global-"+tc.corpID ||
+				migrated.PersistentCode != "persistent-global-"+tc.corpID ||
+				migrated.ClientID != "ding-client-historical" ||
+				migrated.Source != "mcp" {
+				t.Fatalf("migrated token fields were not preserved: %#v", migrated)
+			}
+			orgMirror, err := LoadTokenDataKeychainForCorpID(tc.corpID)
+			if err != nil {
+				t.Fatalf("LoadTokenDataKeychainForCorpID(%q) error = %v", tc.corpID, err)
+			}
+			if orgMirror.AccessToken != "global-"+tc.corpID {
+				t.Fatalf("organization token slot %q = %#v", TokenAccountForCorpID(tc.corpID), orgMirror)
+			}
+			if !tc.tokenHasUID && orgMirror.UserID != "" {
+				t.Fatalf("organization mirror inferred userId %q; want untouched historical blob", orgMirror.UserID)
+			}
+		})
+	}
+
+}
+
+func TestCrossPlatformCoverageV1052RawMultiOrganizationSlotsMigrateEveryIdentity(t *testing.T) {
+	cleanupHistoricalKeychain(t)
+	configDir := t.TempDir()
+	organizations := seedHistoricalV1052MultiOrganizationState(t, configDir)
+
+	// Reading only the current organization must upgrade the complete registry,
+	// including inactive organizations that are not selected.
+	loaded, err := LoadTokenData(configDir)
+	if err != nil {
+		t.Fatalf("LoadTokenData() error = %v", err)
+	}
+	if loaded.CorpID != organizations[1].corpID || loaded.UserID != organizations[1].userID {
+		t.Fatalf("current token after migration = %#v", loaded)
+	}
+	assertHistoricalV1052IdentitySlots(t, organizations, nil)
+}
+
+func TestCrossPlatformCoverageV1052UnresolvedMultiOrganizationProfilesRemainUsable(t *testing.T) {
+	cleanupHistoricalKeychain(t)
+	configDir := t.TempDir()
+	organizations := []historicalV1052Organization{
+		{corpID: "ding_v1052_external_a", corpName: "External Org A", accessToken: "external-access-a"},
+		{corpID: "ding_v1052_external_b", corpName: "External Org B", accessToken: "external-access-b"},
+		{corpID: "ding_v1052_external_c", corpName: "External Org C", accessToken: "external-access-c"},
+	}
+	profiles := make([]historicalV1Profile, 0, len(organizations))
+	for _, organization := range organizations {
+		profiles = append(profiles, historicalV1Profile{
+			name: organization.corpName, corpID: organization.corpID, corpName: organization.corpName,
+		})
+		seedHistoricalTokenSlot(t, TokenAccountForCorpID(organization.corpID), historicalTokenJSON(
+			t, organization.accessToken, organization.corpID, organization.corpName, "", "",
+		))
+	}
+	writeHistoricalV1Profiles(
+		t,
+		configDir,
+		profiles,
+		organizations[0].corpID,
+		organizations[0].corpID,
+		organizations[1].corpID,
+	)
+
+	if err := EnsureProfilesMigration(configDir); err != nil {
+		t.Fatalf("EnsureProfilesMigration() error = %v", err)
+	}
+	for _, organization := range organizations {
+		loaded, err := LoadTokenDataForProfile(configDir, organization.corpID)
+		if err != nil {
+			t.Fatalf("LoadTokenDataForProfile(%q) error = %v", organization.corpID, err)
+		}
+		if loaded.AccessToken != organization.accessToken || loaded.CorpID != organization.corpID || loaded.UserID != "" {
+			t.Fatalf("unresolved organization token for %q = %#v", organization.corpID, loaded)
+		}
+	}
+}
+
+func TestCrossPlatformCoverageV1052FirstSaveMigratesAllOrganizationsBeforeV2Commit(t *testing.T) {
+	cleanupHistoricalKeychain(t)
+	configDir := t.TempDir()
+	organizations := seedHistoricalV1052MultiOrganizationState(t, configDir)
+
+	// A login or refresh can make SaveTokenData the first new-version action.
+	// It must migrate every old organization before profiles.json becomes v2,
+	// otherwise the remaining organization mirrors are stranded permanently.
+	firstWrite := &TokenData{
+		AccessToken:    "new-first-action-access",
+		RefreshToken:   "new-first-action-refresh",
+		PersistentCode: "new-first-action-persistent",
+		CorpID:         organizations[1].corpID,
+		CorpName:       organizations[1].corpName,
+		UserID:         organizations[1].userID,
+		UserName:       organizations[1].userName,
+		ClientID:       "ding-client-new-first-action",
+		Source:         "mcp",
+	}
+	if err := SaveTokenData(configDir, firstWrite); err != nil {
+		t.Fatalf("SaveTokenData(first new-version action) error = %v", err)
+	}
+
+	assertHistoricalV1052IdentitySlots(t, organizations, map[string]string{
+		organizations[1].corpID: firstWrite.AccessToken,
+	})
+	migratedCurrent, err := LoadTokenDataKeychainForIdentity(firstWrite.CorpID, firstWrite.UserID)
+	if err != nil {
+		t.Fatalf("LoadTokenDataKeychainForIdentity(first write) error = %v", err)
+	}
+	if migratedCurrent.RefreshToken != firstWrite.RefreshToken ||
+		migratedCurrent.PersistentCode != firstWrite.PersistentCode ||
+		migratedCurrent.ClientID != firstWrite.ClientID ||
+		migratedCurrent.Source != firstWrite.Source {
+		t.Fatalf("first-write identity token fields were not preserved: %#v", migratedCurrent)
+	}
+	cfg, err := LoadProfiles(configDir)
+	if err != nil {
+		t.Fatalf("LoadProfiles() error = %v", err)
+	}
+	if cfg.Version != profilesVersion || len(cfg.Profiles) != len(organizations) {
+		t.Fatalf("profiles after first save = %#v", cfg)
+	}
+}
+
+func TestCrossPlatformCoverageLegacyGlobalFallbackIsStrictlyScoped(t *testing.T) {
+	t.Run("different organization is never reused", func(t *testing.T) {
+		cleanupHistoricalKeychain(t)
+		configDir := t.TempDir()
+		writeHistoricalV1Profiles(t, configDir, []historicalV1Profile{{
+			name:     "Expected Org",
+			corpID:   "ding_expected",
+			corpName: "Expected Org",
+			userID:   "expected-user",
+			userName: "Expected User",
+			clientID: "ding-client-expected",
+		}}, "ding_expected", "", "ding_expected")
+		seedHistoricalTokenSlot(t, keychain.AccountToken, historicalTokenJSON(
+			t,
+			"wrong-org-access",
+			"ding_other",
+			"Other Org",
+			"other-user",
+			"Other User",
+		))
+
+		if err := EnsureProfilesMigration(configDir); err != nil {
+			t.Fatalf("EnsureProfilesMigration() error = %v", err)
+		}
+		if TokenDataExistsKeychainForCorpID("ding_expected") ||
+			TokenDataExistsKeychainForIdentity("ding_expected", "expected-user") {
+			t.Fatal("global token from another organization was reused")
+		}
+	})
+
+	t.Run("version 2 empty tombstone never imports global slot", func(t *testing.T) {
+		cleanupHistoricalKeychain(t)
+		configDir := t.TempDir()
+		if err := SaveProfiles(configDir, &ProfilesConfig{Version: profilesVersion}); err != nil {
+			t.Fatalf("SaveProfiles() error = %v", err)
+		}
+		seedHistoricalTokenSlot(t, keychain.AccountToken, historicalTokenJSON(
+			t,
+			"stale-v2-global",
+			"ding_v2",
+			"V2 Org",
+			"v2-user",
+			"V2 User",
+		))
+
+		if err := EnsureProfilesMigration(configDir); err != nil {
+			t.Fatalf("EnsureProfilesMigration() error = %v", err)
+		}
+		if TokenDataExistsKeychainForCorpID("ding_v2") ||
+			TokenDataExistsKeychainForIdentity("ding_v2", "v2-user") {
+			t.Fatal("version 2 logout tombstone imported the stale global slot")
+		}
+	})
+
+	t.Run("multiple same organization accounts never receive guessed identity", func(t *testing.T) {
+		cleanupHistoricalKeychain(t)
+		configDir := t.TempDir()
+		writeHistoricalV1Profiles(t, configDir, []historicalV1Profile{
+			{
+				name:     "Shared Org One",
+				corpID:   "ding_shared",
+				corpName: "Shared Org",
+				userID:   "shared-user-one",
+				userName: "Shared User One",
+				clientID: "ding-client-shared",
+			},
+			{
+				name:     "Shared Org Two",
+				corpID:   "ding_shared",
+				corpName: "Shared Org",
+				userID:   "shared-user-two",
+				userName: "Shared User Two",
+				clientID: "ding-client-shared",
+			},
+		}, "ding_shared", "", "ding_shared")
+		seedHistoricalTokenSlot(t, keychain.AccountToken, historicalTokenJSON(
+			t,
+			"shared-without-user",
+			"ding_shared",
+			"Shared Org",
+			"",
+			"",
+		))
+
+		if err := EnsureProfilesMigration(configDir); err != nil {
+			t.Fatalf("EnsureProfilesMigration() error = %v", err)
+		}
+		if !TokenDataExistsKeychainForCorpID("ding_shared") {
+			t.Fatal("matching organization mirror was not restored")
+		}
+		for _, userID := range []string{"shared-user-one", "shared-user-two"} {
+			if TokenDataExistsKeychainForIdentity("ding_shared", userID) {
+				t.Fatalf("ambiguous global token was copied to identity %q", userID)
+			}
+		}
+	})
+}
+
+func TestCrossPlatformCoverageV1053PartialV2RegistryRepairsFromMatchingGlobalSlot(t *testing.T) {
+	tests := []struct {
+		name          string
+		profileUserID string
+		tokenUserID   string
+	}{
+		{
+			name:          "matching token identity",
+			profileUserID: "user_v1053_matching",
+			tokenUserID:   "user_v1053_matching",
+		},
+		{name: "token omitted identity", profileUserID: "user_v1053_token_omitted"},
+		{name: "sole unresolved profile"},
+	}
+
+	for _, tc := range tests {
+		t.Run(tc.name, func(t *testing.T) {
+			cleanupHistoricalKeychain(t)
+			configDir := t.TempDir()
+			suffix := strings.ReplaceAll(tc.name, " ", "_")
+			corpID := "ding_v1053_partial_" + suffix
+			profile := Profile{
+				Name:     "V1053 Partial Org",
+				CorpID:   corpID,
+				CorpName: "V1053 Partial Org",
+				UserID:   tc.profileUserID,
+				UserName: "V1053 Partial User",
+			}
+			identityLoads := 0
+			if profile.UserID == "" {
+				originalLoadIdentity := profilesLoadIdentity
+				profilesLoadIdentity = func(corpID, userID string) (*TokenData, error) {
+					identityLoads++
+					return originalLoadIdentity(corpID, userID)
+				}
+				t.Cleanup(func() { profilesLoadIdentity = originalLoadIdentity })
+			}
+			selector := ProfileSelector(profile)
+			if err := SaveProfiles(configDir, &ProfilesConfig{
+				Version:            profilesVersion,
+				CurrentProfile:     selector,
+				OrgCurrentProfiles: map[string]string{corpID: selector},
+				Profiles:           []Profile{profile},
+			}); err != nil {
+				t.Fatalf("SaveProfiles() error = %v", err)
+			}
+			seedHistoricalTokenSlot(t, keychain.AccountToken, historicalTokenJSON(
+				t,
+				"v1053-global-"+suffix,
+				corpID,
+				profile.CorpName,
+				tc.tokenUserID,
+				"",
+			))
+
+			if TokenDataExistsKeychainForCorpID(corpID) {
+				t.Fatal("partial v2 fixture unexpectedly contained an organization or identity slot")
+			}
+			if profile.UserID != "" && TokenDataExistsKeychainForIdentity(corpID, profile.UserID) {
+				t.Fatal("partial v2 fixture unexpectedly contained an identity slot")
+			}
+			loaded, err := LoadTokenDataForProfile(configDir, selector)
+			if err != nil {
+				t.Fatalf("LoadTokenDataForProfile() error = %v", err)
+			}
+			if loaded.AccessToken != "v1053-global-"+suffix ||
+				loaded.CorpID != corpID || loaded.UserID != profile.UserID {
+				t.Fatalf("repaired v2 token = %#v", loaded)
+			}
+
+			if profile.UserID != "" {
+				identityToken, identityErr := LoadTokenDataKeychainForIdentity(corpID, profile.UserID)
+				if identityErr != nil {
+					t.Fatalf("LoadTokenDataKeychainForIdentity() error = %v", identityErr)
+				}
+				if identityToken.AccessToken != loaded.AccessToken || identityToken.UserID != profile.UserID {
+					t.Fatalf("repaired identity token = %#v", identityToken)
+				}
+			}
+			orgMirror, err := LoadTokenDataKeychainForCorpID(corpID)
+			if err != nil {
+				t.Fatalf("LoadTokenDataKeychainForCorpID() error = %v", err)
+			}
+			if orgMirror.AccessToken != loaded.AccessToken || orgMirror.UserID != tc.tokenUserID {
+				t.Fatalf("repaired organization mirror = %#v", orgMirror)
+			}
+			if identityLoads != 0 {
+				t.Fatalf("unresolved profile caused %d identity-slot reads; want 0", identityLoads)
+			}
+		})
+	}
+
+	t.Run("matching organization among multiple organizations", func(t *testing.T) {
+		cleanupHistoricalKeychain(t)
+		configDir := t.TempDir()
+		profiles := []Profile{
+			{Name: "Partial Org A", CorpID: "ding_v1053_partial_a", UserID: "user_v1053_partial_a"},
+			{Name: "Partial Org B", CorpID: "ding_v1053_partial_b", UserID: "user_v1053_partial_b"},
+		}
+		if err := SaveProfiles(configDir, &ProfilesConfig{
+			Version:        profilesVersion,
+			CurrentProfile: ProfileSelector(profiles[1]),
+			Profiles:       profiles,
+		}); err != nil {
+			t.Fatalf("SaveProfiles() error = %v", err)
+		}
+		seedHistoricalTokenSlot(t, keychain.AccountToken, historicalTokenJSON(
+			t,
+			"v1053-global-multi-org",
+			profiles[1].CorpID,
+			profiles[1].Name,
+			profiles[1].UserID,
+			"",
+		))
+
+		if err := EnsureProfilesMigration(configDir); err != nil {
+			t.Fatalf("EnsureProfilesMigration() error = %v", err)
+		}
+		if TokenDataExistsKeychainForCorpID(profiles[0].CorpID) ||
+			TokenDataExistsKeychainForIdentity(profiles[0].CorpID, profiles[0].UserID) {
+			t.Fatal("global token was copied into the non-matching organization")
+		}
+		repaired, err := LoadTokenDataKeychainForIdentity(profiles[1].CorpID, profiles[1].UserID)
+		if err != nil {
+			t.Fatalf("LoadTokenDataKeychainForIdentity(matching organization) error = %v", err)
+		}
+		if repaired.AccessToken != "v1053-global-multi-org" || repaired.UserID != profiles[1].UserID {
+			t.Fatalf("multi-organization v2 repair token = %#v", repaired)
+		}
+	})
+}
+
+func TestCrossPlatformCoverageV1053PartialV2RegistryRejectsUnsafeGlobalSlot(t *testing.T) {
+	t.Run("global token organization differs", func(t *testing.T) {
+		cleanupHistoricalKeychain(t)
+		configDir := t.TempDir()
+		profile := Profile{Name: "Expected Org", CorpID: "ding_v2_expected", UserID: "user_v2_expected"}
+		if err := SaveProfiles(configDir, &ProfilesConfig{
+			Version:        profilesVersion,
+			CurrentProfile: ProfileSelector(profile),
+			Profiles:       []Profile{profile},
+		}); err != nil {
+			t.Fatalf("SaveProfiles() error = %v", err)
+		}
+		seedHistoricalTokenSlot(t, keychain.AccountToken, historicalTokenJSON(
+			t, "cross-corp-global", "ding_v2_other", "Other Org", profile.UserID, "",
+		))
+
+		if err := EnsureProfilesMigration(configDir); err != nil {
+			t.Fatalf("EnsureProfilesMigration() error = %v", err)
+		}
+		if TokenDataExistsKeychainForCorpID(profile.CorpID) ||
+			TokenDataExistsKeychainForIdentity(profile.CorpID, profile.UserID) {
+			t.Fatal("cross-organization global token was imported")
+		}
+	})
+
+	t.Run("unresolved profile rejects global token identity", func(t *testing.T) {
+		cleanupHistoricalKeychain(t)
+		configDir := t.TempDir()
+		profile := Profile{Name: "Unresolved External", CorpID: "ding_v2_unresolved"}
+		if err := SaveProfiles(configDir, &ProfilesConfig{
+			Version:        profilesVersion,
+			CurrentProfile: ProfileSelector(profile),
+			Profiles:       []Profile{profile},
+		}); err != nil {
+			t.Fatalf("SaveProfiles() error = %v", err)
+		}
+		seedHistoricalTokenSlot(t, keychain.AccountToken, historicalTokenJSON(
+			t, "unexpected-identity-global", profile.CorpID, profile.Name, "user_v2_unexpected", "",
+		))
+
+		if err := EnsureProfilesMigration(configDir); err != nil {
+			t.Fatalf("EnsureProfilesMigration() error = %v", err)
+		}
+		if TokenDataExistsKeychainForCorpID(profile.CorpID) {
+			t.Fatal("global token userId was attached to an unresolved profile")
+		}
+	})
+
+	t.Run("global token identity differs", func(t *testing.T) {
+		cleanupHistoricalKeychain(t)
+		configDir := t.TempDir()
+		profile := Profile{Name: "Expected User", CorpID: "ding_v2_uid", UserID: "user_v2_expected"}
+		if err := SaveProfiles(configDir, &ProfilesConfig{
+			Version:        profilesVersion,
+			CurrentProfile: ProfileSelector(profile),
+			Profiles:       []Profile{profile},
+		}); err != nil {
+			t.Fatalf("SaveProfiles() error = %v", err)
+		}
+		seedHistoricalTokenSlot(t, keychain.AccountToken, historicalTokenJSON(
+			t, "wrong-user-global", profile.CorpID, profile.Name, "user_v2_other", "",
+		))
+
+		if err := EnsureProfilesMigration(configDir); err != nil {
+			t.Fatalf("EnsureProfilesMigration() error = %v", err)
+		}
+		if TokenDataExistsKeychainForCorpID(profile.CorpID) ||
+			TokenDataExistsKeychainForIdentity(profile.CorpID, profile.UserID) {
+			t.Fatal("global token with a different userId was imported")
+		}
+	})
+
+	t.Run("multiple accounts in one organization stay unresolved", func(t *testing.T) {
+		cleanupHistoricalKeychain(t)
+		configDir := t.TempDir()
+		corpID := "ding_v2_shared"
+		profiles := []Profile{
+			{Name: "Shared User One", CorpID: corpID, UserID: "user_v2_shared_one"},
+			{Name: "Shared User Two", CorpID: corpID, UserID: "user_v2_shared_two"},
+		}
+		if err := SaveProfiles(configDir, &ProfilesConfig{
+			Version:        profilesVersion,
+			CurrentProfile: ProfileSelector(profiles[0]),
+			Profiles:       profiles,
+		}); err != nil {
+			t.Fatalf("SaveProfiles() error = %v", err)
+		}
+		seedHistoricalTokenSlot(t, keychain.AccountToken, historicalTokenJSON(
+			t, "ambiguous-global", corpID, "Shared Org", profiles[0].UserID, "",
+		))
+
+		if err := EnsureProfilesMigration(configDir); err != nil {
+			t.Fatalf("EnsureProfilesMigration() error = %v", err)
+		}
+		if TokenDataExistsKeychainForCorpID(corpID) {
+			t.Fatal("multi-account organization imported the mutable global mirror")
+		}
+		for _, profile := range profiles {
+			if TokenDataExistsKeychainForIdentity(corpID, profile.UserID) {
+				t.Fatalf("multi-account global token was copied to identity %q", profile.UserID)
+			}
+		}
+	})
+
+	t.Run("existing exact identity is not overwritten", func(t *testing.T) {
+		cleanupHistoricalKeychain(t)
+		configDir := t.TempDir()
+		profile := Profile{Name: "Existing User", CorpID: "ding_v2_existing", UserID: "user_v2_existing"}
+		if err := SaveProfiles(configDir, &ProfilesConfig{
+			Version:        profilesVersion,
+			CurrentProfile: ProfileSelector(profile),
+			Profiles:       []Profile{profile},
+		}); err != nil {
+			t.Fatalf("SaveProfiles() error = %v", err)
+		}
+		seedHistoricalTokenSlot(t, TokenAccountForIdentity(profile.CorpID, profile.UserID), historicalTokenJSON(
+			t, "existing-exact", profile.CorpID, profile.Name, profile.UserID, "",
+		))
+		seedHistoricalTokenSlot(t, keychain.AccountToken, historicalTokenJSON(
+			t, "stale-global", profile.CorpID, profile.Name, profile.UserID, "",
+		))
+
+		if err := EnsureProfilesMigration(configDir); err != nil {
+			t.Fatalf("EnsureProfilesMigration() error = %v", err)
+		}
+		if TokenDataExistsKeychainForCorpID(profile.CorpID) {
+			t.Fatal("global mirror recreated an organization slot beside an existing exact identity")
+		}
+		exact, err := LoadTokenDataKeychainForIdentity(profile.CorpID, profile.UserID)
+		if err != nil {
+			t.Fatalf("LoadTokenDataKeychainForIdentity() error = %v", err)
+		}
+		if exact.AccessToken != "existing-exact" {
+			t.Fatalf("existing exact identity was overwritten: %#v", exact)
+		}
+	})
+}
+
+func TestCrossPlatformCoverageLegacyMigrationPersistenceErrors(t *testing.T) {
+	oldLoad := profilesLoad
+	oldSave := profilesSave
+	oldLoadLegacy := profilesLoadLegacy
+	oldSaveCorp := profilesSaveCorp
+	oldLoadCorp := profilesLoadCorp
+	oldLoadIdentity := profilesLoadIdentity
+	oldSaveIdentity := profilesSaveIdentity
+	t.Cleanup(func() {
+		profilesLoad = oldLoad
+		profilesSave = oldSave
+		profilesLoadLegacy = oldLoadLegacy
+		profilesSaveCorp = oldSaveCorp
+		profilesLoadCorp = oldLoadCorp
+		profilesLoadIdentity = oldLoadIdentity
+		profilesSaveIdentity = oldSaveIdentity
+	})
+
+	fail := errors.New("legacy migration persistence failed")
+	baseConfig := func(version int) *ProfilesConfig {
+		return &ProfilesConfig{
+			Version: version,
+			Profiles: []Profile{{
+				Name: "Legacy User", CorpID: "ding_legacy_error", UserID: "legacy-user",
+			}},
+		}
+	}
+	profilesSave = func(string, *ProfilesConfig) error { return nil }
+	profilesLoadCorp = func(string) (*TokenData, error) { return nil, ErrTokenDataNotFound }
+	profilesLoadIdentity = func(string, string) (*TokenData, error) { return nil, ErrTokenDataNotFound }
+	profilesSaveIdentity = func(string, string, *TokenData) error { return nil }
+
+	t.Run("global compatibility slot read", func(t *testing.T) {
+		profilesLoad = func(string) (*ProfilesConfig, error) { return baseConfig(1), nil }
+		profilesLoadLegacy = func() (*TokenData, error) { return nil, fail }
+		profilesSaveCorp = func(string, *TokenData) error { return nil }
+		if err := ensureProfilesMigrationLocked("cfg"); !errors.Is(err, fail) {
+			t.Fatalf("ensureProfilesMigrationLocked() error = %v, want %v", err, fail)
+		}
+	})
+
+	t.Run("restored organization slot write", func(t *testing.T) {
+		profilesLoad = func(string) (*ProfilesConfig, error) { return baseConfig(1), nil }
+		profilesLoadLegacy = func() (*TokenData, error) {
+			return &TokenData{CorpID: "ding_legacy_error", AccessToken: "legacy-access"}, nil
+		}
+		profilesSaveCorp = func(string, *TokenData) error { return fail }
+		if err := ensureProfilesMigrationLocked("cfg"); !errors.Is(err, fail) {
+			t.Fatalf("ensureProfilesMigrationLocked() error = %v, want %v", err, fail)
+		}
+	})
+
+	t.Run("repaired identity slot write", func(t *testing.T) {
+		profilesLoad = func(string) (*ProfilesConfig, error) { return baseConfig(profilesVersion), nil }
+		profilesLoadCorp = func(string) (*TokenData, error) {
+			return &TokenData{CorpID: "ding_legacy_error", AccessToken: "legacy-access"}, nil
+		}
+		profilesSaveIdentity = func(string, string, *TokenData) error { return fail }
+		if err := ensureProfilesMigrationLocked("cfg"); !errors.Is(err, fail) {
+			t.Fatalf("ensureProfilesMigrationLocked() error = %v, want %v", err, fail)
+		}
+	})
+
+	t.Run("partial v2 identity slot read", func(t *testing.T) {
+		profilesLoad = func(string) (*ProfilesConfig, error) { return baseConfig(profilesVersion), nil }
+		profilesLoadCorp = func(string) (*TokenData, error) { return nil, ErrTokenDataNotFound }
+		profilesLoadIdentity = func(string, string) (*TokenData, error) { return nil, fail }
+		if err := ensureProfilesMigrationLocked("cfg"); !errors.Is(err, fail) {
+			t.Fatalf("ensureProfilesMigrationLocked() error = %v, want %v", err, fail)
+		}
+	})
+}
+
+type historicalV1Profile struct {
+	name     string
+	corpID   string
+	corpName string
+	userID   string
+	userName string
+	clientID string
+}
+
+type historicalV1052Organization struct {
+	corpID      string
+	corpName    string
+	userID      string
+	userName    string
+	accessToken string
+}
+
+func seedHistoricalV1052MultiOrganizationState(t *testing.T, configDir string) []historicalV1052Organization {
+	t.Helper()
+	organizations := []historicalV1052Organization{
+		{corpID: "ding_v1052_a", corpName: "V1052 Org A", userID: "legacy-user-v1052-a", userName: "V1052 User A", accessToken: "v1052-access-a"},
+		{corpID: "ding_v1052_b", corpName: "V1052 Org B", userID: "legacy-user-v1052-b", userName: "V1052 User B", accessToken: "v1052-access-b"},
+		{corpID: "ding_v1052_c", corpName: "V1052 Org C", userID: "legacy-user-v1052-c", userName: "V1052 User C", accessToken: "v1052-access-c"},
+	}
+	profiles := make([]historicalV1Profile, 0, len(organizations))
+	for _, organization := range organizations {
+		profiles = append(profiles, historicalV1Profile{
+			name:     organization.corpName,
+			corpID:   organization.corpID,
+			corpName: organization.corpName,
+			userID:   organization.userID,
+			userName: organization.userName,
+			clientID: "ding-client-v1052",
+		})
+	}
+	writeHistoricalV1Profiles(
+		t,
+		configDir,
+		profiles,
+		organizations[0].corpID,
+		organizations[0].corpID,
+		organizations[1].corpID,
+	)
+
+	for _, organization := range organizations {
+		// v1.0.52 MCP responses commonly omitted userId while profiles.json
+		// retained a uniquely known identity.
+		seedHistoricalTokenSlot(t, TokenAccountForCorpID(organization.corpID), historicalTokenJSON(
+			t,
+			organization.accessToken,
+			organization.corpID,
+			organization.corpName,
+			"",
+			"",
+		))
+	}
+	// v1.0.52 also mirrored the selected organization into the global account.
+	current := organizations[1]
+	seedHistoricalTokenSlot(t, keychain.AccountToken, historicalTokenJSON(
+		t,
+		current.accessToken,
+		current.corpID,
+		current.corpName,
+		"",
+		"",
+	))
+	return organizations
+}
+
+func assertHistoricalV1052IdentitySlots(
+	t *testing.T,
+	organizations []historicalV1052Organization,
+	accessOverrides map[string]string,
+) {
+	t.Helper()
+	for _, organization := range organizations {
+		migrated, err := LoadTokenDataKeychainForIdentity(organization.corpID, organization.userID)
+		if err != nil {
+			t.Errorf("LoadTokenDataKeychainForIdentity(%q, %q) error = %v", organization.corpID, organization.userID, err)
+			continue
+		}
+		wantAccess := organization.accessToken
+		if override := accessOverrides[organization.corpID]; override != "" {
+			wantAccess = override
+		}
+		if migrated.AccessToken != wantAccess ||
+			migrated.CorpID != organization.corpID ||
+			migrated.UserID != organization.userID {
+			t.Errorf(
+				"migrated token for %q = %#v, want access=%q userId=%q",
+				organization.corpID,
+				migrated,
+				wantAccess,
+				organization.userID,
+			)
+		}
+		if accessOverrides[organization.corpID] == "" &&
+			(migrated.RefreshToken != "refresh-"+organization.accessToken ||
+				migrated.PersistentCode != "persistent-"+organization.accessToken ||
+				migrated.ClientID != "ding-client-historical" ||
+				migrated.Source != "mcp") {
+			t.Errorf("historical token fields for %q were not preserved: %#v", organization.corpID, migrated)
+		}
+	}
+}
+
+func historicalTokenJSON(t *testing.T, accessToken, corpID, corpName, userID, userName string) string {
+	t.Helper()
+	fixture := map[string]any{
+		"access_token":       accessToken,
+		"refresh_token":      "refresh-" + accessToken,
+		"persistent_code":    "persistent-" + accessToken,
+		"expires_at":         "2030-01-02T03:04:05Z",
+		"refresh_expires_at": "2030-02-02T03:04:05Z",
+		"corp_id":            corpID,
+		"corp_name":          corpName,
+		"client_id":          "ding-client-historical",
+		"source":             "mcp",
+	}
+	if userID != "" {
+		fixture["user_id"] = userID
+	}
+	if userName != "" {
+		fixture["user_name"] = userName
+	}
+	data, err := json.MarshalIndent(fixture, "", "  ")
+	if err != nil {
+		t.Fatalf("marshal historical token fixture: %v", err)
+	}
+	return string(data)
+}
+
+func writeHistoricalV1Profiles(
+	t *testing.T,
+	configDir string,
+	profiles []historicalV1Profile,
+	primaryProfile string,
+	previousProfile string,
+	currentProfile string,
+) {
+	t.Helper()
+	rawProfiles := make([]map[string]any, 0, len(profiles))
+	for _, profile := range profiles {
+		rawProfiles = append(rawProfiles, map[string]any{
+			"name":         profile.name,
+			"corpId":       profile.corpID,
+			"corpName":     profile.corpName,
+			"userId":       profile.userID,
+			"userName":     profile.userName,
+			"clientId":     profile.clientID,
+			"status":       "active",
+			"expiresAt":    "2030-01-02T03:04:05Z",
+			"refreshExpAt": "2030-02-02T03:04:05Z",
+		})
+	}
+	fixture := map[string]any{
+		"version":         1,
+		"primaryProfile":  primaryProfile,
+		"currentProfile":  currentProfile,
+		"previousProfile": previousProfile,
+		"profiles":        rawProfiles,
+	}
+	data, err := json.MarshalIndent(fixture, "", "  ")
+	if err != nil {
+		t.Fatalf("marshal historical profiles fixture: %v", err)
+	}
+	if err := os.MkdirAll(configDir, 0o700); err != nil {
+		t.Fatalf("create historical config directory: %v", err)
+	}
+	if err := os.WriteFile(ProfilesPath(configDir), append(data, '\n'), 0o600); err != nil {
+		t.Fatalf("write historical profiles.json: %v", err)
+	}
+}
+
+func seedHistoricalTokenSlot(t *testing.T, account, raw string) {
+	t.Helper()
+	if err := keychain.Set(keychain.Service, account, raw); err != nil {
+		t.Fatalf("seed historical keychain account %q: %v", account, err)
+	}
+}
+
+func cleanupHistoricalKeychain(t *testing.T) {
+	t.Helper()
+	t.Setenv(keychain.DisableKeychainEnv, "1")
+	cleanupKeychain(t)
+}

internal/auth/oauth_helpers.go

@@ -27,7 +27,6 @@ import (
 	"strings"
 	"time"
 
-	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/i18n"
 	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/pkg/config"
 )
 
@@ -39,10 +38,6 @@ var (
 )
 
 func (p *OAuthProvider) exchangeCode(ctx context.Context, code string) (*TokenData, error) {
-	if err := preflightTokenPersistence(p.configDir); err != nil {
-		return nil, fmt.Errorf("%s: %w", i18n.T("本地登录态无法安全更新"), err)
-	}
-
 	// Use MCP mode if clientID is from MCP server
 	if IsClientIDFromMCP() {
 		return p.exchangeCodeViaMCP(ctx, code)
@@ -79,13 +74,21 @@ func (p *OAuthProvider) exchangeCode(ctx context.Context, code string) (*TokenDa
 // the currently configured client credentials.  This is a convenience wrapper
 // around OAuthProvider.exchangeCode for callers outside the auth package.
 func ExchangeCodeForToken(ctx context.Context, configDir, code string) (*TokenData, error) {
+	if err := prepareLoginPersistence(configDir); err != nil {
+		return nil, fmt.Errorf("local login state cannot be safely updated before token exchange: %w", err)
+	}
 	p := &OAuthProvider{
 		configDir:  configDir,
 		clientID:   ClientID(),
 		Output:     io.Discard,
 		httpClient: oauthHTTPClient,
 	}
-	return p.exchangeCode(ctx, code)
+	data, err := p.exchangeCode(ctx, code)
+	if err != nil {
+		return nil, err
+	}
+	data.FreshAuthorization = true
+	return data, nil
 }
 
 // exchangeCodeViaMCP exchanges auth code for token via MCP proxy.
@@ -290,6 +293,24 @@ func (p *OAuthProvider) parseTokenResponse(body []byte) (*TokenData, error) {
 	return data, nil
 }
 
+const legacyMCPRefreshRejectedCode = "invalidParameter.authCode.notFound"
+
+// MCPTokenExchangeError preserves the backend business code so refresh callers
+// can distinguish a legacy credential that requires a new authorization from
+// transient transport failures.
+type MCPTokenExchangeError struct {
+	Code    string
+	Message string
+}
+
+func (e *MCPTokenExchangeError) Error() string {
+	return fmt.Sprintf("MCP token exchange failed: %s - %s", e.Code, e.Message)
+}
+
+func (e *MCPTokenExchangeError) requiresReauthorization() bool {
+	return strings.TrimSpace(e.Code) == legacyMCPRefreshRejectedCode
+}
+
 // parseMCPTokenResponse parses token response from MCP proxy.
 // MCP OAuth response format: {"accessToken": "...", "refreshToken": "...", "expiresIn": 7200, "corpId": "...", "corpName": "..."}
 func (p *OAuthProvider) parseMCPTokenResponse(body []byte) (*TokenData, error) {
@@ -313,7 +334,7 @@ func (p *OAuthProvider) parseMCPTokenResponse(body []byte) (*TokenData, error) {
 	}
 	// Check for error response
 	if resp.ErrorCode != "" || resp.ErrorMsg != "" {
-		return nil, fmt.Errorf("MCP token exchange failed: %s - %s", resp.ErrorCode, resp.ErrorMsg)
+		return nil, &MCPTokenExchangeError{Code: resp.ErrorCode, Message: resp.ErrorMsg}
 	}
 	if resp.AccessToken == "" {
 		return nil, fmt.Errorf("MCP token response missing accessToken (body: %s)", string(body))

internal/auth/oauth_provider.go

@@ -24,6 +24,7 @@ import (
 	"net"
 	"net/http"
 	"os"
+	"strconv"
 	"strings"
 	"sync"
 	"time"
@@ -117,10 +118,13 @@ func (p *OAuthProvider) Login(ctx context.Context, force bool) (*TokenData, erro
 	if !force {
 		data, err := oauthLoadToken(p.configDir)
 		if err != nil && !errors.Is(err, ErrTokenDataNotFound) && !os.IsNotExist(err) {
-			if preflightErr := preflightTokenPersistence(p.configDir); preflightErr != nil {
-				return nil, fmt.Errorf("%s: %w", i18n.T("本地登录态无法安全更新"), preflightErr)
+			// A damaged selected slot must not make browser reauthorization
+			// impossible. The target identity is unknown until token exchange, so
+			// continue into the full flow and let the target-only preflight reject
+			// an unsafe overwrite after identity enrichment.
+			if p.logger != nil {
+				p.logger.Warn(i18n.T("读取现有登录态失败，将尝试扫码登录"), "error", err)
 			}
-			return nil, fmt.Errorf("load existing access token: %w", err)
 		}
 		if err == nil {
 			// Case 1: access_token still valid — no action needed.
@@ -150,7 +154,7 @@ func (p *OAuthProvider) Login(ctx context.Context, force bool) (*TokenData, erro
 			}
 		}
 	}
-	if err := preflightTokenPersistence(p.configDir); err != nil {
+	if err := prepareLoginPersistence(p.configDir); err != nil {
 		return nil, fmt.Errorf("%s: %w", i18n.T("本地登录态无法安全更新"), err)
 	}
 
@@ -662,6 +666,7 @@ func (p *OAuthProvider) GetTokenSnapshot(ctx context.Context) (*TokenData, error
 		}
 		return nil, fmt.Errorf("load access token: %w", err)
 	}
+	profileSelector := StableTokenProfileSelector(p.configDir, data)
 
 	// Fast path: access_token still valid — no lock needed.
 	if data.IsAccessTokenValid() {
@@ -678,19 +683,40 @@ func (p *OAuthProvider) GetTokenSnapshot(ctx context.Context) (*TokenData, error
 		// refresh credential. Keep the profile active so a long-running source
 		// can retry after backoff. Terminal and unknown failures remain fatal.
 		if ClassifyRefreshFailure(rErr) != RefreshFailureTransient {
-			_ = oauthMarkProfile(p.configDir, TokenProfileSelector(data), ProfileStatusExpired)
+			_ = oauthMarkProfile(p.configDir, profileSelector, ProfileStatusExpired)
 		}
 		if p.logger != nil {
 			p.logger.Warn(i18n.T("refresh_token 刷新失败"), "error", rErr)
 		}
+		var exchangeErr *MCPTokenExchangeError
+		if errors.As(rErr, &exchangeErr) && exchangeErr.requiresReauthorization() {
+			return nil, fmt.Errorf(
+				"%s: %w",
+				legacyRefreshReauthorizationGuidance(profileSelector),
+				rErr,
+			)
+		}
 		return nil, fmt.Errorf("%s: %w", i18n.T("refresh_token 刷新失败"), rErr)
 	} else {
-		_ = oauthMarkProfile(p.configDir, TokenProfileSelector(data), ProfileStatusExpired)
+		_ = oauthMarkProfile(p.configDir, profileSelector, ProfileStatusExpired)
 	}
 
 	return nil, fmt.Errorf("%s: %w", i18n.T("所有凭证已失效，请运行 dws auth login 重新登录"), ErrTokenDataNotFound)
 }
 
+func legacyRefreshReauthorizationGuidance(profileSelector string) string {
+	guidance := "旧版登录态已无法由当前认证服务刷新；本地 profile 已保留，请重新运行 dws auth login 完成一次重新授权"
+	profileSelector = strings.TrimSpace(profileSelector)
+	if profileSelector == "" {
+		return guidance
+	}
+	return fmt.Sprintf(
+		"%s；为保留原身份，请把 --profile 参数设置为下方 profile 标识（标识仅作数据展示，不是可执行命令）:\nprofile: %s",
+		guidance,
+		strconv.Quote(profileSelector),
+	)
+}
+
 // GetAccessToken returns a valid access token, auto-refreshing if needed.
 // Uses a file lock with double-check pattern to prevent concurrent refresh
 // from multiple CLI processes.
@@ -763,6 +789,9 @@ func (p *OAuthProvider) lockedRefresh(ctx context.Context) (*TokenData, error) {
 // ExchangeAuthCode takes an AuthCode and an optional UserID provided by an
 // external host, exchanges it for tokens, and persists them.
 func (p *OAuthProvider) ExchangeAuthCode(ctx context.Context, authCode, uid string) (*TokenData, error) {
+	if err := prepareLoginPersistence(p.configDir); err != nil {
+		return nil, fmt.Errorf("%s: %w", i18n.T("本地登录态无法安全更新"), err)
+	}
 	tokenData, err := oauthExchange(p, ctx, authCode)
 	if err != nil {
 		return nil, fmt.Errorf("%s: %w", i18n.T("换取 token 失败"), err)
@@ -802,6 +831,9 @@ func (p *OAuthProvider) persistLoginToken(ctx context.Context, tokenData *TokenD
 		"user_id", strings.TrimSpace(tokenData.UserID),
 		"user_name", strings.TrimSpace(tokenData.UserName),
 	)
+	if err := preflightTokenWritePersistence(p.configDir, tokenData); err != nil {
+		return fmt.Errorf("%s: %w", i18n.T("本地登录态无法安全更新"), err)
+	}
 	if err := oauthSaveToken(p.configDir, tokenData); err != nil {
 		return err
 	}
@@ -812,14 +844,18 @@ func (p *OAuthProvider) prepareLoginToken(ctx context.Context, tokenData *TokenD
 	if tokenData == nil {
 		return fmt.Errorf("token data is empty")
 	}
+	tokenData.FreshAuthorization = true
 	if p != nil && p.IdentityEnricher != nil {
 		if err := p.IdentityEnricher(ctx, tokenData); err != nil {
 			return fmt.Errorf("resolve login identity: %w", err)
 		}
 	}
-	if strings.TrimSpace(tokenData.CorpID) != "" && strings.TrimSpace(tokenData.UserID) == "" {
-		return fmt.Errorf("resolve login identity: userId is required for corpId %q", tokenData.CorpID)
-	}
+	// v1.0.52 and earlier deliberately persisted the freshly exchanged token
+	// before best-effort contact enrichment. External-worker accounts can have a
+	// valid organization token while contact cannot return a userId, so rejecting
+	// that shape here makes an otherwise successful reauthorization impossible.
+	// SaveTokenData remains the safety boundary: an unresolved organization token
+	// cannot overwrite an organization that already has exact account identities.
 	return nil
 }
 
@@ -830,6 +866,9 @@ func (p *OAuthProvider) persistKnownLoginToken(tokenData *TokenData) error {
 	if strings.TrimSpace(tokenData.CorpID) != "" && strings.TrimSpace(tokenData.UserID) == "" {
 		return fmt.Errorf("resolve login identity: userId is required for corpId %q", tokenData.CorpID)
 	}
+	if err := preflightTokenWritePersistence(p.configDir, tokenData); err != nil {
+		return fmt.Errorf("%s: %w", i18n.T("本地登录态无法安全更新"), err)
+	}
 	return oauthSaveToken(p.configDir, tokenData)
 }
 

internal/auth/oauth_provider_coverage_test.go

@@ -303,6 +303,72 @@ func TestCrossPlatformCoverageOAuthLoginCallbackAndAPIs(t *testing.T) {
 	}
 }
 
+func TestOAuthForcedLoginIgnoresUnreadableUnrelatedProfile(t *testing.T) {
+	f := newOAuthLoginFixture(t, func(int32) CLIAuthStatus {
+		return CLIAuthStatus{Success: true, Result: &CLIAuthResult{CLIAuthEnabled: true}}
+	})
+	if err := SaveProfiles(f.configDir, &ProfilesConfig{
+		Version: profilesVersion,
+		Profiles: []Profile{{
+			Name:   "unrelated",
+			CorpID: "corp-unrelated",
+			UserID: "user-unrelated",
+		}},
+	}); err != nil {
+		t.Fatalf("SaveProfiles() error = %v", err)
+	}
+
+	oldGet := authKeychainGet
+	oldValidate := authValidateEntries
+	t.Cleanup(func() {
+		authKeychainGet = oldGet
+		authValidateEntries = oldValidate
+	})
+	var unrelatedReads atomic.Int32
+	var inventoryCalls atomic.Int32
+	authKeychainGet = func(service, account string) (string, error) {
+		if account == TokenAccountForCorpID("corp-unrelated") ||
+			account == TokenAccountForIdentity("corp-unrelated", "user-unrelated") {
+			unrelatedReads.Add(1)
+			return "", errors.New("unrelated profile ciphertext is unreadable")
+		}
+		return oldGet(service, account)
+	}
+	authValidateEntries = func(string) error {
+		inventoryCalls.Add(1)
+		return errors.New("unrelated orphan ciphertext is unreadable")
+	}
+
+	loginDone := startOAuthLogin(t, context.Background(), f)
+	callbackDone := make(chan oauthHTTPResult, 1)
+	go func() {
+		callbackDone <- getHTTPBody(f.callbackBase + CallbackPath + "?code=unrelated-safe")
+	}()
+	waitOAuthSignal(t, f.exchangeEntered, loginDone, "token exchange")
+	closeOAuthRelease(f.exchangeRelease)
+	waitOAuthSignal(t, f.statusEntered, loginDone, "CLI auth status check")
+	closeOAuthRelease(f.statusRelease)
+
+	select {
+	case callback := <-callbackDone:
+		if callback.err != nil || !strings.Contains(callback.body, "<html") {
+			t.Fatalf("OAuth callback body = %q, %v", callback.body, callback.err)
+		}
+	case <-time.After(oauthTestWaitTimeout):
+		t.Fatal("timed out waiting for OAuth callback")
+	}
+	result := awaitOAuthLogin(t, loginDone)
+	if result.err != nil || result.token == nil || result.token.AccessToken != "access" {
+		t.Fatalf("OAuthProvider.Login() = %#v, %v", result.token, result.err)
+	}
+	if got := unrelatedReads.Load(); got != 0 {
+		t.Fatalf("unrelated profile token reads = %d, want 0", got)
+	}
+	if got := inventoryCalls.Load(); got != 0 {
+		t.Fatalf("full inventory validation calls = %d, want 0", got)
+	}
+}
+
 func TestCrossPlatformCoverageOAuthLoginMissingCallbackCode(t *testing.T) {
 	f := newOAuthLoginFixture(t, func(int32) CLIAuthStatus {
 		return CLIAuthStatus{Success: true, Result: &CLIAuthResult{CLIAuthEnabled: true}}
@@ -685,7 +751,10 @@ func TestCrossPlatformCoverageOAuthRefreshAndParsingEdges(t *testing.T) {
 	resetAppConfigCache()
 	oauthHTTPClient = mcpSrv.Client()
 	mcpProvider := &OAuthProvider{configDir: configDir, httpClient: mcpSrv.Client()}
-	mcpOriginal := &TokenData{ClientID: "mcp-client", Source: "mcp", RefreshToken: "refresh", CorpID: "corp"}
+	mcpOriginal := &TokenData{
+		ClientID: "mcp-client", Source: "mcp", RefreshToken: "refresh",
+		CorpID: "corp", UserID: "user",
+	}
 	if updated, err := mcpProvider.refreshViaMCP(context.Background(), mcpOriginal); err != nil || updated.Source != "mcp" {
 		t.Fatalf("MCP refresh = %#v, %v", updated, err)
 	}
@@ -1020,6 +1089,10 @@ func TestCrossPlatformCoverageOAuthHelperRemainingEdges(t *testing.T) {
 	if _, err := p.exchangeCode(context.Background(), "code"); !errors.Is(err, fail) {
 		t.Fatalf("direct exchange request error = %v", err)
 	}
+	oauthHTTPClient = networkClient
+	if _, err := ExchangeCodeForToken(context.Background(), p.configDir, "code"); !errors.Is(err, fail) {
+		t.Fatalf("exchange wrapper request error = %v", err)
+	}
 	p.httpClient = responseClient("{")
 	if _, err := p.exchangeCode(context.Background(), "code"); err == nil {
 		t.Fatal("malformed direct exchange succeeded")

internal/auth/oauth_reauthorization_guidance_test.go

@@ -0,0 +1,62 @@
+// Copyright 2026 Alibaba Group
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package auth
+
+import (
+	"strconv"
+	"strings"
+	"testing"
+)
+
+func TestLegacyRefreshReauthorizationGuidanceTreatsProfileAsDisplayData(t *testing.T) {
+	tests := []struct {
+		name     string
+		selector string
+	}{
+		{name: "command substitution", selector: `external-$(touch marker)`},
+		{name: "backticks", selector: "external-`touch marker`"},
+		{name: "newline", selector: "external\ndws auth reset"},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			guidance := legacyRefreshReauthorizationGuidance(tt.selector)
+
+			if !strings.Contains(guidance, "dws auth login") ||
+				!strings.Contains(guidance, "--profile") ||
+				!strings.Contains(guidance, "profile 标识") {
+				t.Fatalf("guidance lacks stable reauthorization instructions: %q", guidance)
+			}
+			if strings.Contains(guidance, "dws auth login --profile") ||
+				strings.Contains(guidance, "--profile "+strconv.Quote(tt.selector)) {
+				t.Fatalf("guidance embeds untrusted selector in an executable command: %q", guidance)
+			}
+			if !strings.Contains(guidance, "profile: "+strconv.Quote(tt.selector)) {
+				t.Fatalf("guidance does not preserve selector as display data: %q", guidance)
+			}
+			if strings.Contains(tt.selector, "\n") && strings.Contains(guidance, tt.selector) {
+				t.Fatalf("guidance retained a raw selector newline: %q", guidance)
+			}
+		})
+	}
+}
+
+func TestLegacyRefreshReauthorizationGuidanceWithoutProfileStillExplainsLogin(t *testing.T) {
+	guidance := legacyRefreshReauthorizationGuidance("")
+	if !strings.Contains(guidance, "dws auth login") {
+		t.Fatalf("guidance = %q, want login instruction", guidance)
+	}
+	if strings.Contains(guidance, "--profile") || strings.Contains(guidance, "profile:") {
+		t.Fatalf("guidance = %q, should not invent an empty profile value", guidance)
+	}
+}

internal/auth/profiles.go

@@ -15,6 +15,7 @@ package auth
 
 import (
 	"context"
+	"encoding/base64"
 	"encoding/json"
 	"errors"
 	"fmt"
@@ -75,8 +76,11 @@ func withProfilesLock(configDir string, fn func() error) error {
 }
 
 const (
-	profilesJSONFile = "profiles.json"
-	profilesVersion  = 2
+	profilesJSONFile                  = "profiles.json"
+	profilesVersion                   = 2
+	profilesUnresolvedSelectorVersion = 3
+	profilesMaxVersion                = profilesUnresolvedSelectorVersion
+	unresolvedProfileSelectorPrefix   = "@legacy/"
 )
 
 const (
@@ -169,6 +173,7 @@ func SaveProfiles(configDir string, cfg *ProfilesConfig) error {
 		return err
 	}
 	normalizeProfilesConfig(cfg)
+	normalizeProfilesVersionForSelectors(cfg)
 	if err := profilesMkdirAll(configDir, config.DirPerm); err != nil {
 		return fmt.Errorf("create config dir: %w", err)
 	}
@@ -207,13 +212,16 @@ func ensureProfilesMigrationLocked(configDir string) error {
 	if err != nil {
 		return err
 	}
-	if cfg.Version > profilesVersion {
+	if cfg.Version > profilesMaxVersion {
 		return nil
 	}
 	if len(cfg.Profiles) == 0 {
 		// Version 2 with no profiles is an intentional logged-out tombstone.
 		// Never resurrect a stale legacy mirror after logout/reset.
 		if cfg.Version >= profilesVersion {
+			if normalizeProfilesVersionForSelectors(cfg) {
+				return profilesSave(configDir, cfg)
+			}
 			return nil
 		}
 		if !profilesTokenExists() {
@@ -240,6 +248,9 @@ func ensureProfilesMigrationLocked(configDir string) error {
 		cfg.OrgCurrentProfiles = make(map[string]string)
 	}
 	orgTokens := make(map[string]*TokenData)
+	var legacyToken *TokenData
+	var legacyTokenErr error
+	legacyTokenLoaded := false
 	for i := range cfg.Profiles {
 		p := &cfg.Profiles[i]
 		corpID := strings.TrimSpace(p.CorpID)
@@ -255,12 +266,75 @@ func ensureProfilesMigrationLocked(configDir string) error {
 			if loadErr != nil {
 				token = nil
 			}
+			var v2RepairProfile *Profile
+			if !legacySelectionState && errors.Is(loadErr, ErrTokenDataNotFound) {
+				v2RepairProfile = uniqueV2GlobalRepairProfile(cfg, corpID)
+				if v2RepairProfile != nil && strings.TrimSpace(v2RepairProfile.UserID) != "" {
+					_, identityErr := profilesLoadIdentity(corpID, v2RepairProfile.UserID)
+					switch {
+					case identityErr == nil:
+						// The exact identity is already usable. Do not let a stale
+						// global compatibility mirror recreate the organization slot.
+						v2RepairProfile = nil
+					case !errors.Is(identityErr, ErrTokenDataNotFound):
+						return identityErr
+					}
+				}
+			}
+			if (legacySelectionState || v2RepairProfile != nil) && errors.Is(loadErr, ErrTokenDataNotFound) {
+				// v1.0.50/1.0.51 installations can retain the selected
+				// organization only in the global compatibility slot. Consult
+				// that slot while migrating v1, or while repairing a non-empty
+				// v2 registry left half-migrated by an earlier CLI. The v2 path
+				// additionally requires one unambiguous profile, a missing exact
+				// slot when its userId is known, and a non-conflicting token userId.
+				// Persist the untouched organization mirror before identity
+				// enrichment below.
+				if !legacyTokenLoaded {
+					legacyToken, legacyTokenErr = profilesLoadLegacy()
+					legacyTokenLoaded = true
+				}
+				if legacyTokenErr != nil && !errors.Is(legacyTokenErr, ErrTokenDataNotFound) {
+					return legacyTokenErr
+				}
+				legacyMatchesProfile := legacySelectionState ||
+					legacyTokenMatchesV2RepairProfile(legacyToken, v2RepairProfile)
+				if legacyTokenErr == nil &&
+					legacyToken != nil &&
+					strings.TrimSpace(legacyToken.CorpID) == corpID &&
+					legacyMatchesProfile {
+					token = legacyToken
+					if err := profilesSaveCorp(corpID, token); err != nil {
+						return err
+					}
+				}
+			}
 			orgToken = token
 			orgTokens[corpID] = orgToken
 		}
 		if orgToken == nil {
 			continue
 		}
+		// v1.0.52 stored one token per organization. Some of those token blobs
+		// predate userId persistence even though profiles.json already recorded
+		// the account identity. Version 2 loads exact identities and therefore
+		// cannot safely use an organization mirror with no userId. A single
+		// profile with a known userId makes that association unambiguous,
+		// including when an earlier migration already bumped profiles.json to v2
+		// but failed before writing the identity slot. Enrich only the copy saved
+		// to that exact slot; never infer an identity for an organization with
+		// multiple accounts.
+		identityToken := orgToken
+		if strings.TrimSpace(orgToken.UserID) == "" &&
+			strings.TrimSpace(p.UserID) != "" &&
+			len(profilesForCorpID(cfg, corpID)) == 1 {
+			enriched := *orgToken
+			enriched.UserID = strings.TrimSpace(p.UserID)
+			if strings.TrimSpace(enriched.UserName) == "" {
+				enriched.UserName = strings.TrimSpace(p.UserName)
+			}
+			identityToken = &enriched
+		}
 		if strings.TrimSpace(p.UserID) == "" && strings.TrimSpace(orgToken.UserID) != "" {
 			if existing := findExactProfile(cfg, corpID, orgToken.UserID); existing != nil && existing != p {
 				p.CorpID = ""
@@ -273,12 +347,12 @@ func ensureProfilesMigrationLocked(configDir string) error {
 			}
 			changed = true
 		}
-		if strings.TrimSpace(p.UserID) == "" || strings.TrimSpace(orgToken.UserID) != strings.TrimSpace(p.UserID) {
+		if strings.TrimSpace(p.UserID) == "" || strings.TrimSpace(identityToken.UserID) != strings.TrimSpace(p.UserID) {
 			continue
 		}
 		_, identityErr := profilesLoadIdentity(corpID, p.UserID)
 		if errors.Is(identityErr, ErrTokenDataNotFound) {
-			if err := profilesSaveIdentity(corpID, p.UserID, orgToken); err != nil {
+			if err := profilesSaveIdentity(corpID, p.UserID, identityToken); err != nil {
 				return err
 			}
 		} else if identityErr != nil {
@@ -324,6 +398,10 @@ func ensureProfilesMigrationLocked(configDir string) error {
 		cfg.CurrentProfile = exact
 		changed = true
 	}
+	if exact := canonicalStoredSelector(cfg, cfg.PrimaryProfile); exact != "" && exact != cfg.PrimaryProfile {
+		cfg.PrimaryProfile = exact
+		changed = true
+	}
 	if legacySelectionState && cfg.CurrentProfile == "" {
 		if exact := canonicalStoredSelector(cfg, cfg.PrimaryProfile); exact != "" {
 			cfg.CurrentProfile = exact
@@ -345,12 +423,38 @@ func ensureProfilesMigrationLocked(configDir string) error {
 		cfg.Version = profilesVersion
 		changed = true
 	}
+	if normalizeProfilesVersionForSelectors(cfg) {
+		changed = true
+	}
 	if changed {
 		return profilesSave(configDir, cfg)
 	}
 	return nil
 }
 
+// uniqueV2GlobalRepairProfile returns the only profile that may safely be
+// recovered from the legacy global token mirror. A v2 registry with multiple
+// accounts in one organization is deliberately ineligible, even if one token
+// happens to carry a matching userId: the global slot is a mutable compatibility
+// mirror and is not authoritative account-selection state. A sole unresolved
+// profile is eligible only for organization-slot repair; the token matcher below
+// rejects any global token that tries to attach a userId to it.
+func uniqueV2GlobalRepairProfile(cfg *ProfilesConfig, corpID string) *Profile {
+	profiles := profilesForCorpID(cfg, corpID)
+	if len(profiles) != 1 {
+		return nil
+	}
+	return profiles[0]
+}
+
+func legacyTokenMatchesV2RepairProfile(data *TokenData, profile *Profile) bool {
+	if data == nil || strings.TrimSpace(data.CorpID) != strings.TrimSpace(profile.CorpID) {
+		return false
+	}
+	tokenUserID := strings.TrimSpace(data.UserID)
+	return tokenUserID == "" || tokenUserID == strings.TrimSpace(profile.UserID)
+}
+
 // UpsertProfileFromToken updates profiles.json after a successful login or refresh.
 func UpsertProfileFromToken(configDir string, data *TokenData) error {
 	return UpsertProfileFromTokenWithCurrent(configDir, data, true)
@@ -384,7 +488,9 @@ func upsertProfileFromToken(configDir string, cfg *ProfilesConfig, data *TokenDa
 		return err
 	}
 	normalizeProfilesConfig(cfg)
-	cfg.Version = profilesVersion
+	if cfg.Version < profilesVersion {
+		cfg.Version = profilesVersion
+	}
 	now := time.Now().Format(time.RFC3339)
 	userID := strings.TrimSpace(data.UserID)
 	var previousCurrent *Profile
@@ -392,7 +498,11 @@ func upsertProfileFromToken(configDir string, cfg *ProfilesConfig, data *TokenDa
 		previousCurrent, _, _ = resolveProfileSelection(configDir, cfg, cfg.CurrentProfile)
 	}
 	idx := profileIndexByIdentity(cfg, corpID, userID)
-	if idx < 0 && userID != "" {
+	if idx < 0 && userID != "" && len(profilesForCorpID(cfg, corpID)) == 1 {
+		// Upgrade an organization-scoped v1 profile only when it is the sole
+		// account in that organization. If exact identities already coexist
+		// with a blank profile, consuming the blank profile here would silently
+		// discard that unresolved historical account.
 		idx = legacyProfileIndexByCorpID(cfg, corpID)
 	}
 	if idx < 0 {
@@ -409,6 +519,7 @@ func upsertProfileFromToken(configDir string, cfg *ProfilesConfig, data *TokenDa
 			UpdatedAt:   now,
 		}
 		cfg.Profiles = append(cfg.Profiles, profile)
+		idx = len(cfg.Profiles) - 1
 	} else {
 		p := &cfg.Profiles[idx]
 		if userID != "" {
@@ -431,17 +542,33 @@ func upsertProfileFromToken(configDir string, cfg *ProfilesConfig, data *TokenDa
 		p.LastUsedAt = now
 		p.UpdatedAt = now
 	}
+	storedProfile := &cfg.Profiles[idx]
+	if userID == "" && len(profilesForCorpID(cfg, corpID)) > 1 &&
+		(strings.TrimSpace(storedProfile.Name) == corpID || profileNameTakenByOtherIdentity(cfg, storedProfile.Name, corpID, "")) {
+		// A blank profile needs a stable name when exact identities coexist in
+		// the same organization; the corpId selector denotes the organization
+		// as a whole and is therefore not an exact account selector.
+		storedProfile.Name = chooseProfileName(cfg, data)
+	}
 	if makeCurrent {
-		newSelector := profileSelector(corpID, userID)
-		if previousCurrent != nil && ProfileSelector(*previousCurrent) != newSelector {
-			cfg.PreviousProfile = ProfileSelector(*previousCurrent)
+		newSelector := storedProfileSelector(cfg, storedProfile)
+		if previousCurrent != nil && storedProfileSelector(cfg, previousCurrent) != newSelector {
+			cfg.PreviousProfile = storedProfileSelector(cfg, previousCurrent)
 		}
 		cfg.CurrentProfile = newSelector
-		setOrgCurrentProfile(cfg, corpID, newSelector)
+		if userID == "" {
+			delete(cfg.OrgCurrentProfiles, corpID)
+		} else {
+			setOrgCurrentProfile(cfg, corpID, newSelector)
+		}
 	}
 	if cfg.CurrentProfile == "" {
-		cfg.CurrentProfile = profileSelector(corpID, userID)
-		setOrgCurrentProfile(cfg, corpID, cfg.CurrentProfile)
+		cfg.CurrentProfile = storedProfileSelector(cfg, storedProfile)
+		if userID == "" {
+			delete(cfg.OrgCurrentProfiles, corpID)
+		} else {
+			setOrgCurrentProfile(cfg, corpID, cfg.CurrentProfile)
+		}
 	}
 	return profilesSave(configDir, cfg)
 }
@@ -452,6 +579,87 @@ func ProfileSelector(profile Profile) string {
 	return profileSelector(profile.CorpID, profile.UserID)
 }
 
+// storedProfileSelector returns a selector that remains exact inside
+// profiles.json. A blank userId has only an organization selector in the
+// public compatibility surface; when other accounts share that organization,
+// use the profile's unique local name so current/previous pointers do not
+// accidentally resolve to an exact account through OrgCurrentProfiles.
+func storedProfileSelector(cfg *ProfilesConfig, profile *Profile) string {
+	if profile == nil {
+		return ""
+	}
+	if strings.TrimSpace(profile.UserID) != "" {
+		return ProfileSelector(*profile)
+	}
+	if cfg == nil {
+		return strings.TrimSpace(profile.CorpID)
+	}
+	if len(profilesForCorpID(cfg, profile.CorpID)) <= 1 {
+		return strings.TrimSpace(profile.CorpID)
+	}
+	name := strings.TrimSpace(profile.Name)
+	if localProfileSelectorIsSafe(cfg, profile, name) {
+		return name
+	}
+	return unresolvedProfileSelector(profile.CorpID)
+}
+
+// ProfileSelectionSelector returns the stable selector used for one profile.
+// Exact identities use corpId:userId. A historical profile without userId
+// keeps the organization selector while it is the only account, and otherwise
+// uses either an unambiguous local name or a reserved, reversible selector.
+func ProfileSelectionSelector(profile Profile, cfg *ProfilesConfig) string {
+	return storedProfileSelector(cfg, &profile)
+}
+
+func localProfileSelectorIsSafe(cfg *ProfilesConfig, profile *Profile, name string) bool {
+	if cfg == nil || profile == nil {
+		return false
+	}
+	name = strings.TrimSpace(name)
+	if name == "" || strings.Contains(name, ":") || strings.HasPrefix(name, unresolvedProfileSelectorPrefix) {
+		return false
+	}
+	nameMatches := 0
+	for i := range cfg.Profiles {
+		candidate := &cfg.Profiles[i]
+		if strings.TrimSpace(candidate.Name) == name {
+			nameMatches++
+		}
+		// Organization selectors are resolved before ordinary local names.
+		// Never persist a local selector that can be captured by that grammar.
+		if strings.TrimSpace(candidate.CorpID) == name || strings.TrimSpace(candidate.CorpName) == name {
+			return false
+		}
+	}
+	return nameMatches == 1
+}
+
+func unresolvedProfileSelector(corpID string) string {
+	corpID = strings.TrimSpace(corpID)
+	if corpID == "" {
+		return ""
+	}
+	return unresolvedProfileSelectorPrefix + base64.RawURLEncoding.EncodeToString([]byte(corpID))
+}
+
+func parseUnresolvedProfileSelector(selector string) (string, bool) {
+	selector = strings.TrimSpace(selector)
+	if !strings.HasPrefix(selector, unresolvedProfileSelectorPrefix) {
+		return "", false
+	}
+	encoded := strings.TrimPrefix(selector, unresolvedProfileSelectorPrefix)
+	decoded, err := base64.RawURLEncoding.DecodeString(encoded)
+	if err != nil {
+		return "", false
+	}
+	corpID := strings.TrimSpace(string(decoded))
+	if corpID == "" || unresolvedProfileSelector(corpID) != selector {
+		return "", false
+	}
+	return corpID, true
+}
+
 // TokenProfileSelector returns the exact identity selector for token data when
 // its userId is known, otherwise it returns the historical corpId selector.
 func TokenProfileSelector(data *TokenData) string {
@@ -461,6 +669,34 @@ func TokenProfileSelector(data *TokenData) string {
 	return profileSelector(data.CorpID, data.UserID)
 }
 
+// StableTokenProfileSelector preserves the exact selector that loaded a token.
+// This matters for an unresolved historical account sharing an organization
+// with exact identities: reducing its selector to corpId would follow
+// OrgCurrentProfiles and could mark or reauthorize a different account.
+func StableTokenProfileSelector(configDir string, data *TokenData) string {
+	if selector := strings.TrimSpace(RuntimeProfile()); selector != "" {
+		return selector
+	}
+	fallback := TokenProfileSelector(data)
+	if data == nil {
+		return fallback
+	}
+	cfg, err := LoadProfiles(configDir)
+	if err != nil || cfg == nil || strings.TrimSpace(cfg.CurrentProfile) == "" {
+		return fallback
+	}
+	selector := canonicalStoredSelector(cfg, cfg.CurrentProfile)
+	if selector == "" {
+		selector = strings.TrimSpace(cfg.CurrentProfile)
+	}
+	profile, _, err := resolveProfileSelection(configDir, cfg, selector)
+	if err != nil || profile == nil ||
+		!sameProfileIdentity(profile.CorpID, profile.UserID, data.CorpID, data.UserID) {
+		return fallback
+	}
+	return storedProfileSelector(cfg, profile)
+}
+
 func profileSelector(corpID, userID string) string {
 	corpID = strings.TrimSpace(corpID)
 	userID = strings.TrimSpace(userID)
@@ -609,7 +845,13 @@ func setCurrentProfileLocked(configDir, selector string) (*Profile, error) {
 		return nil, err
 	}
 	originalCfg := cloneProfilesConfig(cfg)
-	mirrors, err := snapshotProfileSelectionMirrors(configDir, p.CorpID)
+	syncOrganization := shouldSyncOrganizationMirror(cfg, *p)
+	if !syncOrganization {
+		if err := validateIdentityOnlyProfileToken(*p); err != nil {
+			return nil, err
+		}
+	}
+	mirrors, err := snapshotProfileSelectionMirrors(configDir, p.CorpID, syncOrganization)
 	if err != nil {
 		return nil, err
 	}
@@ -617,20 +859,26 @@ func setCurrentProfileLocked(configDir, selector string) (*Profile, error) {
 	if strings.TrimSpace(cfg.CurrentProfile) != "" {
 		previousCurrent, _, _ = resolveProfileSelection(configDir, cfg, cfg.CurrentProfile)
 	}
-	storedSelector := ProfileSelector(*p)
+	storedSelector := storedProfileSelector(cfg, p)
 	if cfg.CurrentProfile != storedSelector {
 		if previousCurrent != nil {
-			cfg.PreviousProfile = ProfileSelector(*previousCurrent)
+			cfg.PreviousProfile = storedProfileSelector(cfg, previousCurrent)
 		}
 		cfg.CurrentProfile = storedSelector
 	}
-	setOrgCurrentProfile(cfg, p.CorpID, storedSelector)
+	if strings.TrimSpace(p.UserID) == "" {
+		delete(cfg.OrgCurrentProfiles, strings.TrimSpace(p.CorpID))
+	} else {
+		setOrgCurrentProfile(cfg, p.CorpID, storedSelector)
+	}
 	touchProfileUsage(p)
 	if err := profilesSave(configDir, cfg); err != nil {
 		return nil, err
 	}
-	if err := syncOrganizationTokenMirrorForProfile(*p); err != nil {
-		return nil, rollbackProfileSelection(configDir, originalCfg, p.CorpID, mirrors, err)
+	if syncOrganization {
+		if err := syncOrganizationTokenMirrorForProfile(*p); err != nil {
+			return nil, rollbackProfileSelection(configDir, originalCfg, p.CorpID, mirrors, err)
+		}
 	}
 	if err := profilesSyncLegacyMirror(configDir); err != nil {
 		return nil, rollbackProfileSelection(configDir, originalCfg, p.CorpID, mirrors, err)
@@ -669,7 +917,13 @@ func usePreviousProfileLocked(configDir string) (*Profile, error) {
 		return nil, fmt.Errorf("resolve previous profile %q: %w", prev, err)
 	}
 	originalCfg := cloneProfilesConfig(cfg)
-	mirrors, err := snapshotProfileSelectionMirrors(configDir, p.CorpID)
+	syncOrganization := shouldSyncOrganizationMirror(cfg, *p)
+	if !syncOrganization {
+		if err := validateIdentityOnlyProfileToken(*p); err != nil {
+			return nil, err
+		}
+	}
+	mirrors, err := snapshotProfileSelectionMirrors(configDir, p.CorpID, syncOrganization)
 	if err != nil {
 		return nil, err
 	}
@@ -677,19 +931,25 @@ func usePreviousProfileLocked(configDir string) (*Profile, error) {
 	if strings.TrimSpace(cfg.CurrentProfile) != "" {
 		current, _, _ = resolveProfileSelection(configDir, cfg, cfg.CurrentProfile)
 	}
-	cfg.CurrentProfile = ProfileSelector(*p)
+	cfg.CurrentProfile = storedProfileSelector(cfg, p)
 	if current != nil {
-		cfg.PreviousProfile = ProfileSelector(*current)
+		cfg.PreviousProfile = storedProfileSelector(cfg, current)
 	} else {
 		cfg.PreviousProfile = ""
 	}
-	setOrgCurrentProfile(cfg, p.CorpID, ProfileSelector(*p))
+	if strings.TrimSpace(p.UserID) == "" {
+		delete(cfg.OrgCurrentProfiles, strings.TrimSpace(p.CorpID))
+	} else {
+		setOrgCurrentProfile(cfg, p.CorpID, ProfileSelector(*p))
+	}
 	touchProfileUsage(p)
 	if err := profilesSave(configDir, cfg); err != nil {
 		return nil, err
 	}
-	if err := syncOrganizationTokenMirrorForProfile(*p); err != nil {
-		return nil, rollbackProfileSelection(configDir, originalCfg, p.CorpID, mirrors, err)
+	if syncOrganization {
+		if err := syncOrganizationTokenMirrorForProfile(*p); err != nil {
+			return nil, rollbackProfileSelection(configDir, originalCfg, p.CorpID, mirrors, err)
+		}
 	}
 	if err := profilesSyncLegacyMirror(configDir); err != nil {
 		return nil, rollbackProfileSelection(configDir, originalCfg, p.CorpID, mirrors, err)
@@ -733,6 +993,21 @@ func removeProfileLocked(configDir, selector string) (*Profile, error) {
 		return nil, err
 	}
 	removed := *p
+	originalCurrentSelector := strings.TrimSpace(cfg.CurrentProfile)
+	originalOrganizationCurrent := strings.TrimSpace(cfg.OrgCurrentProfiles[strings.TrimSpace(removed.CorpID)])
+	pointers := []*string{&cfg.PrimaryProfile, &cfg.CurrentProfile, &cfg.PreviousProfile}
+	pointerMatches := make([]bool, len(pointers))
+	for i, pointer := range pointers {
+		selected, _, resolveErr := resolveProfileSelection(configDir, cfg, *pointer)
+		if resolveErr == nil && selected != nil {
+			if exact {
+				pointerMatches[i] =
+					sameProfileIdentity(selected.CorpID, selected.UserID, removed.CorpID, removed.UserID)
+			} else {
+				pointerMatches[i] = strings.TrimSpace(selected.CorpID) == strings.TrimSpace(removed.CorpID)
+			}
+		}
+	}
 	kept := cfg.Profiles[:0]
 	for _, profile := range cfg.Profiles {
 		remove := profile.CorpID == removed.CorpID
@@ -748,7 +1023,7 @@ func removeProfileLocked(configDir, selector string) (*Profile, error) {
 	if exact && len(profilesForCorpID(cfg, removed.CorpID)) > 0 {
 		remaining := profilesForCorpID(cfg, removed.CorpID)
 		if len(remaining) == 1 {
-			replacementSelector = ProfileSelector(*remaining[0])
+			replacementSelector = storedProfileSelector(cfg, remaining[0])
 		}
 	}
 	if exact {
@@ -762,15 +1037,14 @@ func removeProfileLocked(configDir, selector string) (*Profile, error) {
 	} else {
 		delete(cfg.OrgCurrentProfiles, removed.CorpID)
 	}
-	pointers := []*string{&cfg.PrimaryProfile, &cfg.CurrentProfile, &cfg.PreviousProfile}
-	for _, pointer := range pointers {
+	for i, pointer := range pointers {
 		if exact {
-			if selectorMatchesIdentity(*pointer, removed) {
+			if pointerMatches[i] {
 				*pointer = replacementSelector
 			}
 			continue
 		}
-		if selectorTargetsCorp(*pointer, removed.CorpID) {
+		if pointerMatches[i] || selectorTargetsCorp(*pointer, removed.CorpID) {
 			*pointer = ""
 		}
 	}
@@ -779,7 +1053,35 @@ func removeProfileLocked(configDir, selector string) (*Profile, error) {
 			cfg.CurrentProfile = previous
 			cfg.PreviousProfile = ""
 		} else if len(cfg.Profiles) == 1 {
-			cfg.CurrentProfile = ProfileSelector(cfg.Profiles[0])
+			cfg.CurrentProfile = storedProfileSelector(cfg, &cfg.Profiles[0])
+		}
+	}
+	if cfg.PreviousProfile != "" && cfg.PreviousProfile == cfg.CurrentProfile {
+		cfg.PreviousProfile = ""
+	}
+	currentSelectionChanged := strings.TrimSpace(cfg.CurrentProfile) != originalCurrentSelector
+	organizationCurrentChanged := strings.TrimSpace(cfg.OrgCurrentProfiles[strings.TrimSpace(removed.CorpID)]) != originalOrganizationCurrent
+	if strings.TrimSpace(cfg.CurrentProfile) != "" {
+		if current, _, resolveErr := resolveProfileSelection(configDir, cfg, cfg.CurrentProfile); resolveErr == nil && current != nil {
+			if (currentSelectionChanged || organizationCurrentChanged) &&
+				strings.TrimSpace(current.CorpID) == strings.TrimSpace(removed.CorpID) &&
+				unresolvedProfileForCorp(cfg, removed.CorpID) != nil {
+				if strings.TrimSpace(current.UserID) == "" {
+					delete(cfg.OrgCurrentProfiles, strings.TrimSpace(current.CorpID))
+				} else {
+					setOrgCurrentProfile(cfg, current.CorpID, storedProfileSelector(cfg, current))
+				}
+			}
+		}
+	}
+	// Removing the exact identity that forced a blank sibling to use the v3
+	// reserved selector can make that blank profile the sole account in its
+	// organization. Re-canonicalize every surviving pointer against the final
+	// profile set before SaveProfiles derives the schema version, so the pointer
+	// collapses back to the v2 corpId grammar instead of pinning the file at v3.
+	for _, pointer := range pointers {
+		if canonical := canonicalStoredSelector(cfg, *pointer); canonical != "" {
+			*pointer = canonical
 		}
 	}
 	if cfg.PreviousProfile != "" && cfg.PreviousProfile == cfg.CurrentProfile {
@@ -808,6 +1110,9 @@ func selectorTargetsCorp(selector, corpID string) bool {
 	if selector == corpID {
 		return true
 	}
+	if selectedCorpID, unresolved := parseUnresolvedProfileSelector(selector); unresolved {
+		return selectedCorpID == corpID
+	}
 	selectedCorpID, _, exact := ParseIdentitySelector(selector)
 	return exact && selectedCorpID == corpID
 }
@@ -840,28 +1145,72 @@ func markProfileStatusLocked(configDir, selector, status string) error {
 }
 
 func ensureProfilesWritable(cfg *ProfilesConfig) error {
-	if cfg != nil && cfg.Version > profilesVersion {
+	if cfg != nil && cfg.Version > profilesMaxVersion {
 		return fmt.Errorf(
 			"profiles.json version %d is newer than supported version %d; upgrade dws before changing profiles",
 			cfg.Version,
-			profilesVersion,
+			profilesMaxVersion,
 		)
 	}
 	return nil
 }
 
+// normalizeProfilesVersionForSelectors derives the persisted schema version
+// from the final normalized selector grammar. Keep v3 only while a legal
+// reserved unresolved-identity selector remains on disk; once completion,
+// deletion, or canonicalization removes that grammar, the file is again safe
+// for v2 clients and should downgrade to v2.
+func normalizeProfilesVersionForSelectors(cfg *ProfilesConfig) bool {
+	if cfg == nil || cfg.Version > profilesMaxVersion {
+		return false
+	}
+
+	target := cfg.Version
+	if profilesConfigContainsUnresolvedSelector(cfg) {
+		target = profilesUnresolvedSelectorVersion
+	} else if cfg.Version >= profilesVersion {
+		target = profilesVersion
+	}
+	if target == cfg.Version {
+		return false
+	}
+	cfg.Version = target
+	return true
+}
+
+func profilesConfigContainsUnresolvedSelector(cfg *ProfilesConfig) bool {
+	if cfg == nil {
+		return false
+	}
+	for _, selector := range []string{cfg.PrimaryProfile, cfg.CurrentProfile, cfg.PreviousProfile} {
+		if _, unresolved := parseUnresolvedProfileSelector(selector); unresolved {
+			return true
+		}
+	}
+	for _, selector := range cfg.OrgCurrentProfiles {
+		if _, unresolved := parseUnresolvedProfileSelector(selector); unresolved {
+			return true
+		}
+	}
+	return false
+}
+
 type profileSelectionMirrorSnapshot struct {
 	organization tokenSlotSnapshot
 	legacy       tokenSlotSnapshot
 	marker       tokenMarkerSnapshot
 }
 
-func snapshotProfileSelectionMirrors(configDir, corpID string) (profileSelectionMirrorSnapshot, error) {
-	organization, err := snapshotTokenSlot(func() (*TokenData, error) {
-		return profilesLoadCorp(corpID)
-	})
-	if err != nil {
-		return profileSelectionMirrorSnapshot{}, err
+func snapshotProfileSelectionMirrors(configDir, corpID string, includeOrganization bool) (profileSelectionMirrorSnapshot, error) {
+	var organization tokenSlotSnapshot
+	if includeOrganization {
+		var err error
+		organization, err = snapshotTokenSlot(func() (*TokenData, error) {
+			return profilesLoadCorp(corpID)
+		})
+		if err != nil {
+			return profileSelectionMirrorSnapshot{}, err
+		}
 	}
 	legacy, err := snapshotTokenSlot(profilesLoadLegacy)
 	if err != nil {
@@ -889,12 +1238,14 @@ func rollbackProfileSelection(
 	if err := profilesSave(configDir, cloneProfilesConfig(cfg)); err != nil {
 		rollbackErr = errors.Join(rollbackErr, err)
 	}
-	if mirrors.organization.exists {
-		if err := profilesSaveCorp(corpID, mirrors.organization.token); err != nil {
+	if mirrors.organization.known {
+		if mirrors.organization.exists {
+			if err := profilesSaveCorp(corpID, mirrors.organization.token); err != nil {
+				rollbackErr = errors.Join(rollbackErr, err)
+			}
+		} else if err := profilesDeleteCorp(corpID); err != nil {
 			rollbackErr = errors.Join(rollbackErr, err)
 		}
-	} else if err := profilesDeleteCorp(corpID); err != nil {
-		rollbackErr = errors.Join(rollbackErr, err)
 	}
 	if mirrors.legacy.exists {
 		if err := profilesSaveLegacy(mirrors.legacy.token); err != nil {
@@ -975,33 +1326,48 @@ func syncOrganizationTokenMirrorForProfile(profile Profile) error {
 }
 
 func loadTokenForProfileIdentity(profile Profile) (*TokenData, error) {
-	if strings.TrimSpace(profile.UserID) != "" {
-		data, err := profilesLoadIdentity(profile.CorpID, profile.UserID)
-		if err == nil {
-			return data, nil
-		}
-		if !errors.Is(err, ErrTokenDataNotFound) {
+	if strings.TrimSpace(profile.UserID) == "" {
+		data, err := profilesLoadCorp(profile.CorpID)
+		if err != nil {
 			return nil, err
 		}
-		orgData, orgErr := profilesLoadCorp(profile.CorpID)
-		if orgErr != nil {
-			if errors.Is(orgErr, ErrTokenDataNotFound) {
-				return nil, err
-			}
-			return nil, orgErr
+		if data == nil {
+			return nil, ErrTokenDataNotFound
 		}
-		if strings.TrimSpace(orgData.UserID) == "" {
-			return nil, fmt.Errorf("organization token mirror for corpId %q has no userId; cannot use it for profile %q", profile.CorpID, ProfileSelector(profile))
+		if strings.TrimSpace(data.UserID) != "" {
+			return nil, fmt.Errorf(
+				"organization token mirror for corpId %q belongs to userId %q; cannot use it for unresolved profile %q",
+				profile.CorpID,
+				data.UserID,
+				profile.Name,
+			)
 		}
-		if strings.TrimSpace(orgData.UserID) != strings.TrimSpace(profile.UserID) {
-			return nil, err
-		}
-		if saveErr := profilesSaveIdentity(profile.CorpID, profile.UserID, orgData); saveErr != nil {
-			return nil, saveErr
-		}
-		return orgData, nil
+		return data, nil
 	}
-	return profilesLoadCorp(profile.CorpID)
+	data, err := profilesLoadIdentity(profile.CorpID, profile.UserID)
+	if err == nil {
+		return data, nil
+	}
+	if !errors.Is(err, ErrTokenDataNotFound) {
+		return nil, err
+	}
+	orgData, orgErr := profilesLoadCorp(profile.CorpID)
+	if orgErr != nil {
+		if errors.Is(orgErr, ErrTokenDataNotFound) {
+			return nil, err
+		}
+		return nil, orgErr
+	}
+	if strings.TrimSpace(orgData.UserID) == "" {
+		return nil, fmt.Errorf("organization token mirror for corpId %q has no userId; cannot use it for profile %q", profile.CorpID, ProfileSelector(profile))
+	}
+	if strings.TrimSpace(orgData.UserID) != strings.TrimSpace(profile.UserID) {
+		return nil, err
+	}
+	if saveErr := profilesSaveIdentity(profile.CorpID, profile.UserID, orgData); saveErr != nil {
+		return nil, saveErr
+	}
+	return orgData, nil
 }
 
 func normalizeProfilesConfig(cfg *ProfilesConfig) {
@@ -1127,6 +1493,12 @@ func resolveProfileSelection(_ string, cfg *ProfilesConfig, selector string) (*P
 	if selector == "" {
 		return nil, false, fmt.Errorf("profile selector is empty")
 	}
+	if corpID, unresolved := parseUnresolvedProfileSelector(selector); unresolved {
+		if profile := unresolvedProfileForCorp(cfg, corpID); profile != nil {
+			return profile, true, nil
+		}
+		return nil, true, fmt.Errorf("historical profile for organization %q not found", corpID)
+	}
 
 	if organization, account, compound := ParseIdentitySelector(selector); compound {
 		corpID, err := resolveOrganizationCorpID(cfg, organization)
@@ -1218,6 +1590,12 @@ func resolveProfileDeletionSelection(cfg *ProfilesConfig, selector string) (*Pro
 	if selector == "" {
 		return nil, false, fmt.Errorf("profile selector is empty")
 	}
+	if corpID, unresolved := parseUnresolvedProfileSelector(selector); unresolved {
+		if profile := unresolvedProfileForCorp(cfg, corpID); profile != nil {
+			return profile, true, nil
+		}
+		return nil, true, fmt.Errorf("historical profile for organization %q not found", corpID)
+	}
 	if _, _, compound := ParseIdentitySelector(selector); compound {
 		return resolveProfileSelection("", cfg, selector)
 	}
@@ -1303,6 +1681,12 @@ func resolveOrganizationDefault(cfg *ProfilesConfig, corpID, displaySelector str
 			return p, false, nil
 		}
 	}
+	if unresolved := unresolvedProfileForCorp(cfg, corpID); unresolved != nil {
+		// With no exact organization-current selection, the organization slot
+		// belongs to the sole unresolved historical account. Do not choose an
+		// arbitrary exact identity merely because it shares the corpId.
+		return unresolved, false, nil
+	}
 	if len(profiles) == 1 {
 		return profiles[0], false, nil
 	}
@@ -1314,12 +1698,18 @@ func resolveOrganizationDefault(cfg *ProfilesConfig, corpID, displaySelector str
 }
 
 func profileSelectorCandidates(profiles []*Profile) []string {
+	cfg := &ProfilesConfig{Profiles: make([]Profile, 0, len(profiles))}
+	for _, profile := range profiles {
+		if profile != nil {
+			cfg.Profiles = append(cfg.Profiles, *profile)
+		}
+	}
 	candidates := make([]string, 0, len(profiles))
 	for _, p := range profiles {
 		if p == nil {
 			continue
 		}
-		candidates = append(candidates, ProfileSelector(*p))
+		candidates = append(candidates, storedProfileSelector(cfg, p))
 	}
 	sort.Strings(candidates)
 	return candidates
@@ -1341,18 +1731,39 @@ func canonicalStoredSelector(cfg *ProfilesConfig, selector string) string {
 	if selector == "" {
 		return ""
 	}
+	if corpID, unresolved := parseUnresolvedProfileSelector(selector); unresolved {
+		if profile := unresolvedProfileForCorp(cfg, corpID); profile != nil {
+			return storedProfileSelector(cfg, profile)
+		}
+		return ""
+	}
 	if corpID, userID, exact := ParseIdentitySelector(selector); exact {
 		if p := findExactProfile(cfg, corpID, userID); p != nil {
 			return ProfileSelector(*p)
 		}
+		// A colon-containing legacy local name is recoverable only when it does
+		// not name a real exact identity. Exact corpId:userId always wins.
+		if profile := unresolvedProfileForLocalName(cfg, selector); profile != nil {
+			return storedProfileSelector(cfg, profile)
+		}
 		return ""
 	}
+	// Older multi-account writers stored the unresolved profile's local name.
+	// Recover it only when no profile gives the same text organization-selector
+	// meaning. CorpId and CorpName have always outranked local names in the
+	// public resolver; migration must preserve that precedence instead of
+	// silently redirecting one organization's selector to another blank profile.
+	if !selectorConflictsWithOrganizationGrammar(cfg, selector) {
+		if profile := unresolvedProfileForLocalName(cfg, selector); profile != nil {
+			return storedProfileSelector(cfg, profile)
+		}
+	}
 	if profiles := profilesForCorpID(cfg, selector); len(profiles) > 0 {
 		if exact := exactProfileSelectorForCorp(cfg, selector, cfg.OrgCurrentProfiles[selector]); exact != "" {
 			return exact
 		}
 		if len(profiles) == 1 {
-			return ProfileSelector(*profiles[0])
+			return storedProfileSelector(cfg, profiles[0])
 		}
 		return ""
 	}
@@ -1360,7 +1771,21 @@ func canonicalStoredSelector(cfg *ProfilesConfig, selector string) string {
 	if err != nil || p == nil {
 		return ""
 	}
-	return ProfileSelector(*p)
+	return storedProfileSelector(cfg, p)
+}
+
+func selectorConflictsWithOrganizationGrammar(cfg *ProfilesConfig, selector string) bool {
+	if cfg == nil {
+		return false
+	}
+	selector = strings.TrimSpace(selector)
+	for i := range cfg.Profiles {
+		if strings.TrimSpace(cfg.Profiles[i].CorpID) == selector ||
+			strings.TrimSpace(cfg.Profiles[i].CorpName) == selector {
+			return true
+		}
+	}
+	return false
 }
 
 func setOrgCurrentProfile(cfg *ProfilesConfig, corpID, selector string) {
@@ -1445,6 +1870,79 @@ func profilesForCorpID(cfg *ProfilesConfig, corpID string) []*Profile {
 	return result
 }
 
+func unresolvedProfileForCorp(cfg *ProfilesConfig, corpID string) *Profile {
+	if cfg == nil {
+		return nil
+	}
+	corpID = strings.TrimSpace(corpID)
+	for i := range cfg.Profiles {
+		profile := &cfg.Profiles[i]
+		if strings.TrimSpace(profile.CorpID) == corpID && strings.TrimSpace(profile.UserID) == "" {
+			return profile
+		}
+	}
+	return nil
+}
+
+func unresolvedProfileForLocalName(cfg *ProfilesConfig, name string) *Profile {
+	if cfg == nil {
+		return nil
+	}
+	name = strings.TrimSpace(name)
+	if name == "" {
+		return nil
+	}
+	var match *Profile
+	for i := range cfg.Profiles {
+		profile := &cfg.Profiles[i]
+		if strings.TrimSpace(profile.UserID) != "" || strings.TrimSpace(profile.Name) != name ||
+			len(profilesForCorpID(cfg, profile.CorpID)) <= 1 {
+			continue
+		}
+		if match != nil {
+			return nil
+		}
+		match = profile
+	}
+	return match
+}
+
+// When a blank profile coexists with exact accounts, the organization slot is
+// that unresolved profile's only canonical credential. Exact identities must
+// remain in their identity slots and may still become global current without
+// overwriting the organization slot.
+func shouldSyncOrganizationMirror(cfg *ProfilesConfig, profile Profile) bool {
+	return strings.TrimSpace(profile.UserID) == "" || unresolvedProfileForCorp(cfg, profile.CorpID) == nil
+}
+
+// validateIdentityOnlyProfileToken verifies the canonical token slot before a
+// profile selection is persisted. This path is used only when an unresolved
+// profile owns the organization slot, so an exact identity must not fall back
+// to that slot. It is deliberately read-only: a rejected switch leaves every
+// selection pointer and compatibility mirror untouched.
+func validateIdentityOnlyProfileToken(profile Profile) error {
+	corpID := strings.TrimSpace(profile.CorpID)
+	userID := strings.TrimSpace(profile.UserID)
+	if corpID == "" || userID == "" {
+		return ErrTokenDataNotFound
+	}
+	data, err := profilesLoadIdentity(corpID, userID)
+	if err != nil {
+		return fmt.Errorf("load token for profile %q: %w", ProfileSelector(profile), err)
+	}
+	if data == nil {
+		return fmt.Errorf("load token for profile %q: %w", ProfileSelector(profile), ErrTokenDataNotFound)
+	}
+	if !sameProfileIdentity(data.CorpID, data.UserID, corpID, userID) {
+		return fmt.Errorf(
+			"token in profile slot %q belongs to %q; identity does not match selected profile",
+			ProfileSelector(profile),
+			profileSelector(data.CorpID, data.UserID),
+		)
+	}
+	return nil
+}
+
 func profileSelectorReferenceExists(cfg *ProfilesConfig, selector string) bool {
 	if cfg == nil {
 		return false
@@ -1453,6 +1951,12 @@ func profileSelectorReferenceExists(cfg *ProfilesConfig, selector string) bool {
 	if selector == "" {
 		return false
 	}
+	if corpID, unresolved := parseUnresolvedProfileSelector(selector); unresolved {
+		return unresolvedProfileForCorp(cfg, corpID) != nil
+	}
+	if unresolvedProfileForLocalName(cfg, selector) != nil {
+		return true
+	}
 	if corpID, userID, exact := ParseIdentitySelector(selector); exact {
 		if findExactProfile(cfg, corpID, userID) != nil {
 			return true

internal/auth/refresh_failure_test.go

@@ -10,6 +10,7 @@ import (
 	"net/http"
 	"net/http/httptest"
 	"net/url"
+	"strconv"
 	"strings"
 	"testing"
 	"time"
@@ -152,4 +153,103 @@ func TestCrossPlatformCoverageGetTokenSnapshotOnlyExpiresProfileForNonTransientR
 	if markCalls != 1 {
 		t.Fatalf("terminal refresh marked profile expired %d times, want 1", markCalls)
 	}
+
+	oauthRefreshToken = func(*OAuthProvider, context.Context, *TokenData) (*TokenData, error) {
+		return nil, &MCPTokenExchangeError{
+			Code:    legacyMCPRefreshRejectedCode,
+			Message: "不合法的临时授权码",
+		}
+	}
+	_, err := provider.GetTokenSnapshot(context.Background())
+	if err == nil || !strings.Contains(err.Error(), "dws auth login") ||
+		!strings.Contains(err.Error(), "--profile") ||
+		!strings.Contains(err.Error(), `profile: "corp:user"`) ||
+		strings.Contains(err.Error(), `dws auth login --profile "corp:user"`) ||
+		!strings.Contains(err.Error(), legacyMCPRefreshRejectedCode) {
+		t.Fatalf("legacy MCP refresh guidance = %v", err)
+	}
+	var exchangeErr *MCPTokenExchangeError
+	if !errors.As(err, &exchangeErr) || !exchangeErr.requiresReauthorization() {
+		t.Fatalf("legacy MCP refresh cause was not preserved: %v", err)
+	}
+	if markCalls != 2 {
+		t.Fatalf("legacy MCP rejection marked profile expired %d times, want 2", markCalls)
+	}
+
+	SetRuntimeProfile("External Worker")
+	_, err = provider.GetTokenSnapshot(context.Background())
+	SetRuntimeProfile("")
+	if err == nil || !strings.Contains(err.Error(), "dws auth login") ||
+		!strings.Contains(err.Error(), `profile: "External Worker"`) ||
+		strings.Contains(err.Error(), `dws auth login --profile "External Worker"`) {
+		t.Fatalf("legacy MCP refresh guidance did not isolate spaced selector as display data: %v", err)
+	}
+	if markCalls != 3 {
+		t.Fatalf("spaced legacy MCP rejection marked profile expired %d times, want 3", markCalls)
+	}
+}
+
+func TestCrossPlatformCoverageLegacyRefreshFailureKeepsBlankCurrentSelectorIsolated(t *testing.T) {
+	fixture := seedBlankProfileSelectorFixture(t, "Fixture Organization", "Fixture Organization", true)
+	expired := *fixture.blankToken
+	expired.ExpiresAt = time.Now().Add(-time.Hour)
+	expired.RefreshExpAt = time.Now().Add(time.Hour)
+
+	oldLoad := oauthLoadToken
+	oldLoadLocked := oauthLoadTokenLocked
+	oldAcquire := oauthAcquireLock
+	oldRefresh := oauthRefreshToken
+	oldMark := oauthMarkProfile
+	oldEdition := edition.Get()
+	t.Cleanup(func() {
+		oauthLoadToken = oldLoad
+		oauthLoadTokenLocked = oldLoadLocked
+		oauthAcquireLock = oldAcquire
+		oauthRefreshToken = oldRefresh
+		oauthMarkProfile = oldMark
+		edition.Override(oldEdition)
+	})
+	edition.Override(&edition.Hooks{})
+	oauthLoadToken = func(string) (*TokenData, error) { return &expired, nil }
+	oauthLoadTokenLocked = func(string, string) (*TokenData, error) { return &expired, nil }
+	oauthAcquireLock = func(context.Context, string) (*DualLock, error) { return &DualLock{}, nil }
+	oauthRefreshToken = func(*OAuthProvider, context.Context, *TokenData) (*TokenData, error) {
+		return nil, &MCPTokenExchangeError{
+			Code:    legacyMCPRefreshRejectedCode,
+			Message: "legacy refresh rejected",
+		}
+	}
+	var markedSelector string
+	oauthMarkProfile = func(configDir, selector, status string) error {
+		markedSelector = selector
+		return MarkProfileStatus(configDir, selector, status)
+	}
+
+	provider := NewOAuthProvider(fixture.configDir, nil)
+	_, err := provider.GetTokenSnapshot(context.Background())
+	if err == nil || !strings.Contains(err.Error(), "dws auth login") ||
+		!strings.Contains(err.Error(), "--profile") ||
+		!strings.Contains(err.Error(), "profile: "+strconv.Quote(fixture.blankSelector)) ||
+		strings.Contains(err.Error(), "dws auth login --profile") {
+		t.Fatalf("legacy blank refresh guidance = %v, want selector %q", err, fixture.blankSelector)
+	}
+	if markedSelector != fixture.blankSelector {
+		t.Fatalf("marked selector = %q, want blank %q", markedSelector, fixture.blankSelector)
+	}
+	cfg, loadErr := LoadProfiles(fixture.configDir)
+	if loadErr != nil {
+		t.Fatalf("LoadProfiles() error = %v", loadErr)
+	}
+	for _, profile := range cfg.Profiles {
+		switch profile.UserID {
+		case "":
+			if profile.Status != ProfileStatusExpired {
+				t.Fatalf("blank profile status = %q, want expired", profile.Status)
+			}
+		case fixture.exactUserID:
+			if profile.Status != ProfileStatusActive {
+				t.Fatalf("exact profile status = %q, want active", profile.Status)
+			}
+		}
+	}
 }

internal/auth/remaining_coverage_test.go

@@ -43,6 +43,9 @@ func TestCrossPlatformCoverageTokenPersistencePreflightRemainingEdges(t *testing
 	})
 
 	edition.Override(&edition.Hooks{SaveToken: func(string, []byte) error { return nil }})
+	if err := prepareLoginPersistence(t.TempDir()); err != nil {
+		t.Fatal(err)
+	}
 	if err := preflightTokenPersistence(t.TempDir()); err != nil {
 		t.Fatal(err)
 	}
@@ -55,6 +58,9 @@ func TestCrossPlatformCoverageTokenPersistencePreflightRemainingEdges(t *testing
 	authValidateEntries = func(string) error { return nil }
 	profileFail := errors.New("profile load failed")
 	profilesReadFile = func(string) ([]byte, error) { return nil, profileFail }
+	if err := prepareLoginPersistence(t.TempDir()); !errors.Is(err, profileFail) {
+		t.Fatalf("schema preflight profile load error = %v", err)
+	}
 	if err := preflightTokenPersistence(t.TempDir()); !errors.Is(err, profileFail) {
 		t.Fatalf("profile load error = %v", err)
 	}

internal/auth/token.go

@@ -59,27 +59,28 @@ var (
 		profile, _, err := resolveProfileForLoadLocked(configDir, selector)
 		return profile, err
 	}
-	tokenResolveDeletion        = resolveProfileDeletionSelection
-	tokenResolveSelection       = resolveProfileSelection
-	tokenUpsertProfile          = upsertProfileFromTokenWithCurrentLocked
-	tokenRemoveProfile          = removeProfileLocked
-	tokenSyncLegacyMirror       = syncLegacyTokenMirrorLocked
-	tokenSyncOrganizationMirror = syncOrganizationTokenMirrorForProfile
-	tokenLoadProfiles           = LoadProfiles
-	tokenSaveProfiles           = SaveProfiles
-	tokenWriteMarker            = WriteTokenMarker
-	tokenWriteManualMarker      = WriteManualTokenMarker
-	tokenDeleteMarker           = DeleteTokenMarker
-	tokenParseURL               = url.Parse
-	tokenNewRequest             = http.NewRequestWithContext
-	tokenDefaultConfigDir       = getDefaultConfigDir
-	tokenLoadData               = LoadTokenData
-	tokenRevokeURL              = GetRevokeTokenURL
-	tokenMCPBaseURL             = GetMCPBaseURL
-	tokenLogoutURL              = LogoutURL
-	tokenLogoutContinueURL      = LogoutContinueURL
-	tokenLogoutHTTPClient       = &http.Client{Timeout: 10 * time.Second, CheckRedirect: func(*http.Request, []*http.Request) error { return http.ErrUseLastResponse }}
-	tokenRevokeHTTPClient       = &http.Client{Timeout: 10 * time.Second}
+	tokenResolveDeletion         = resolveProfileDeletionSelection
+	tokenResolveSelection        = resolveProfileSelection
+	tokenUpsertProfile           = upsertProfileFromTokenWithCurrentLocked
+	tokenRemoveProfile           = removeProfileLocked
+	tokenSyncLegacyMirror        = syncLegacyTokenMirrorLocked
+	tokenSyncOrganizationMirror  = syncOrganizationTokenMirrorForProfile
+	tokenLoadProfiles            = LoadProfiles
+	tokenEnsureProfilesMigration = ensureProfilesMigrationLocked
+	tokenSaveProfiles            = SaveProfiles
+	tokenWriteMarker             = WriteTokenMarker
+	tokenWriteManualMarker       = WriteManualTokenMarker
+	tokenDeleteMarker            = DeleteTokenMarker
+	tokenParseURL                = url.Parse
+	tokenNewRequest              = http.NewRequestWithContext
+	tokenDefaultConfigDir        = getDefaultConfigDir
+	tokenLoadData                = LoadTokenData
+	tokenRevokeURL               = GetRevokeTokenURL
+	tokenMCPBaseURL              = GetMCPBaseURL
+	tokenLogoutURL               = LogoutURL
+	tokenLogoutContinueURL       = LogoutContinueURL
+	tokenLogoutHTTPClient        = &http.Client{Timeout: 10 * time.Second, CheckRedirect: func(*http.Request, []*http.Request) error { return http.ErrUseLastResponse }}
+	tokenRevokeHTTPClient        = &http.Client{Timeout: 10 * time.Second}
 )
 
 // TokenData holds the OAuth token set persisted to disk.
@@ -96,6 +97,96 @@ type TokenData struct {
 	ClientID       string    `json:"client_id,omitempty"` // Associated app client ID for refresh
 	UpdatedAt      string    `json:"updated_at,omitempty"`
 	Source         string    `json:"source,omitempty"`
+	// LegacyOrgScopedProfile is an in-memory destination for an explicitly
+	// matched historical profile whose userId was never resolved. It is never
+	// persisted as token material.
+	LegacyOrgScopedProfile string `json:"-"`
+	// FreshAuthorization distinguishes a new OAuth/device exchange from a
+	// refresh of the credential already selected locally. Persistence uses this
+	// transient marker to reject ambiguous UID-less logins without breaking
+	// legitimate refreshes of unresolved accounts.
+	FreshAuthorization bool `json:"-"`
+}
+
+// tokenPersistenceWritePlan is the single source of truth for deciding which
+// credential slots a token publication can touch. Both the write path and its
+// read-only preflight must use this plan so a slot cannot be newly written
+// without first being checked for unreadable data.
+//
+// The plan is intentionally pure: callers supply the already-loaded profile
+// registry and process-local runtime selector, and no storage is read or
+// mutated while the decision is made.
+type tokenPersistenceWritePlan struct {
+	CorpID                         string
+	UserID                         string
+	RuntimeSelector                string
+	PersistenceSelector            string
+	ExactSelector                  string
+	MakeCurrent                    bool
+	ExistingIdentity               bool
+	UpgradesLegacyProfile          bool
+	PreserveUnresolvedOrganization bool
+	WriteIdentity                  bool
+	WriteOrganization              bool
+	WriteGlobal                    bool
+}
+
+func planTokenPersistenceWrites(
+	cfg *ProfilesConfig,
+	data *TokenData,
+	runtimeSelector string,
+) tokenPersistenceWritePlan {
+	plan := tokenPersistenceWritePlan{
+		RuntimeSelector: strings.TrimSpace(runtimeSelector),
+		// Manual and organization-bound publications both snapshot the global
+		// compatibility slot before they can replace or resynchronize it.
+		WriteGlobal: true,
+	}
+	if data == nil {
+		return plan
+	}
+
+	plan.CorpID = strings.TrimSpace(data.CorpID)
+	plan.UserID = strings.TrimSpace(data.UserID)
+	if plan.CorpID == "" {
+		return plan
+	}
+
+	plan.PersistenceSelector = plan.RuntimeSelector
+	if plan.PersistenceSelector == "" && plan.UserID == "" {
+		plan.PersistenceSelector = strings.TrimSpace(data.LegacyOrgScopedProfile)
+	}
+	plan.MakeCurrent = plan.PersistenceSelector == ""
+	plan.ExactSelector = profileSelector(plan.CorpID, plan.UserID)
+	plan.ExistingIdentity = profileIndexByIdentity(cfg, plan.CorpID, plan.UserID) >= 0
+	plan.UpgradesLegacyProfile = !plan.ExistingIdentity &&
+		plan.UserID != "" &&
+		len(profilesForCorpID(cfg, plan.CorpID)) == 1 &&
+		legacyProfileIndexByCorpID(cfg, plan.CorpID) >= 0
+	plan.PreserveUnresolvedOrganization = plan.UserID != "" &&
+		unresolvedProfileForCorp(cfg, plan.CorpID) != nil &&
+		!plan.UpgradesLegacyProfile
+	plan.WriteIdentity = plan.UserID != ""
+	orgCurrentSelector := ""
+	if cfg != nil {
+		orgCurrentSelector = cfg.OrgCurrentProfiles[plan.CorpID]
+	}
+
+	// A sole unresolved profile is being completed in place during explicit
+	// reauthorization. Its organization slot must move with the newly exact
+	// identity even when an explicit runtime selector keeps it from becoming
+	// process-global current.
+	plan.WriteOrganization = plan.UserID == "" ||
+		plan.UpgradesLegacyProfile ||
+		(!plan.PreserveUnresolvedOrganization &&
+			(plan.MakeCurrent ||
+				exactProfileSelectorForCorp(
+					cfg,
+					plan.CorpID,
+					orgCurrentSelector,
+				) == plan.ExactSelector))
+
+	return plan
 }
 
 // IsAccessTokenValid returns true if the access token has not expired.
@@ -219,6 +310,30 @@ func SaveTokenData(configDir string, data *TokenData) error {
 	})
 }
 
+// SaveLoginTokenData is the safe persistence boundary for credentials produced
+// by a new login entry point (OAuth/device/PAT authorization or --token). It
+// repairs a uniquely recoverable half-migrated legacy login before any global
+// mirror can be replaced, marks the incoming credential as a fresh
+// authorization for UID-less account isolation, and preflights every slot the
+// write plan can touch.
+//
+// Refresh paths must continue to use SaveTokenData after their refresh-specific
+// preflight; treating a refresh as a fresh authorization would incorrectly
+// reject the selected unresolved account in a multi-account organization.
+func SaveLoginTokenData(configDir string, data *TokenData) error {
+	if data == nil {
+		return fmt.Errorf("token data is empty")
+	}
+	if err := prepareLoginPersistence(configDir); err != nil {
+		return fmt.Errorf("local login state cannot be safely updated: %w", err)
+	}
+	data.FreshAuthorization = true
+	if err := preflightTokenWritePersistence(configDir, data); err != nil {
+		return fmt.Errorf("local login state cannot be safely updated: %w", err)
+	}
+	return SaveTokenData(configDir, data)
+}
+
 // saveTokenDataLocked performs the keychain + profiles.json + legacy mirror
 // writes assuming the auth dual-layer lock is already held. Callers that
 // already hold the lock (OAuthProvider refresh path, the legacy secure->keychain
@@ -235,35 +350,56 @@ func saveTokenDataLocked(configDir string, data *TokenData) error {
 		if err != nil {
 			return err
 		}
+		// A login may be the first operation after upgrading. Finish a v1
+		// registry migration before an upsert can raise profiles.json to v2;
+		// otherwise untouched organizations would permanently lose their chance
+		// to receive exact identity token slots. Do not re-run v2 repair here:
+		// refresh has already rotated the remote credential at this point, and an
+		// unrelated damaged identity slot must not prevent the new token from
+		// being committed. Normal load/preflight paths repair v2 before exchange.
+		if cfg.Version < profilesVersion {
+			if err := tokenEnsureProfilesMigration(configDir); err != nil {
+				return err
+			}
+			cfg, err = tokenLoadProfiles(configDir)
+			if err != nil {
+				return err
+			}
+		}
 		if err := ensureProfilesWritable(cfg); err != nil {
 			return err
 		}
-		runtimeSelector := strings.TrimSpace(RuntimeProfile())
-		makeCurrent := runtimeSelector == ""
-		exactSelector := profileSelector(corpID, userID)
-		mirrorOrg := makeCurrent ||
-			exactProfileSelectorForCorp(cfg, corpID, cfg.OrgCurrentProfiles[corpID]) == exactSelector
-		existingIdentity := profileIndexByIdentity(cfg, corpID, userID) >= 0
-		upgradesLegacyProfile := !existingIdentity && userID != "" && legacyProfileIndexByCorpID(cfg, corpID) >= 0
+		plan := planTokenPersistenceWrites(cfg, data, RuntimeProfile())
+		if err := validateTokenPersistenceWritePlan(cfg, data, plan); err != nil {
+			return err
+		}
 		logging.AuthDebug(
 			"auth.token.persist.plan",
 			"corp_id", corpID,
 			"user_id", userID,
 			"user_name", strings.TrimSpace(data.UserName),
-			"identity_selector", exactSelector,
-			"existing_identity", existingIdentity,
-			"upgrades_legacy_profile", upgradesLegacyProfile,
+			"identity_selector", plan.ExactSelector,
+			"existing_identity", plan.ExistingIdentity,
+			"upgrades_legacy_profile", plan.UpgradesLegacyProfile,
 			"profiles_before", len(cfg.Profiles),
-			"runtime_profile", runtimeSelector,
-			"write_identity_slot", userID != "",
-			"write_org_mirror", mirrorOrg,
-			"write_global_mirror", makeCurrent,
+			"runtime_profile", plan.RuntimeSelector,
+			"persistence_profile", plan.PersistenceSelector,
+			"write_identity_slot", plan.WriteIdentity,
+			"write_org_mirror", plan.WriteOrganization,
+			"write_global_mirror", plan.WriteGlobal,
+			"publish_incoming_global", plan.MakeCurrent,
+		)
+		snapshot, err := snapshotTokenPersistence(
+			configDir,
+			cfg,
+			plan.CorpID,
+			plan.UserID,
+			plan.WriteOrganization,
 		)
-		snapshot, err := snapshotTokenPersistence(configDir, cfg, corpID, userID, mirrorOrg)
 		if err != nil {
 			return err
 		}
-		preserveManualDefault := !makeCurrent &&
+		preserveManualDefault := !plan.MakeCurrent &&
 			snapshot.marker.known &&
 			snapshot.marker.exists &&
 			snapshot.marker.manual
@@ -273,32 +409,28 @@ func saveTokenDataLocked(configDir string, data *TokenData) error {
 			}
 			return operationErr
 		}
-		if userID != "" {
+		if plan.WriteIdentity {
 			if err := tokenSaveKeychainForIdentity(corpID, userID, data); err != nil {
 				return rollback(err)
 			}
-		} else {
-			for _, profile := range cfg.Profiles {
-				if strings.TrimSpace(profile.CorpID) == corpID && strings.TrimSpace(profile.UserID) != "" {
-					return fmt.Errorf("cannot store profile for corpId %q without userId because account identities already exist", corpID)
-				}
-			}
 		}
-		if err := tokenUpsertProfile(configDir, data, makeCurrent); err != nil {
+		if err := tokenUpsertProfile(configDir, data, plan.MakeCurrent); err != nil {
 			return rollback(err)
 		}
-		if mirrorOrg {
+		if plan.WriteOrganization {
 			if err := tokenSaveKeychainForCorpID(corpID, data); err != nil {
 				return rollback(err)
 			}
 		}
-		if makeCurrent {
-			if err := tokenSaveKeychain(data); err != nil {
-				return rollback(err)
-			}
-		} else if !preserveManualDefault {
-			if err := tokenSyncLegacyMirror(configDir); err != nil {
-				return rollback(err)
+		if plan.WriteGlobal {
+			if plan.MakeCurrent {
+				if err := tokenSaveKeychain(data); err != nil {
+					return rollback(err)
+				}
+			} else if !preserveManualDefault {
+				if err := tokenSyncLegacyMirror(configDir); err != nil {
+					return rollback(err)
+				}
 			}
 		}
 		if preserveManualDefault {
@@ -313,10 +445,11 @@ func saveTokenDataLocked(configDir string, data *TokenData) error {
 			"corp_id", corpID,
 			"user_id", userID,
 			"user_name", strings.TrimSpace(data.UserName),
-			"identity_selector", exactSelector,
-			"write_identity_slot", userID != "",
-			"write_org_mirror", mirrorOrg,
-			"write_global_mirror", makeCurrent,
+			"identity_selector", plan.ExactSelector,
+			"write_identity_slot", plan.WriteIdentity,
+			"write_org_mirror", plan.WriteOrganization,
+			"write_global_mirror", plan.WriteGlobal && !preserveManualDefault,
+			"publish_incoming_global", plan.MakeCurrent,
 		)
 		return nil
 	}
@@ -428,7 +561,7 @@ func loadTokenDataForProfileLocked(configDir, profile string) (*TokenData, error
 		// as a different organization (the legacy mirror may have drifted).
 		if legacy, lerr := tokenLoadKeychain(); lerr == nil && legacy != nil &&
 			strings.TrimSpace(legacy.CorpID) == strings.TrimSpace(selected.CorpID) &&
-			(strings.TrimSpace(selected.UserID) == "" || strings.TrimSpace(legacy.UserID) == strings.TrimSpace(selected.UserID)) {
+			strings.TrimSpace(legacy.UserID) == strings.TrimSpace(selected.UserID) {
 			return legacy, nil
 		} else if lerr != nil && !errors.Is(lerr, ErrTokenDataNotFound) {
 			return nil, lerr
@@ -458,35 +591,129 @@ func loadTokenDataForProfileLocked(configDir, profile string) (*TokenData, error
 }
 
 func tokenLoadProfileIdentity(profile Profile) (*TokenData, error) {
+	corpID := strings.TrimSpace(profile.CorpID)
+	userID := strings.TrimSpace(profile.UserID)
 	if strings.TrimSpace(profile.UserID) == "" {
-		return tokenLoadKeychainForCorpID(profile.CorpID)
+		data, err := tokenLoadKeychainForCorpID(corpID)
+		if err != nil {
+			return nil, err
+		}
+		if data == nil {
+			return nil, ErrTokenDataNotFound
+		}
+		if strings.TrimSpace(data.CorpID) != corpID {
+			return nil, fmt.Errorf(
+				"organization token mirror for corpId %q contains token for corpId %q; cannot use it for unresolved profile %q",
+				corpID,
+				data.CorpID,
+				profile.Name,
+			)
+		}
+		if strings.TrimSpace(data.UserID) != "" {
+			return nil, fmt.Errorf(
+				"organization token mirror for corpId %q belongs to userId %q; cannot use it for unresolved profile %q",
+				corpID,
+				data.UserID,
+				profile.Name,
+			)
+		}
+		return data, nil
 	}
-	data, err := tokenLoadKeychainIdentity(profile.CorpID, profile.UserID)
+	data, err := tokenLoadKeychainIdentity(corpID, userID)
 	if err == nil {
+		if data == nil {
+			return nil, ErrTokenDataNotFound
+		}
+		if strings.TrimSpace(data.CorpID) != corpID || strings.TrimSpace(data.UserID) != userID {
+			return nil, fmt.Errorf(
+				"identity token slot %q contains token for %q; cannot use it for profile %q",
+				TokenAccountForIdentity(corpID, userID),
+				profileSelector(data.CorpID, data.UserID),
+				ProfileSelector(profile),
+			)
+		}
 		return data, nil
 	}
 	if !errors.Is(err, ErrTokenDataNotFound) {
 		return nil, err
 	}
-	orgData, orgErr := tokenLoadKeychainForCorpID(profile.CorpID)
+	orgData, orgErr := tokenLoadKeychainForCorpID(corpID)
 	if orgErr != nil {
 		if errors.Is(orgErr, ErrTokenDataNotFound) {
 			return nil, err
 		}
 		return nil, orgErr
 	}
-	if strings.TrimSpace(orgData.UserID) == "" {
-		return nil, fmt.Errorf("organization token mirror for corpId %q has no userId; cannot use it for profile %q", profile.CorpID, ProfileSelector(profile))
-	}
-	if strings.TrimSpace(orgData.UserID) != strings.TrimSpace(profile.UserID) {
+	if orgData == nil {
 		return nil, err
 	}
-	if saveErr := tokenSaveKeychainForIdentity(profile.CorpID, profile.UserID, orgData); saveErr != nil {
+	if strings.TrimSpace(orgData.CorpID) != corpID {
+		return nil, fmt.Errorf(
+			"organization token mirror for corpId %q contains token for corpId %q; cannot use it for profile %q",
+			corpID,
+			orgData.CorpID,
+			ProfileSelector(profile),
+		)
+	}
+	if strings.TrimSpace(orgData.UserID) == "" {
+		return nil, fmt.Errorf("organization token mirror for corpId %q has no userId; cannot use it for profile %q", corpID, ProfileSelector(profile))
+	}
+	if strings.TrimSpace(orgData.UserID) != userID {
+		return nil, err
+	}
+	if saveErr := tokenSaveKeychainForIdentity(corpID, userID, orgData); saveErr != nil {
 		return nil, saveErr
 	}
 	return orgData, nil
 }
 
+// validateTokenPersistenceWritePlan prevents a freshly authorized UID-less
+// credential from being attached to an existing unresolved sibling merely
+// because both accounts share a corpId. An explicit selector is a storage
+// boundary, but it is not proof that an exact account completed OAuth. The only
+// safe overwrite is when that selector itself resolves to the existing
+// unresolved profile. A blank selector remains allowed for ordinary refreshes
+// of the already-selected unresolved token.
+func validateTokenPersistenceWritePlan(
+	cfg *ProfilesConfig,
+	data *TokenData,
+	plan tokenPersistenceWritePlan,
+) error {
+	if data == nil || plan.CorpID == "" || plan.UserID != "" {
+		return nil
+	}
+	unresolved := unresolvedProfileForCorp(cfg, plan.CorpID)
+	if unresolved == nil {
+		return nil
+	}
+	targetSelector := plan.RuntimeSelector
+	if targetSelector == "" {
+		targetSelector = strings.TrimSpace(data.LegacyOrgScopedProfile)
+	}
+	if targetSelector == "" {
+		if data.FreshAuthorization && len(profilesForCorpID(cfg, plan.CorpID)) > 1 {
+			return fmt.Errorf(
+				"refusing to save a fresh UID-less token over existing unresolved profile %q in multi-account organization %q; retry with that unresolved profile selector or require server-provided userId",
+				storedProfileSelector(cfg, unresolved),
+				plan.CorpID,
+			)
+		}
+		return nil
+	}
+	selected, _, err := resolveProfileSelection("", cfg, targetSelector)
+	if err == nil && selected != nil &&
+		strings.TrimSpace(selected.CorpID) == plan.CorpID &&
+		strings.TrimSpace(selected.UserID) == "" {
+		return nil
+	}
+	return fmt.Errorf(
+		"refusing to save UID-less token selected as profile %q over existing unresolved profile %q in organization %q; retry with the unresolved profile selector or require server-provided userId",
+		targetSelector,
+		storedProfileSelector(cfg, unresolved),
+		plan.CorpID,
+	)
+}
+
 // DeleteTokenData removes token data. Edition hooks and the default keychain
 // path are both serialized with refresh through the auth dual lock.
 func DeleteTokenData(configDir string) error {
@@ -561,12 +788,20 @@ func deleteTokenDataForProfileLocked(configDir, profile string) error {
 		removeSelector := removed.CorpID
 		orgCurrent := false
 		if exact {
-			removeSelector = ProfileSelector(removed)
-			orgCurrent = exactProfileSelectorForCorp(
-				cfg,
-				removed.CorpID,
-				cfg.OrgCurrentProfiles[removed.CorpID],
-			) == ProfileSelector(removed)
+			if strings.TrimSpace(removed.UserID) == "" {
+				// A blank profile is exact only when it was selected by its unique
+				// local name. Converting it back to corpId here would turn a
+				// one-profile logout into whole-organization deletion.
+				removeSelector = effectiveSelector
+				orgCurrent = true
+			} else {
+				removeSelector = ProfileSelector(removed)
+				orgCurrent = exactProfileSelectorForCorp(
+					cfg,
+					removed.CorpID,
+					cfg.OrgCurrentProfiles[removed.CorpID],
+				) == ProfileSelector(removed)
+			}
 		}
 		if _, err := tokenRemoveProfile(configDir, removeSelector); err != nil {
 			return err
@@ -592,7 +827,8 @@ func deleteTokenDataForProfileLocked(configDir, profile string) error {
 				return rollback(loadErr)
 			}
 			replacementSelector := updated.OrgCurrentProfiles[removed.CorpID]
-			if exact && replacementSelector != "" {
+			preserveUnresolvedOrg := unresolvedProfileForCorp(updated, removed.CorpID) != nil
+			if exact && replacementSelector != "" && !preserveUnresolvedOrg {
 				replacement, _, resolveErr := tokenResolveSelection(configDir, updated, replacementSelector)
 				if resolveErr != nil {
 					return rollback(resolveErr)
@@ -600,8 +836,10 @@ func deleteTokenDataForProfileLocked(configDir, profile string) error {
 				if err := tokenSyncOrganizationMirror(*replacement); err != nil {
 					return rollback(err)
 				}
-			} else if err := tokenDeleteKeychainForCorpID(removed.CorpID); err != nil {
-				return rollback(err)
+			} else if !preserveUnresolvedOrg {
+				if err := tokenDeleteKeychainForCorpID(removed.CorpID); err != nil {
+					return rollback(err)
+				}
 			}
 		}
 		preserveManualDefault := markerSnapshot.known &&

internal/auth/token_preflight_test.go

@@ -21,6 +21,7 @@ import (
 	"encoding/json"
 	"errors"
 	"io"
+	"net"
 	"net/http"
 	"net/http/httptest"
 	"os"
@@ -210,10 +211,48 @@ func TestExactNonOrgCurrentRefreshIgnoresUnreadableOrgMirror(t *testing.T) {
 	}
 }
 
-func TestExchangeAuthCodePreflightsOrphanProfileCiphertextBeforeHTTP(t *testing.T) {
+func TestCrossPlatformCoverageExactRefreshAndSwitchIgnoreUnreadableReservedBlankOrgSlot(t *testing.T) {
+	t.Setenv(keychain.DisableKeychainEnv, "1")
+	fixture := seedLegacyBlankAndExactIdentitySlots(t)
+	if err := os.WriteFile(profileCiphertextPathForTest(fixture.corpID), []byte("corrupt reserved blank slot"), 0o600); err != nil {
+		t.Fatalf("WriteFile(reserved blank ciphertext) error = %v", err)
+	}
+
+	SetRuntimeProfile("")
+	if err := preflightTokenRefreshPersistence(fixture.configDir, fixture.beta); err != nil {
+		t.Fatalf("preflightTokenRefreshPersistence(current exact with reserved blank) error = %v", err)
+	}
+	refreshed := *fixture.beta
+	refreshed.AccessToken = "at_identity_beta_refreshed"
+	if err := SaveTokenData(fixture.configDir, &refreshed); err != nil {
+		t.Fatalf("SaveTokenData(current exact with reserved blank) error = %v", err)
+	}
+	if raw, err := os.ReadFile(profileCiphertextPathForTest(fixture.corpID)); err != nil || string(raw) != "corrupt reserved blank slot" {
+		t.Fatalf("reserved blank slot changed during exact refresh: %q, %v", raw, err)
+	}
+
+	if selected, err := SetCurrentProfile(fixture.configDir, profileSelector(fixture.alpha.CorpID, fixture.alpha.UserID)); err != nil || selected.UserID != fixture.alpha.UserID {
+		t.Fatalf("SetCurrentProfile(exact with reserved blank) = %#v, %v", selected, err)
+	}
+	if selected, err := UsePreviousProfile(fixture.configDir); err != nil || selected.UserID != fixture.beta.UserID {
+		t.Fatalf("UsePreviousProfile(exact with reserved blank) = %#v, %v", selected, err)
+	}
+	if raw, err := os.ReadFile(profileCiphertextPathForTest(fixture.corpID)); err != nil || string(raw) != "corrupt reserved blank slot" {
+		t.Fatalf("reserved blank slot changed during exact switches: %q, %v", raw, err)
+	}
+
+	blankRefresh := *fixture.blank
+	blankRefresh.LegacyOrgScopedProfile = fixture.blankName
+	SetRuntimeProfile(fixture.blankName)
+	if err := preflightTokenRefreshPersistence(fixture.configDir, &blankRefresh); err == nil ||
+		!strings.Contains(err.Error(), "profile token slot") {
+		t.Fatalf("blank refresh preflight error = %v, want unreadable reserved slot", err)
+	}
+}
+
+func TestFullTokenPersistenceInventoryDetectsOrphanProfileCiphertext(t *testing.T) {
 	cleanupKeychain(t)
 	t.Setenv(keychain.DisableKeychainEnv, "1")
-	setPreflightTestCredentials(t)
 	configDir := t.TempDir()
 	data := testToken("at_orphan", "corp_orphan", "Orphan Org")
 
@@ -230,21 +269,12 @@ func TestExchangeAuthCodePreflightsOrphanProfileCiphertextBeforeHTTP(t *testing.
 		t.Fatalf("WriteFile(replacement DEK) error = %v", err)
 	}
 
-	var calls atomic.Int32
-	provider := NewOAuthProvider(configDir, nil)
-	provider.httpClient = &http.Client{Transport: preflightRoundTripFunc(func(*http.Request) (*http.Response, error) {
-		calls.Add(1)
-		return nil, errors.New("unexpected HTTP request")
-	})}
-	_, err := provider.ExchangeAuthCode(context.Background(), "auth-code", "")
+	err := preflightTokenPersistence(configDir)
 	if err == nil || !strings.Contains(err.Error(), "auth token ciphertext inventory") {
-		t.Fatalf("ExchangeAuthCode() error = %v, want orphan ciphertext preflight error", err)
+		t.Fatalf("preflightTokenPersistence() error = %v, want orphan ciphertext inventory error", err)
 	}
 	if !keychain.IsCiphertextKeyMismatch(err) {
-		t.Fatalf("ExchangeAuthCode() error = %v, want ciphertext key mismatch in error chain", err)
-	}
-	if got := calls.Load(); got != 0 {
-		t.Fatalf("HTTP calls = %d, want 0", got)
+		t.Fatalf("preflightTokenPersistence() error = %v, want ciphertext key mismatch in error chain", err)
 	}
 }
 
@@ -304,7 +334,7 @@ func TestRefreshPreflightIgnoresUnreadableUnrelatedProfile(t *testing.T) {
 	}
 }
 
-func TestOAuthLoginPreflightsTokenPersistence(t *testing.T) {
+func TestOAuthLoginUnreadableGlobalFailsClosedBeforeAuthorizationStart(t *testing.T) {
 	setPreflightTestCredentials(t)
 	for _, force := range []bool{false, true} {
 		t.Run("force="+map[bool]string{false: "false", true: "true"}[force], func(t *testing.T) {
@@ -312,33 +342,60 @@ func TestOAuthLoginPreflightsTokenPersistence(t *testing.T) {
 			configDir := t.TempDir()
 			seedUnreadableTokenStorage(t, configDir, testToken("at_login", "corp_login", "Login Org"))
 
-			ctx, cancel := context.WithCancel(context.Background())
-			cancel()
+			listenErr := errors.New("authorization listener reached")
+			var calls atomic.Int32
+			oldListen := oauthListen
+			oauthListen = func(string, string) (net.Listener, error) {
+				calls.Add(1)
+				return nil, listenErr
+			}
+			t.Cleanup(func() { oauthListen = oldListen })
+
 			provider := NewOAuthProvider(configDir, nil)
 			provider.NoBrowser = true
-			_, err := provider.Login(ctx, force)
-			if err == nil || !strings.Contains(err.Error(), "legacy token slot") {
-				t.Fatalf("Login(force=%v) error = %v, want token persistence preflight error", force, err)
+			_, err := provider.Login(context.Background(), force)
+			if err == nil || !strings.Contains(err.Error(), "refusing to overwrite") {
+				t.Fatalf("Login(force=%v) error = %v, want unreadable-global protection", force, err)
+			}
+			if errors.Is(err, listenErr) {
+				t.Fatalf("Login(force=%v) reached authorization listener: %v", force, err)
+			}
+			if got := calls.Load(); got != 0 {
+				t.Fatalf("Login(force=%v) listener calls = %d, want 0", force, got)
 			}
 		})
 	}
 }
 
-func TestExchangeAuthCodePreflightsBeforeHTTP(t *testing.T) {
+func TestExchangeAuthCodeRejectsUnreadableGlobalBeforeHTTP(t *testing.T) {
 	cleanupKeychain(t)
 	setPreflightTestCredentials(t)
 	configDir := t.TempDir()
-	seedUnreadableTokenStorage(t, configDir, testToken("at_exchange", "corp_exchange", "Exchange Org"))
+	existing := testToken("at_exchange", "corp_exchange", "Exchange Org")
+	seedUnreadableTokenStorage(t, configDir, existing)
 
 	var calls atomic.Int32
+	var saveCalls atomic.Int32
+	oldSave := oauthSaveToken
+	oauthSaveToken = func(string, *TokenData) error {
+		saveCalls.Add(1)
+		return nil
+	}
+	t.Cleanup(func() { oauthSaveToken = oldSave })
 	provider := NewOAuthProvider(configDir, nil)
 	provider.httpClient = &http.Client{Transport: preflightRoundTripFunc(func(*http.Request) (*http.Response, error) {
 		calls.Add(1)
-		return nil, errors.New("unexpected HTTP request")
+		return &http.Response{
+			StatusCode: http.StatusOK,
+			Header:     make(http.Header),
+			Body: io.NopCloser(strings.NewReader(
+				`{"accessToken":"new-access","refreshToken":"new-refresh","expiresIn":7200,"corpId":"corp_exchange"}`,
+			)),
+		}, nil
 	})}
-	_, err := provider.ExchangeAuthCode(context.Background(), "auth-code", "")
+	_, err := provider.ExchangeAuthCode(context.Background(), "auth-code", existing.UserID)
 	if err == nil || !strings.Contains(err.Error(), "legacy token slot") {
-		t.Fatalf("ExchangeAuthCode() error = %v, want token persistence preflight error", err)
+		t.Fatalf("ExchangeAuthCode() error = %v, want target token persistence error", err)
 	}
 	if !keychain.IsCiphertextKeyMismatch(err) {
 		t.Fatalf("ExchangeAuthCode() error = %v, want ciphertext key mismatch in error chain", err)
@@ -346,9 +403,12 @@ func TestExchangeAuthCodePreflightsBeforeHTTP(t *testing.T) {
 	if got := calls.Load(); got != 0 {
 		t.Fatalf("HTTP calls = %d, want 0", got)
 	}
+	if got := saveCalls.Load(); got != 0 {
+		t.Fatalf("SaveTokenData calls = %d, want 0", got)
+	}
 }
 
-func TestDeviceFlowLoginPreflightsBeforeDeviceCodeRequest(t *testing.T) {
+func TestDeviceFlowLoginRejectsUnreadableGlobalBeforeDeviceCodeRequest(t *testing.T) {
 	cleanupKeychain(t)
 	setPreflightTestCredentials(t)
 	configDir := t.TempDir()
@@ -366,7 +426,7 @@ func TestDeviceFlowLoginPreflightsBeforeDeviceCodeRequest(t *testing.T) {
 	provider.SetBaseURL(server.URL)
 	_, err := provider.Login(context.Background())
 	if err == nil || !strings.Contains(err.Error(), "legacy token slot") {
-		t.Fatalf("DeviceFlowProvider.Login() error = %v, want token persistence preflight error", err)
+		t.Fatalf("DeviceFlowProvider.Login() error = %v, want unreadable-global protection", err)
 	}
 	if got := calls.Load(); got != 0 {
 		t.Fatalf("device code requests = %d, want 0", got)
@@ -418,7 +478,7 @@ func TestLockedRefreshRejectsFutureProfilesVersionBeforeHTTP(t *testing.T) {
 	if err != nil {
 		t.Fatalf("LoadProfiles() error = %v", err)
 	}
-	cfg.Version = profilesVersion + 1
+	cfg.Version = profilesMaxVersion + 1
 	raw, err := json.Marshal(cfg)
 	if err != nil {
 		t.Fatalf("json.Marshal() error = %v", err)

internal/cli/schema_agent_metadata/index.json

@@ -1,18 +1,18 @@
 {
   "version": 1,
-  "source_hash": "sha256:5df496973c41b3b4f7ae8c856ff0e9e99bcabd4455419b7d41bb23c44df6f1af",
-  "surface_hash": "sha256:7ef588f38052f0104e027c8daff5fefd68698781f2c87715461183d4058288ed",
+  "source_hash": "sha256:f7d9911baa7c829febb0f461c1ae773d9bd4747b73a79c6fc4cd8a0e3956561c",
+  "surface_hash": "sha256:4cf8460240b19f896c3a330c69982cbb5f57aa8576cdf30373172082eea893ed",
   "coverage": {
-    "surface_products": 22,
-    "products_with_metadata": 22,
-    "surface_tools": 572,
-    "tools_with_metadata": 572,
-    "tools_with_agent_summary": 572,
-    "tools_with_use_when": 572,
-    "tools_with_avoid_when": 572,
-    "tools_with_examples": 572,
-    "tools_with_interface_mode": 572,
-    "unmatched_skill_tools": 120,
+    "surface_products": 26,
+    "products_with_metadata": 26,
+    "surface_tools": 813,
+    "tools_with_metadata": 813,
+    "tools_with_agent_summary": 813,
+    "tools_with_use_when": 813,
+    "tools_with_avoid_when": 813,
+    "tools_with_examples": 813,
+    "tools_with_interface_mode": 813,
+    "unmatched_skill_tools": 118,
     "unreviewed_skill_tools": 7
   },
   "products": {
@@ -444,20 +444,20 @@
       ]
     },
     "contact": {
-      "agent_summary": "查询通讯录与花名册，并创建企业、企业账号或邀请员工",
+      "agent_summary": "查询通讯录与花名册，并管理企业、部门、员工及企业账号",
       "agent_summary_source": "dws-agent-selection/contact",
       "avoid_when": [
         "职责/上级等语义找人优先 aisearch person；不要用 contact 发消息；写操作前确认当前企业和目标信息"
       ],
       "field_provenance": {
         "agent_summary": {
-          "value": "查询通讯录与花名册，并创建企业、企业账号或邀请员工",
+          "value": "查询通讯录与花名册，并管理企业、部门、员工及企业账号",
           "source": "internal/cli/schema_hints/selection/contact.json",
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
           "candidates": [
             {
-              "value": "查询通讯录与花名册，并创建企业、企业账号或邀请员工",
+              "value": "查询通讯录与花名册，并管理企业、部门、员工及企业账号",
               "source": "internal/cli/schema_hints/selection/contact.json",
               "precedence": "reviewed_explicit",
               "selected": true
@@ -585,6 +585,74 @@
         "创建/配置开放平台应用、机器人、权限、事件订阅或发布版本"
       ]
     },
+    "devapp": {
+      "agent_summary": "管理钉钉开放平台企业内部应用、成员、权限、机器人、事件与版本",
+      "agent_summary_source": "dws-agent-selection/devapp",
+      "avoid_when": [
+        "开放平台接口文档搜索使用 devdoc；普通钉钉业务数据使用对应产品命令"
+      ],
+      "field_provenance": {
+        "agent_summary": {
+          "value": "管理钉钉开放平台企业内部应用、成员、权限、机器人、事件与版本",
+          "source": "internal/cli/schema_hints/selection/devapp.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "candidates": [
+            {
+              "value": "管理钉钉开放平台企业内部应用、成员、权限、机器人、事件与版本",
+              "source": "internal/cli/schema_hints/selection/devapp.json",
+              "precedence": "reviewed_explicit",
+              "selected": true
+            }
+          ]
+        },
+        "avoid_when": {
+          "value": [
+            "开放平台接口文档搜索使用 devdoc；普通钉钉业务数据使用对应产品命令"
+          ],
+          "source": "internal/cli/schema_hints/selection/devapp.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "candidates": [
+            {
+              "value": [
+                "开放平台接口文档搜索使用 devdoc；普通钉钉业务数据使用对应产品命令"
+              ],
+              "source": "internal/cli/schema_hints/selection/devapp.json",
+              "precedence": "reviewed_explicit",
+              "selected": true
+            }
+          ]
+        },
+        "use_when": {
+          "value": [
+            "请求涉及企业内部应用的查询、创建、配置、成员权限、机器人、事件订阅或版本管理"
+          ],
+          "source": "internal/cli/schema_hints/selection/devapp.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "candidates": [
+            {
+              "value": [
+                "请求涉及企业内部应用的查询、创建、配置、成员权限、机器人、事件订阅或版本管理"
+              ],
+              "source": "internal/cli/schema_hints/selection/devapp.json",
+              "precedence": "reviewed_explicit",
+              "selected": true
+            }
+          ]
+        }
+      },
+      "source_refs": [
+        "CommandRegistry:product=devapp",
+        "internal/cli/schema_command_registry.json",
+        "internal/cli/schema_hints/selection/devapp.json",
+        "internal/shortcut/devapp"
+      ],
+      "use_when": [
+        "请求涉及企业内部应用的查询、创建、配置、成员权限、机器人、事件订阅或版本管理"
+      ]
+    },
     "devdoc": {
       "agent_summary": "搜索钉钉开放平台开发文档与错误排查资料",
       "agent_summary_source": "dws-agent-selection/devdoc",
@@ -941,6 +1009,78 @@
         "需要实时监听个人消息接收、已读、撤回或表情回应事件，或管理个人事件订阅生命周期"
       ]
     },
+    "hrbrain": {
+      "agent_summary": "钉钉组织大脑：人才池管理、员工档案查询与人才搜索",
+      "agent_summary_source": "dws-agent-selection/hrbrain",
+      "avoid_when": [
+        "要操作通讯录/组织架构基础信息时改用 contact 产品",
+        "要提交/查询战略解码、经营合约、目标等 OKR 能力时改用 agoal（CLI-only，未接入 Agent Schema）"
+      ],
+      "field_provenance": {
+        "agent_summary": {
+          "value": "钉钉组织大脑：人才池管理、员工档案查询与人才搜索",
+          "source": "internal/cli/schema_hints/selection/hrbrain.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "candidates": [
+            {
+              "value": "钉钉组织大脑：人才池管理、员工档案查询与人才搜索",
+              "source": "internal/cli/schema_hints/selection/hrbrain.json",
+              "precedence": "reviewed_explicit",
+              "selected": true
+            }
+          ]
+        },
+        "avoid_when": {
+          "value": [
+            "要操作通讯录/组织架构基础信息时改用 contact 产品",
+            "要提交/查询战略解码、经营合约、目标等 OKR 能力时改用 agoal（CLI-only，未接入 Agent Schema）"
+          ],
+          "source": "internal/cli/schema_hints/selection/hrbrain.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "candidates": [
+            {
+              "value": [
+                "要操作通讯录/组织架构基础信息时改用 contact 产品",
+                "要提交/查询战略解码、经营合约、目标等 OKR 能力时改用 agoal（CLI-only，未接入 Agent Schema）"
+              ],
+              "source": "internal/cli/schema_hints/selection/hrbrain.json",
+              "precedence": "reviewed_explicit",
+              "selected": true
+            }
+          ]
+        },
+        "use_when": {
+          "value": [
+            "需要查询人才池、员工档案（元数据/标签/职业历程/绩效）或搜索员工时"
+          ],
+          "source": "internal/cli/schema_hints/selection/hrbrain.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "candidates": [
+            {
+              "value": [
+                "需要查询人才池、员工档案（元数据/标签/职业历程/绩效）或搜索员工时"
+              ],
+              "source": "internal/cli/schema_hints/selection/hrbrain.json",
+              "precedence": "reviewed_explicit",
+              "selected": true
+            }
+          ]
+        }
+      },
+      "source_refs": [
+        "CommandRegistry:product=hrbrain",
+        "Skill:skills/mono/references/products/hrbrain.md",
+        "Skill:skills/multi/dingtalk-hrbrain/SKILL.md",
+        "internal/cli/schema_hints/selection/hrbrain.json",
+        "skills/mono/SKILL.md"
+      ],
+      "use_when": [
+        "需要查询人才池、员工档案（元数据/标签/职业历程/绩效）或搜索员工时"
+      ]
+    },
     "live": {
       "agent_summary": "查询当前用户发起的直播列表",
       "agent_summary_source": "dws-agent-selection/live",
@@ -1089,6 +1229,142 @@
         "查收、搜索、阅读、回复、发送或整理邮件"
       ]
     },
+    "markdown": {
+      "agent_summary": "跨钉盘与文档空间创建、获取、覆盖和局部修补原生 Markdown 文件",
+      "agent_summary_source": "dws-agent-selection/markdown",
+      "avoid_when": [
+        "在线文档正文操作使用 doc；普通二进制文件上传下载使用 drive"
+      ],
+      "field_provenance": {
+        "agent_summary": {
+          "value": "跨钉盘与文档空间创建、获取、覆盖和局部修补原生 Markdown 文件",
+          "source": "internal/cli/schema_hints/selection/markdown.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "candidates": [
+            {
+              "value": "跨钉盘与文档空间创建、获取、覆盖和局部修补原生 Markdown 文件",
+              "source": "internal/cli/schema_hints/selection/markdown.json",
+              "precedence": "reviewed_explicit",
+              "selected": true
+            }
+          ]
+        },
+        "avoid_when": {
+          "value": [
+            "在线文档正文操作使用 doc；普通二进制文件上传下载使用 drive"
+          ],
+          "source": "internal/cli/schema_hints/selection/markdown.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "candidates": [
+            {
+              "value": [
+                "在线文档正文操作使用 doc；普通二进制文件上传下载使用 drive"
+              ],
+              "source": "internal/cli/schema_hints/selection/markdown.json",
+              "precedence": "reviewed_explicit",
+              "selected": true
+            }
+          ]
+        },
+        "use_when": {
+          "value": [
+            "目标是原生 .md 文件，并需要在 Drive/Doc 路由间安全处理内容时"
+          ],
+          "source": "internal/cli/schema_hints/selection/markdown.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "candidates": [
+            {
+              "value": [
+                "目标是原生 .md 文件，并需要在 Drive/Doc 路由间安全处理内容时"
+              ],
+              "source": "internal/cli/schema_hints/selection/markdown.json",
+              "precedence": "reviewed_explicit",
+              "selected": true
+            }
+          ]
+        }
+      },
+      "source_refs": [
+        "CommandRegistry:product=markdown",
+        "Wukong-parity:3306c3307",
+        "internal/cli/schema_hints/selection/markdown.json",
+        "skills/mono/SKILL.md"
+      ],
+      "use_when": [
+        "目标是原生 .md 文件，并需要在 Drive/Doc 路由间安全处理内容时"
+      ]
+    },
+    "mcp": {
+      "agent_summary": "解析和管理当前身份可用的 MCP 服务连接信息",
+      "agent_summary_source": "dws-agent-selection/mcp",
+      "avoid_when": [
+        "查询普通钉钉业务数据时使用对应产品命令，不要使用 mcp"
+      ],
+      "field_provenance": {
+        "agent_summary": {
+          "value": "解析和管理当前身份可用的 MCP 服务连接信息",
+          "source": "internal/cli/schema_hints/selection/mcp.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "candidates": [
+            {
+              "value": "解析和管理当前身份可用的 MCP 服务连接信息",
+              "source": "internal/cli/schema_hints/selection/mcp.json",
+              "precedence": "reviewed_explicit",
+              "selected": true
+            }
+          ]
+        },
+        "avoid_when": {
+          "value": [
+            "查询普通钉钉业务数据时使用对应产品命令，不要使用 mcp"
+          ],
+          "source": "internal/cli/schema_hints/selection/mcp.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "candidates": [
+            {
+              "value": [
+                "查询普通钉钉业务数据时使用对应产品命令，不要使用 mcp"
+              ],
+              "source": "internal/cli/schema_hints/selection/mcp.json",
+              "precedence": "reviewed_explicit",
+              "selected": true
+            }
+          ]
+        },
+        "use_when": {
+          "value": [
+            "需要把钉钉 MCP 市场中的服务连接到支持 Streamable HTTP 的 Agent 或客户端"
+          ],
+          "source": "internal/cli/schema_hints/selection/mcp.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "candidates": [
+            {
+              "value": [
+                "需要把钉钉 MCP 市场中的服务连接到支持 Streamable HTTP 的 Agent 或客户端"
+              ],
+              "source": "internal/cli/schema_hints/selection/mcp.json",
+              "precedence": "reviewed_explicit",
+              "selected": true
+            }
+          ]
+        }
+      },
+      "source_refs": [
+        "CommandRegistry:product=mcp",
+        "cobra-help:dws mcp --help",
+        "internal/cli/schema_command_registry.json#mcp",
+        "internal/cli/schema_hints/selection/mcp.json"
+      ],
+      "use_when": [
+        "需要把钉钉 MCP 市场中的服务连接到支持 Streamable HTTP 的 Agent 或客户端"
+      ]
+    },
     "minutes": {
       "agent_summary": "查询和维护钉钉听记的转写、摘要、待办、权限、录音、标签、说话人总结及文件上传会话。",
       "agent_summary_source": "dws-agent-selection/minutes",
@@ -1450,20 +1726,20 @@
       ]
     },
     "todo": {
-      "agent_summary": "管理待办任务、子任务、执行人、参与人、评论、附件与提醒",
+      "agent_summary": "管理待办任务、标签、子任务、执行人、参与人、评论、附件与提醒",
       "agent_summary_source": "dws-agent-selection/todo",
       "avoid_when": [
         "不要用于 OA 审批流转、工作日志提交或日历日程管理"
       ],
       "field_provenance": {
         "agent_summary": {
-          "value": "管理待办任务、子任务、执行人、参与人、评论、附件与提醒",
+          "value": "管理待办任务、标签、子任务、执行人、参与人、评论、附件与提醒",
           "source": "internal/cli/schema_hints/selection/todo.json",
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
           "candidates": [
             {
-              "value": "管理待办任务、子任务、执行人、参与人、评论、附件与提醒",
+              "value": "管理待办任务、标签、子任务、执行人、参与人、评论、附件与提醒",
               "source": "internal/cli/schema_hints/selection/todo.json",
               "precedence": "reviewed_explicit",
               "selected": true
@@ -1598,13 +1874,17 @@
     "chat",
     "contact",
     "dev",
+    "devapp",
     "devdoc",
     "ding",
     "doc",
     "drive",
     "event",
+    "hrbrain",
     "live",
     "mail",
+    "markdown",
+    "mcp",
     "minutes",
     "oa",
     "pat",

internal/cli/schema_agent_metadata/mcp.json

@@ -0,0 +1,267 @@
+{
+  "product_id": "mcp",
+  "tools": {
+    "mcp url get": {
+      "agent_summary": "按 MCP 市场 mcpId 获取当前用户和组织可用的 Streamable HTTP 地址",
+      "agent_summary_source": "dws-agent-selection/mcp",
+      "availability": "available",
+      "avoid_when": [
+        "只是查询 DWS 已公开命令或参数时使用 dws schema",
+        "用户要求把返回的凭据 URL 发送到群聊、文档、邮件、日志或代码仓库时不要执行或传播"
+      ],
+      "confirmation": "not_required",
+      "effect": "read",
+      "effect_source": "agent-hint",
+      "examples": [
+        "dws mcp url get 10043 --format json"
+      ],
+      "field_provenance": {
+        "agent_summary": {
+          "value": "按 MCP 市场 mcpId 获取当前用户和组织可用的 Streamable HTTP 地址",
+          "source": "internal/cli/schema_hints/selection/mcp.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "正式公开 MCP URL 解析命令的 Agent 选型文案，并明确凭据 URL 只能返回给当前用户、不得二次传播。",
+          "candidates": [
+            {
+              "value": "按 MCP 市场 mcpId 获取当前用户和组织可用的 Streamable HTTP 地址",
+              "source": "internal/cli/schema_hints/selection/mcp.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "正式公开 MCP URL 解析命令的 Agent 选型文案，并明确凭据 URL 只能返回给当前用户、不得二次传播。"
+            }
+          ]
+        },
+        "availability": {
+          "value": "available",
+          "source": "internal/cli/schema_hints/metadata/mcp.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Expose the reviewed MCP URL resolver through the public CLI and Agent Schema while preserving the helper endpoint as a non-product supplemental server. The returned URL contains user- and organization-scoped credentials and must be treated as sensitive output.",
+          "candidates": [
+            {
+              "value": "available",
+              "source": "internal/cli/schema_hints/metadata/mcp.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Expose the reviewed MCP URL resolver through the public CLI and Agent Schema while preserving the helper endpoint as a non-product supplemental server. The returned URL contains user- and organization-scoped credentials and must be treated as sensitive output."
+            }
+          ]
+        },
+        "avoid_when": {
+          "value": [
+            "只是查询 DWS 已公开命令或参数时使用 dws schema",
+            "用户要求把返回的凭据 URL 发送到群聊、文档、邮件、日志或代码仓库时不要执行或传播"
+          ],
+          "source": "internal/cli/schema_hints/selection/mcp.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "正式公开 MCP URL 解析命令的 Agent 选型文案，并明确凭据 URL 只能返回给当前用户、不得二次传播。",
+          "candidates": [
+            {
+              "value": [
+                "只是查询 DWS 已公开命令或参数时使用 dws schema",
+                "用户要求把返回的凭据 URL 发送到群聊、文档、邮件、日志或代码仓库时不要执行或传播"
+              ],
+              "source": "internal/cli/schema_hints/selection/mcp.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "正式公开 MCP URL 解析命令的 Agent 选型文案，并明确凭据 URL 只能返回给当前用户、不得二次传播。"
+            }
+          ]
+        },
+        "confirmation": {
+          "value": "not_required",
+          "source": "internal/cli/schema_hints/metadata/mcp.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Expose the reviewed MCP URL resolver through the public CLI and Agent Schema while preserving the helper endpoint as a non-product supplemental server. The returned URL contains user- and organization-scoped credentials and must be treated as sensitive output.",
+          "candidates": [
+            {
+              "value": "not_required",
+              "source": "internal/cli/schema_hints/metadata/mcp.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Expose the reviewed MCP URL resolver through the public CLI and Agent Schema while preserving the helper endpoint as a non-product supplemental server. The returned URL contains user- and organization-scoped credentials and must be treated as sensitive output."
+            }
+          ]
+        },
+        "effect": {
+          "value": "read",
+          "source": "internal/cli/schema_hints/metadata/mcp.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Expose the reviewed MCP URL resolver through the public CLI and Agent Schema while preserving the helper endpoint as a non-product supplemental server. The returned URL contains user- and organization-scoped credentials and must be treated as sensitive output.",
+          "candidates": [
+            {
+              "value": "read",
+              "source": "internal/cli/schema_hints/metadata/mcp.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Expose the reviewed MCP URL resolver through the public CLI and Agent Schema while preserving the helper endpoint as a non-product supplemental server. The returned URL contains user- and organization-scoped credentials and must be treated as sensitive output."
+            }
+          ]
+        },
+        "examples": {
+          "value": [
+            "dws mcp url get 10043 --format json"
+          ],
+          "source": "internal/cli/schema_hints/selection/mcp.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "正式公开 MCP URL 解析命令的 Agent 选型文案，并明确凭据 URL 只能返回给当前用户、不得二次传播。",
+          "candidates": [
+            {
+              "value": [
+                "dws mcp url get 10043 --format json"
+              ],
+              "source": "internal/cli/schema_hints/selection/mcp.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "正式公开 MCP URL 解析命令的 Agent 选型文案，并明确凭据 URL 只能返回给当前用户、不得二次传播。"
+            }
+          ]
+        },
+        "idempotency": {
+          "value": "idempotent",
+          "source": "internal/cli/schema_hints/metadata/mcp.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Expose the reviewed MCP URL resolver through the public CLI and Agent Schema while preserving the helper endpoint as a non-product supplemental server. The returned URL contains user- and organization-scoped credentials and must be treated as sensitive output.",
+          "candidates": [
+            {
+              "value": "idempotent",
+              "source": "internal/cli/schema_hints/metadata/mcp.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Expose the reviewed MCP URL resolver through the public CLI and Agent Schema while preserving the helper endpoint as a non-product supplemental server. The returned URL contains user- and organization-scoped credentials and must be treated as sensitive output."
+            }
+          ]
+        },
+        "interface_mode": {
+          "value": "composite",
+          "source": "internal/cli/schema_hints/metadata/mcp.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Expose the reviewed MCP URL resolver through the public CLI and Agent Schema while preserving the helper endpoint as a non-product supplemental server. The returned URL contains user- and organization-scoped credentials and must be treated as sensitive output.",
+          "candidates": [
+            {
+              "value": "composite",
+              "source": "internal/cli/schema_hints/metadata/mcp.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Expose the reviewed MCP URL resolver through the public CLI and Agent Schema while preserving the helper endpoint as a non-product supplemental server. The returned URL contains user- and organization-scoped credentials and must be treated as sensitive output."
+            }
+          ]
+        },
+        "interface_reason": {
+          "value": "Reviewed unpinned remote adapter: the public CLI wrapper calls the helper-only mcp-meta/get_mcp_server_url endpoint, which is intentionally absent from the public product catalog and pinned MCP metadata.",
+          "source": "internal/cli/schema_hints/metadata/mcp.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Expose the reviewed MCP URL resolver through the public CLI and Agent Schema while preserving the helper endpoint as a non-product supplemental server. The returned URL contains user- and organization-scoped credentials and must be treated as sensitive output.",
+          "candidates": [
+            {
+              "value": "Reviewed unpinned remote adapter: the public CLI wrapper calls the helper-only mcp-meta/get_mcp_server_url endpoint, which is intentionally absent from the public product catalog and pinned MCP metadata.",
+              "source": "internal/cli/schema_hints/metadata/mcp.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Expose the reviewed MCP URL resolver through the public CLI and Agent Schema while preserving the helper endpoint as a non-product supplemental server. The returned URL contains user- and organization-scoped credentials and must be treated as sensitive output."
+            }
+          ]
+        },
+        "interface_ref": {
+          "value": null,
+          "source": "internal/cli/schema_hints/metadata/mcp.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "interface_disposition_matrix",
+          "review_reason": "final interface mode composite forbids a direct MCP interface_ref",
+          "candidates": [
+            {
+              "value": null,
+              "source": "internal/cli/schema_hints/metadata/mcp.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "final interface mode composite forbids a direct MCP interface_ref"
+            }
+          ]
+        },
+        "reviewed": {
+          "value": true,
+          "source": "internal/cli/schema_hints/metadata/mcp.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Expose the reviewed MCP URL resolver through the public CLI and Agent Schema while preserving the helper endpoint as a non-product supplemental server. The returned URL contains user- and organization-scoped credentials and must be treated as sensitive output.",
+          "candidates": [
+            {
+              "value": true,
+              "source": "internal/cli/schema_hints/metadata/mcp.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Expose the reviewed MCP URL resolver through the public CLI and Agent Schema while preserving the helper endpoint as a non-product supplemental server. The returned URL contains user- and organization-scoped credentials and must be treated as sensitive output."
+            },
+            {
+              "value": true,
+              "source": "internal/cli/schema_hints/selection/mcp.json",
+              "precedence": "reviewed_explicit",
+              "selected": false,
+              "review_reason": "正式公开 MCP URL 解析命令的 Agent 选型文案，并明确凭据 URL 只能返回给当前用户、不得二次传播。"
+            }
+          ]
+        },
+        "risk": {
+          "value": "medium",
+          "source": "internal/cli/schema_hints/metadata/mcp.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Expose the reviewed MCP URL resolver through the public CLI and Agent Schema while preserving the helper endpoint as a non-product supplemental server. The returned URL contains user- and organization-scoped credentials and must be treated as sensitive output.",
+          "candidates": [
+            {
+              "value": "medium",
+              "source": "internal/cli/schema_hints/metadata/mcp.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Expose the reviewed MCP URL resolver through the public CLI and Agent Schema while preserving the helper endpoint as a non-product supplemental server. The returned URL contains user- and organization-scoped credentials and must be treated as sensitive output."
+            }
+          ]
+        },
+        "use_when": {
+          "value": [
+            "已知钉钉 MCP 市场 mcpId，需要获得当前身份可用的 Streamable HTTP 连接地址"
+          ],
+          "source": "internal/cli/schema_hints/selection/mcp.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "正式公开 MCP URL 解析命令的 Agent 选型文案，并明确凭据 URL 只能返回给当前用户、不得二次传播。",
+          "candidates": [
+            {
+              "value": [
+                "已知钉钉 MCP 市场 mcpId，需要获得当前身份可用的 Streamable HTTP 连接地址"
+              ],
+              "source": "internal/cli/schema_hints/selection/mcp.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "正式公开 MCP URL 解析命令的 Agent 选型文案，并明确凭据 URL 只能返回给当前用户、不得二次传播。"
+            }
+          ]
+        }
+      },
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "interface_reason": "Reviewed unpinned remote adapter: the public CLI wrapper calls the helper-only mcp-meta/get_mcp_server_url endpoint, which is intentionally absent from the public product catalog and pinned MCP metadata.",
+      "reviewed": true,
+      "risk": "medium",
+      "source_refs": [
+        "cobra-help:dws mcp url get --help",
+        "internal/app/mcp_url_command.go",
+        "internal/cli/schema_command_registry.json#mcp.url_get",
+        "internal/cli/schema_hints/metadata/mcp.json",
+        "internal/cli/schema_hints/selection/mcp.json",
+        "pkg/edition/default.go#openSupplementServers"
+      ],
+      "use_when": [
+        "已知钉钉 MCP 市场 mcpId，需要获得当前身份可用的 Streamable HTTP 连接地址"
+      ]
+    }
+  }
+}

internal/cli/schema_agent_metadata/report.json

@@ -1,6 +1,525 @@
 {
   "product_id": "report",
   "tools": {
+    "report +inbox-list": {
+      "agent_summary": "列出我收到的日报（按时间范围分页）",
+      "agent_summary_source": "dws-agent-selection/report",
+      "availability": "available",
+      "avoid_when": [
+        "需要该 Shortcut 未公开的底层参数、原始响应或不同执行语义时，改用对应原子命令"
+      ],
+      "confirmation": "not_required",
+      "effect": "read",
+      "effect_source": "agent-hint",
+      "examples": [
+        "dws report +inbox-list --start \"2026-03-10T00:00:00+08:00\" --end \"2026-03-10T23:59:59+08:00\" --cursor 0 --size 20"
+      ],
+      "field_provenance": {
+        "agent_summary": {
+          "value": "列出我收到的日报（按时间范围分页）",
+          "source": "internal/cli/schema_hints/selection/report.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts.",
+          "candidates": [
+            {
+              "value": "列出我收到的日报（按时间范围分页）",
+              "source": "internal/cli/schema_hints/selection/report.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts."
+            }
+          ]
+        },
+        "availability": {
+          "value": "available",
+          "source": "internal/cli/schema_hints/metadata/report.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": "available",
+              "source": "internal/cli/schema_hints/metadata/report.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            }
+          ]
+        },
+        "avoid_when": {
+          "value": [
+            "需要该 Shortcut 未公开的底层参数、原始响应或不同执行语义时，改用对应原子命令"
+          ],
+          "source": "internal/cli/schema_hints/selection/report.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts.",
+          "candidates": [
+            {
+              "value": [
+                "需要该 Shortcut 未公开的底层参数、原始响应或不同执行语义时，改用对应原子命令"
+              ],
+              "source": "internal/cli/schema_hints/selection/report.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts."
+            }
+          ]
+        },
+        "confirmation": {
+          "value": "not_required",
+          "source": "internal/cli/schema_hints/metadata/report.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": "not_required",
+              "source": "internal/cli/schema_hints/metadata/report.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            }
+          ]
+        },
+        "effect": {
+          "value": "read",
+          "source": "internal/cli/schema_hints/metadata/report.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": "read",
+              "source": "internal/cli/schema_hints/metadata/report.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            }
+          ]
+        },
+        "examples": {
+          "value": [
+            "dws report +inbox-list --start \"2026-03-10T00:00:00+08:00\" --end \"2026-03-10T23:59:59+08:00\" --cursor 0 --size 20"
+          ],
+          "source": "internal/cli/schema_hints/selection/report.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts.",
+          "candidates": [
+            {
+              "value": [
+                "dws report +inbox-list --start \"2026-03-10T00:00:00+08:00\" --end \"2026-03-10T23:59:59+08:00\" --cursor 0 --size 20"
+              ],
+              "source": "internal/cli/schema_hints/selection/report.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts."
+            }
+          ]
+        },
+        "idempotency": {
+          "value": "idempotent",
+          "source": "internal/cli/schema_hints/metadata/report.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": "idempotent",
+              "source": "internal/cli/schema_hints/metadata/report.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            }
+          ]
+        },
+        "interface_mode": {
+          "value": "composite",
+          "source": "internal/cli/schema_hints/metadata/report.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": "composite",
+              "source": "internal/cli/schema_hints/metadata/report.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            }
+          ]
+        },
+        "interface_reason": {
+          "value": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+          "source": "internal/cli/schema_hints/metadata/report.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+              "source": "internal/cli/schema_hints/metadata/report.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            }
+          ]
+        },
+        "interface_ref": {
+          "value": null,
+          "source": "internal/cli/schema_hints/metadata/report.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "interface_disposition_matrix",
+          "review_reason": "final interface mode composite forbids a direct MCP interface_ref",
+          "candidates": [
+            {
+              "value": null,
+              "source": "internal/cli/schema_hints/metadata/report.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "final interface mode composite forbids a direct MCP interface_ref"
+            }
+          ]
+        },
+        "reviewed": {
+          "value": true,
+          "source": "internal/cli/schema_hints/metadata/report.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": true,
+              "source": "internal/cli/schema_hints/metadata/report.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            },
+            {
+              "value": true,
+              "source": "internal/cli/schema_hints/selection/report.json",
+              "precedence": "reviewed_explicit",
+              "selected": false,
+              "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts."
+            }
+          ]
+        },
+        "risk": {
+          "value": "low",
+          "source": "internal/cli/schema_hints/metadata/report.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": "low",
+              "source": "internal/cli/schema_hints/metadata/report.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            }
+          ]
+        },
+        "use_when": {
+          "value": [
+            "当你要查看下属或同事发给自己的日报周报、想在某个时间段内浏览或审阅收到的汇报时使用；输入起止时间（ISO-8601），可按发送人 staffId 过滤，分页返回收到的日报列表及其 reportId，供后续 +entry-get 读正文。"
+          ],
+          "source": "internal/cli/schema_hints/selection/report.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts.",
+          "candidates": [
+            {
+              "value": [
+                "当你要查看下属或同事发给自己的日报周报、想在某个时间段内浏览或审阅收到的汇报时使用；输入起止时间（ISO-8601），可按发送人 staffId 过滤，分页返回收到的日报列表及其 reportId，供后续 +entry-get 读正文。"
+              ],
+              "source": "internal/cli/schema_hints/selection/report.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts."
+            }
+          ]
+        }
+      },
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "risk": "low",
+      "source_refs": [
+        "ShortcutRegistry:report +inbox-list",
+        "cobra-help:dws report +inbox-list",
+        "internal/cli/schema_command_registry.json#report.shortcut_inbox_list",
+        "internal/cli/schema_hints/metadata/report.json",
+        "internal/cli/schema_hints/selection/report.json"
+      ],
+      "use_when": [
+        "当你要查看下属或同事发给自己的日报周报、想在某个时间段内浏览或审阅收到的汇报时使用；输入起止时间（ISO-8601），可按发送人 staffId 过滤，分页返回收到的日报列表及其 reportId，供后续 +entry-get 读正文。"
+      ]
+    },
+    "report +outbox-list": {
+      "agent_summary": "列出我发出的日报（可选时间/模版名过滤）",
+      "agent_summary_source": "dws-agent-selection/report",
+      "availability": "available",
+      "avoid_when": [
+        "需要该 Shortcut 未公开的底层参数、原始响应或不同执行语义时，改用对应原子命令"
+      ],
+      "confirmation": "not_required",
+      "effect": "read",
+      "effect_source": "agent-hint",
+      "examples": [
+        "dws report +outbox-list --cursor 0 --size 20",
+        "dws report +outbox-list --cursor 0 --size 20 --template-name \"日报\""
+      ],
+      "field_provenance": {
+        "agent_summary": {
+          "value": "列出我发出的日报（可选时间/模版名过滤）",
+          "source": "internal/cli/schema_hints/selection/report.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts.",
+          "candidates": [
+            {
+              "value": "列出我发出的日报（可选时间/模版名过滤）",
+              "source": "internal/cli/schema_hints/selection/report.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts."
+            }
+          ]
+        },
+        "availability": {
+          "value": "available",
+          "source": "internal/cli/schema_hints/metadata/report.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": "available",
+              "source": "internal/cli/schema_hints/metadata/report.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            }
+          ]
+        },
+        "avoid_when": {
+          "value": [
+            "需要该 Shortcut 未公开的底层参数、原始响应或不同执行语义时，改用对应原子命令"
+          ],
+          "source": "internal/cli/schema_hints/selection/report.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts.",
+          "candidates": [
+            {
+              "value": [
+                "需要该 Shortcut 未公开的底层参数、原始响应或不同执行语义时，改用对应原子命令"
+              ],
+              "source": "internal/cli/schema_hints/selection/report.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts."
+            }
+          ]
+        },
+        "confirmation": {
+          "value": "not_required",
+          "source": "internal/cli/schema_hints/metadata/report.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": "not_required",
+              "source": "internal/cli/schema_hints/metadata/report.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            }
+          ]
+        },
+        "effect": {
+          "value": "read",
+          "source": "internal/cli/schema_hints/metadata/report.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": "read",
+              "source": "internal/cli/schema_hints/metadata/report.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            }
+          ]
+        },
+        "examples": {
+          "value": [
+            "dws report +outbox-list --cursor 0 --size 20",
+            "dws report +outbox-list --cursor 0 --size 20 --template-name \"日报\""
+          ],
+          "source": "internal/cli/schema_hints/selection/report.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts.",
+          "candidates": [
+            {
+              "value": [
+                "dws report +outbox-list --cursor 0 --size 20",
+                "dws report +outbox-list --cursor 0 --size 20 --template-name \"日报\""
+              ],
+              "source": "internal/cli/schema_hints/selection/report.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts."
+            }
+          ]
+        },
+        "idempotency": {
+          "value": "idempotent",
+          "source": "internal/cli/schema_hints/metadata/report.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": "idempotent",
+              "source": "internal/cli/schema_hints/metadata/report.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            }
+          ]
+        },
+        "interface_mode": {
+          "value": "composite",
+          "source": "internal/cli/schema_hints/metadata/report.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": "composite",
+              "source": "internal/cli/schema_hints/metadata/report.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            }
+          ]
+        },
+        "interface_reason": {
+          "value": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+          "source": "internal/cli/schema_hints/metadata/report.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+              "source": "internal/cli/schema_hints/metadata/report.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            }
+          ]
+        },
+        "interface_ref": {
+          "value": null,
+          "source": "internal/cli/schema_hints/metadata/report.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "interface_disposition_matrix",
+          "review_reason": "final interface mode composite forbids a direct MCP interface_ref",
+          "candidates": [
+            {
+              "value": null,
+              "source": "internal/cli/schema_hints/metadata/report.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "final interface mode composite forbids a direct MCP interface_ref"
+            }
+          ]
+        },
+        "reviewed": {
+          "value": true,
+          "source": "internal/cli/schema_hints/metadata/report.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": true,
+              "source": "internal/cli/schema_hints/metadata/report.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            },
+            {
+              "value": true,
+              "source": "internal/cli/schema_hints/selection/report.json",
+              "precedence": "reviewed_explicit",
+              "selected": false,
+              "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts."
+            }
+          ]
+        },
+        "risk": {
+          "value": "low",
+          "source": "internal/cli/schema_hints/metadata/report.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": "low",
+              "source": "internal/cli/schema_hints/metadata/report.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            }
+          ]
+        },
+        "use_when": {
+          "value": [
+            "当你要回顾自己写过、提交过的日报周报，比如确认某天是否已交、找回历史汇报内容或统计提交情况时使用；可按创建/修改时间范围和模版名过滤，分页返回自己发出的日报列表及 reportId，供后续 +entry-get 查看正文。"
+          ],
+          "source": "internal/cli/schema_hints/selection/report.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts.",
+          "candidates": [
+            {
+              "value": [
+                "当你要回顾自己写过、提交过的日报周报，比如确认某天是否已交、找回历史汇报内容或统计提交情况时使用；可按创建/修改时间范围和模版名过滤，分页返回自己发出的日报列表及 reportId，供后续 +entry-get 查看正文。"
+              ],
+              "source": "internal/cli/schema_hints/selection/report.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts."
+            }
+          ]
+        }
+      },
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "risk": "low",
+      "source_refs": [
+        "ShortcutRegistry:report +outbox-list",
+        "cobra-help:dws report +outbox-list",
+        "internal/cli/schema_command_registry.json#report.shortcut_outbox_list",
+        "internal/cli/schema_hints/metadata/report.json",
+        "internal/cli/schema_hints/selection/report.json"
+      ],
+      "use_when": [
+        "当你要回顾自己写过、提交过的日报周报，比如确认某天是否已交、找回历史汇报内容或统计提交情况时使用；可按创建/修改时间范围和模版名过滤，分页返回自己发出的日报列表及 reportId，供后续 +entry-get 查看正文。"
+      ]
+    },
     "report entry get": {
       "agent_summary": "获取指定一篇日志的详情信息",
       "agent_summary_source": "dws-agent-selection/report",

internal/cli/schema_agent_metadata/sheet.json

@@ -1,6 +1,522 @@
 {
   "product_id": "sheet",
   "tools": {
+    "sheet +list-sheets": {
+      "agent_summary": "获取表格文档中全部工作表列表",
+      "agent_summary_source": "dws-agent-selection/sheet",
+      "availability": "available",
+      "avoid_when": [
+        "需要该 Shortcut 未公开的底层参数、原始响应或不同执行语义时，改用对应原子命令"
+      ],
+      "confirmation": "not_required",
+      "effect": "read",
+      "effect_source": "agent-hint",
+      "examples": [
+        "dws sheet +list-sheets --node NODE_ID"
+      ],
+      "field_provenance": {
+        "agent_summary": {
+          "value": "获取表格文档中全部工作表列表",
+          "source": "internal/cli/schema_hints/selection/sheet.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts.",
+          "candidates": [
+            {
+              "value": "获取表格文档中全部工作表列表",
+              "source": "internal/cli/schema_hints/selection/sheet.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts."
+            }
+          ]
+        },
+        "availability": {
+          "value": "available",
+          "source": "internal/cli/schema_hints/metadata/sheet.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": "available",
+              "source": "internal/cli/schema_hints/metadata/sheet.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            }
+          ]
+        },
+        "avoid_when": {
+          "value": [
+            "需要该 Shortcut 未公开的底层参数、原始响应或不同执行语义时，改用对应原子命令"
+          ],
+          "source": "internal/cli/schema_hints/selection/sheet.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts.",
+          "candidates": [
+            {
+              "value": [
+                "需要该 Shortcut 未公开的底层参数、原始响应或不同执行语义时，改用对应原子命令"
+              ],
+              "source": "internal/cli/schema_hints/selection/sheet.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts."
+            }
+          ]
+        },
+        "confirmation": {
+          "value": "not_required",
+          "source": "internal/cli/schema_hints/metadata/sheet.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": "not_required",
+              "source": "internal/cli/schema_hints/metadata/sheet.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            }
+          ]
+        },
+        "effect": {
+          "value": "read",
+          "source": "internal/cli/schema_hints/metadata/sheet.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": "read",
+              "source": "internal/cli/schema_hints/metadata/sheet.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            }
+          ]
+        },
+        "examples": {
+          "value": [
+            "dws sheet +list-sheets --node NODE_ID"
+          ],
+          "source": "internal/cli/schema_hints/selection/sheet.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts.",
+          "candidates": [
+            {
+              "value": [
+                "dws sheet +list-sheets --node NODE_ID"
+              ],
+              "source": "internal/cli/schema_hints/selection/sheet.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts."
+            }
+          ]
+        },
+        "idempotency": {
+          "value": "idempotent",
+          "source": "internal/cli/schema_hints/metadata/sheet.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": "idempotent",
+              "source": "internal/cli/schema_hints/metadata/sheet.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            }
+          ]
+        },
+        "interface_mode": {
+          "value": "composite",
+          "source": "internal/cli/schema_hints/metadata/sheet.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": "composite",
+              "source": "internal/cli/schema_hints/metadata/sheet.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            }
+          ]
+        },
+        "interface_reason": {
+          "value": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+          "source": "internal/cli/schema_hints/metadata/sheet.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+              "source": "internal/cli/schema_hints/metadata/sheet.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            }
+          ]
+        },
+        "interface_ref": {
+          "value": null,
+          "source": "internal/cli/schema_hints/metadata/sheet.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "interface_disposition_matrix",
+          "review_reason": "final interface mode composite forbids a direct MCP interface_ref",
+          "candidates": [
+            {
+              "value": null,
+              "source": "internal/cli/schema_hints/metadata/sheet.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "final interface mode composite forbids a direct MCP interface_ref"
+            }
+          ]
+        },
+        "reviewed": {
+          "value": true,
+          "source": "internal/cli/schema_hints/metadata/sheet.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": true,
+              "source": "internal/cli/schema_hints/metadata/sheet.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            },
+            {
+              "value": true,
+              "source": "internal/cli/schema_hints/selection/sheet.json",
+              "precedence": "reviewed_explicit",
+              "selected": false,
+              "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts."
+            }
+          ]
+        },
+        "risk": {
+          "value": "low",
+          "source": "internal/cli/schema_hints/metadata/sheet.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": "low",
+              "source": "internal/cli/schema_hints/metadata/sheet.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            }
+          ]
+        },
+        "use_when": {
+          "value": [
+            "当你拿到一个表格文档、想先了解它里面有哪些工作表（sheet）以及各自的 sheetId 时使用，通常作为读写具体数据前的第一步；传入表格文档 ID 或 URL，返回工作表清单。"
+          ],
+          "source": "internal/cli/schema_hints/selection/sheet.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts.",
+          "candidates": [
+            {
+              "value": [
+                "当你拿到一个表格文档、想先了解它里面有哪些工作表（sheet）以及各自的 sheetId 时使用，通常作为读写具体数据前的第一步；传入表格文档 ID 或 URL，返回工作表清单。"
+              ],
+              "source": "internal/cli/schema_hints/selection/sheet.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts."
+            }
+          ]
+        }
+      },
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "risk": "low",
+      "source_refs": [
+        "ShortcutRegistry:sheet +list-sheets",
+        "cobra-help:dws sheet +list-sheets",
+        "internal/cli/schema_command_registry.json#sheet.shortcut_list_sheets",
+        "internal/cli/schema_hints/metadata/sheet.json",
+        "internal/cli/schema_hints/selection/sheet.json"
+      ],
+      "use_when": [
+        "当你拿到一个表格文档、想先了解它里面有哪些工作表（sheet）以及各自的 sheetId 时使用，通常作为读写具体数据前的第一步；传入表格文档 ID 或 URL，返回工作表清单。"
+      ]
+    },
+    "sheet +read": {
+      "agent_summary": "读取工作表指定范围的结构化单元格数据",
+      "agent_summary_source": "dws-agent-selection/sheet",
+      "availability": "available",
+      "avoid_when": [
+        "需要该 Shortcut 未公开的底层参数、原始响应或不同执行语义时，改用对应原子命令"
+      ],
+      "confirmation": "not_required",
+      "effect": "read",
+      "effect_source": "agent-hint",
+      "examples": [
+        "dws sheet +read --node NODE_ID --sheet-id SHEET_ID --range \"A1:D10\""
+      ],
+      "field_provenance": {
+        "agent_summary": {
+          "value": "读取工作表指定范围的结构化单元格数据",
+          "source": "internal/cli/schema_hints/selection/sheet.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts.",
+          "candidates": [
+            {
+              "value": "读取工作表指定范围的结构化单元格数据",
+              "source": "internal/cli/schema_hints/selection/sheet.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts."
+            }
+          ]
+        },
+        "availability": {
+          "value": "available",
+          "source": "internal/cli/schema_hints/metadata/sheet.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": "available",
+              "source": "internal/cli/schema_hints/metadata/sheet.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            }
+          ]
+        },
+        "avoid_when": {
+          "value": [
+            "需要该 Shortcut 未公开的底层参数、原始响应或不同执行语义时，改用对应原子命令"
+          ],
+          "source": "internal/cli/schema_hints/selection/sheet.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts.",
+          "candidates": [
+            {
+              "value": [
+                "需要该 Shortcut 未公开的底层参数、原始响应或不同执行语义时，改用对应原子命令"
+              ],
+              "source": "internal/cli/schema_hints/selection/sheet.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts."
+            }
+          ]
+        },
+        "confirmation": {
+          "value": "not_required",
+          "source": "internal/cli/schema_hints/metadata/sheet.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": "not_required",
+              "source": "internal/cli/schema_hints/metadata/sheet.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            }
+          ]
+        },
+        "effect": {
+          "value": "read",
+          "source": "internal/cli/schema_hints/metadata/sheet.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": "read",
+              "source": "internal/cli/schema_hints/metadata/sheet.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            }
+          ]
+        },
+        "examples": {
+          "value": [
+            "dws sheet +read --node NODE_ID --sheet-id SHEET_ID --range \"A1:D10\""
+          ],
+          "source": "internal/cli/schema_hints/selection/sheet.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts.",
+          "candidates": [
+            {
+              "value": [
+                "dws sheet +read --node NODE_ID --sheet-id SHEET_ID --range \"A1:D10\""
+              ],
+              "source": "internal/cli/schema_hints/selection/sheet.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts."
+            }
+          ]
+        },
+        "idempotency": {
+          "value": "idempotent",
+          "source": "internal/cli/schema_hints/metadata/sheet.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": "idempotent",
+              "source": "internal/cli/schema_hints/metadata/sheet.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            }
+          ]
+        },
+        "interface_mode": {
+          "value": "composite",
+          "source": "internal/cli/schema_hints/metadata/sheet.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": "composite",
+              "source": "internal/cli/schema_hints/metadata/sheet.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            }
+          ]
+        },
+        "interface_reason": {
+          "value": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+          "source": "internal/cli/schema_hints/metadata/sheet.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+              "source": "internal/cli/schema_hints/metadata/sheet.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            }
+          ]
+        },
+        "interface_ref": {
+          "value": null,
+          "source": "internal/cli/schema_hints/metadata/sheet.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "interface_disposition_matrix",
+          "review_reason": "final interface mode composite forbids a direct MCP interface_ref",
+          "candidates": [
+            {
+              "value": null,
+              "source": "internal/cli/schema_hints/metadata/sheet.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "final interface mode composite forbids a direct MCP interface_ref"
+            }
+          ]
+        },
+        "reviewed": {
+          "value": true,
+          "source": "internal/cli/schema_hints/metadata/sheet.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": true,
+              "source": "internal/cli/schema_hints/metadata/sheet.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            },
+            {
+              "value": true,
+              "source": "internal/cli/schema_hints/selection/sheet.json",
+              "precedence": "reviewed_explicit",
+              "selected": false,
+              "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts."
+            }
+          ]
+        },
+        "risk": {
+          "value": "low",
+          "source": "internal/cli/schema_hints/metadata/sheet.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": "low",
+              "source": "internal/cli/schema_hints/metadata/sheet.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            }
+          ]
+        },
+        "use_when": {
+          "value": [
+            "当你需要按单元格逐格获取数据（含类型、公式或格式化值等结构化信息）以便程序处理时使用；传入表格与可选范围（A1 表示法，不传则全部），可指定取格式化值/原始值/公式，返回结构化单元格数组。若只想要纯文本可改用 +csv-get。"
+          ],
+          "source": "internal/cli/schema_hints/selection/sheet.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts.",
+          "candidates": [
+            {
+              "value": [
+                "当你需要按单元格逐格获取数据（含类型、公式或格式化值等结构化信息）以便程序处理时使用；传入表格与可选范围（A1 表示法，不传则全部），可指定取格式化值/原始值/公式，返回结构化单元格数组。若只想要纯文本可改用 +csv-get。"
+              ],
+              "source": "internal/cli/schema_hints/selection/sheet.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts."
+            }
+          ]
+        }
+      },
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "risk": "low",
+      "source_refs": [
+        "ShortcutRegistry:sheet +read",
+        "cobra-help:dws sheet +read",
+        "internal/cli/schema_command_registry.json#sheet.shortcut_read",
+        "internal/cli/schema_hints/metadata/sheet.json",
+        "internal/cli/schema_hints/selection/sheet.json"
+      ],
+      "use_when": [
+        "当你需要按单元格逐格获取数据（含类型、公式或格式化值等结构化信息）以便程序处理时使用；传入表格与可选范围（A1 表示法，不传则全部），可指定取格式化值/原始值/公式，返回结构化单元格数组。若只想要纯文本可改用 +csv-get。"
+      ]
+    },
     "sheet add-dimension": {
       "agent_summary": "在工作表末尾追加空行或空列。",
       "agent_summary_source": "dws-agent-selection/sheet",

internal/cli/schema_agent_metadata/wiki.json

@@ -1,6 +1,264 @@
 {
   "product_id": "wiki",
   "tools": {
+    "wiki +space-search": {
+      "agent_summary": "搜索知识库",
+      "agent_summary_source": "dws-agent-selection/wiki",
+      "availability": "available",
+      "avoid_when": [
+        "需要该 Shortcut 未公开的底层参数、原始响应或不同执行语义时，改用对应原子命令"
+      ],
+      "confirmation": "not_required",
+      "effect": "read",
+      "effect_source": "agent-hint",
+      "examples": [
+        "dws wiki +space-search --query \"产品文档\""
+      ],
+      "field_provenance": {
+        "agent_summary": {
+          "value": "搜索知识库",
+          "source": "internal/cli/schema_hints/selection/wiki.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts.",
+          "candidates": [
+            {
+              "value": "搜索知识库",
+              "source": "internal/cli/schema_hints/selection/wiki.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts."
+            }
+          ]
+        },
+        "availability": {
+          "value": "available",
+          "source": "internal/cli/schema_hints/metadata/wiki.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": "available",
+              "source": "internal/cli/schema_hints/metadata/wiki.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            }
+          ]
+        },
+        "avoid_when": {
+          "value": [
+            "需要该 Shortcut 未公开的底层参数、原始响应或不同执行语义时，改用对应原子命令"
+          ],
+          "source": "internal/cli/schema_hints/selection/wiki.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts.",
+          "candidates": [
+            {
+              "value": [
+                "需要该 Shortcut 未公开的底层参数、原始响应或不同执行语义时，改用对应原子命令"
+              ],
+              "source": "internal/cli/schema_hints/selection/wiki.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts."
+            }
+          ]
+        },
+        "confirmation": {
+          "value": "not_required",
+          "source": "internal/cli/schema_hints/metadata/wiki.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": "not_required",
+              "source": "internal/cli/schema_hints/metadata/wiki.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            }
+          ]
+        },
+        "effect": {
+          "value": "read",
+          "source": "internal/cli/schema_hints/metadata/wiki.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": "read",
+              "source": "internal/cli/schema_hints/metadata/wiki.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            }
+          ]
+        },
+        "examples": {
+          "value": [
+            "dws wiki +space-search --query \"产品文档\""
+          ],
+          "source": "internal/cli/schema_hints/selection/wiki.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts.",
+          "candidates": [
+            {
+              "value": [
+                "dws wiki +space-search --query \"产品文档\""
+              ],
+              "source": "internal/cli/schema_hints/selection/wiki.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts."
+            }
+          ]
+        },
+        "idempotency": {
+          "value": "idempotent",
+          "source": "internal/cli/schema_hints/metadata/wiki.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": "idempotent",
+              "source": "internal/cli/schema_hints/metadata/wiki.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            }
+          ]
+        },
+        "interface_mode": {
+          "value": "composite",
+          "source": "internal/cli/schema_hints/metadata/wiki.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": "composite",
+              "source": "internal/cli/schema_hints/metadata/wiki.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            }
+          ]
+        },
+        "interface_reason": {
+          "value": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+          "source": "internal/cli/schema_hints/metadata/wiki.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+              "source": "internal/cli/schema_hints/metadata/wiki.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            }
+          ]
+        },
+        "interface_ref": {
+          "value": null,
+          "source": "internal/cli/schema_hints/metadata/wiki.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "interface_disposition_matrix",
+          "review_reason": "final interface mode composite forbids a direct MCP interface_ref",
+          "candidates": [
+            {
+              "value": null,
+              "source": "internal/cli/schema_hints/metadata/wiki.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "final interface mode composite forbids a direct MCP interface_ref"
+            }
+          ]
+        },
+        "reviewed": {
+          "value": true,
+          "source": "internal/cli/schema_hints/metadata/wiki.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": true,
+              "source": "internal/cli/schema_hints/metadata/wiki.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            },
+            {
+              "value": true,
+              "source": "internal/cli/schema_hints/selection/wiki.json",
+              "precedence": "reviewed_explicit",
+              "selected": false,
+              "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts."
+            }
+          ]
+        },
+        "risk": {
+          "value": "low",
+          "source": "internal/cli/schema_hints/metadata/wiki.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+          "candidates": [
+            {
+              "value": "low",
+              "source": "internal/cli/schema_hints/metadata/wiki.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity."
+            }
+          ]
+        },
+        "use_when": {
+          "value": [
+            "当你只记得知识库名称的部分关键词、想快速按名称定位某个知识库时使用；输入关键词返回匹配的知识库列表，比逐页 +space-list 更快找到目标 workspaceId。"
+          ],
+          "source": "internal/cli/schema_hints/selection/wiki.json",
+          "precedence": "reviewed_explicit",
+          "resolution": "highest_precedence",
+          "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts.",
+          "candidates": [
+            {
+              "value": [
+                "当你只记得知识库名称的部分关键词、想快速按名称定位某个知识库时使用；输入关键词返回匹配的知识库列表，比逐页 +space-list 更快找到目标 workspaceId。"
+              ],
+              "source": "internal/cli/schema_hints/selection/wiki.json",
+              "precedence": "reviewed_explicit",
+              "selected": true,
+              "review_reason": "Agent-authored and reviewed from the built-in Shortcut intent, executable Cobra path, and declared outcome; it affects selection only and does not alter execution or safety facts."
+            }
+          ]
+        }
+      },
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "risk": "low",
+      "source_refs": [
+        "ShortcutRegistry:wiki +space-search",
+        "cobra-help:dws wiki +space-search",
+        "internal/cli/schema_command_registry.json#wiki.shortcut_space_search",
+        "internal/cli/schema_hints/metadata/wiki.json",
+        "internal/cli/schema_hints/selection/wiki.json"
+      ],
+      "use_when": [
+        "当你只记得知识库名称的部分关键词、想快速按名称定位某个知识库时使用；输入关键词返回匹配的知识库列表，比逐页 +space-list 更快找到目标 workspaceId。"
+      ]
+    },
     "wiki member add": {
       "agent_summary": "为指定知识库添加一个或多个成员，并授予指定角色",
       "agent_summary_source": "dws-agent-selection/wiki",

internal/cli/schema_command_exclusions.json

@@ -48,223 +48,6 @@
         "version"
       ]
     },
-    {
-      "id": "shortcut-catalog-surface",
-      "reason": "Shortcut commands are high-level task entries discovered through dws shortcut list and product skills; they are intentionally separate from the runtime Schema catalog of base Agent tools.",
-      "reviewed": true,
-      "commands": [
-        "aitable +base-get",
-        "aitable +base-list",
-        "aitable +base-search",
-        "aitable +chart-get",
-        "aitable +chart-widgets-example",
-        "aitable +dashboard-config-example",
-        "aitable +dashboard-get",
-        "aitable +field-get",
-        "aitable +find-record",
-        "aitable +form-field-list",
-        "aitable +form-list",
-        "aitable +form-share-get",
-        "aitable +list-tables",
-        "aitable +record-history-list",
-        "aitable +record-query",
-        "aitable +record-query-empty",
-        "aitable +record-share-links",
-        "aitable +record-share-url",
-        "aitable +resolve-base",
-        "aitable +resolve-table",
-        "aitable +role-list",
-        "aitable +section-list-empty",
-        "aitable +section-list-nodes",
-        "aitable +table-get",
-        "aitable +template-search",
-        "aitable +view-get",
-        "aitable +view-get-frozen-cols",
-        "aitable +view-get-lock",
-        "aitable +view-get-row-height",
-        "attendance +check-record",
-        "attendance +check-result",
-        "attendance +get-adjustment-rule",
-        "attendance +get-approve-template",
-        "attendance +get-checkin-record",
-        "attendance +get-leave-records",
-        "attendance +get-overtime-rule",
-        "attendance +get-schedule",
-        "attendance +get-self-setting",
-        "attendance +get-summary",
-        "attendance +list-approve",
-        "attendance +list-leave-types",
-        "attendance +my-attendance",
-        "attendance +query-report-data",
-        "attendance +search-adjustment-rule",
-        "attendance +search-class",
-        "attendance +search-group",
-        "attendance +search-overtime-rule",
-        "attendance +this-month",
-        "calendar +agenda",
-        "calendar +attendee-list",
-        "calendar +book",
-        "calendar +book-list",
-        "calendar +book-search",
-        "calendar +cancel-event",
-        "calendar +conflicts",
-        "calendar +free",
-        "calendar +free-slots",
-        "calendar +freebusy",
-        "calendar +invite",
-        "calendar +my-free",
-        "calendar +next-event",
-        "calendar +reschedule",
-        "calendar +room-groups",
-        "calendar +room-search",
-        "calendar +suggest-time",
-        "calendar +today",
-        "calendar +tomorrow",
-        "calendar +week",
-        "chat +at-me",
-        "chat +bot-find",
-        "chat +bot-search",
-        "chat +broadcast",
-        "chat +category-create",
-        "chat +category-delete",
-        "chat +category-list",
-        "chat +category-rename",
-        "chat +chat-bots",
-        "chat +chat-dismiss",
-        "chat +chat-invite-url",
-        "chat +chat-list-all",
-        "chat +chat-list-join-requests",
-        "chat +chat-list-mine",
-        "chat +chat-mute",
-        "chat +chat-role-add",
-        "chat +chat-role-list",
-        "chat +chat-role-query-user",
-        "chat +chat-role-set-user",
-        "chat +chat-role-update",
-        "chat +chat-search",
-        "chat +chat-set-admin",
-        "chat +chat-set-history",
-        "chat +chat-update-alias",
-        "chat +chat-update-nick",
-        "chat +conversation-clear-all-red-point",
-        "chat +conversation-info",
-        "chat +conversation-list",
-        "chat +conversation-list-top",
-        "chat +dm",
-        "chat +group-members",
-        "chat +messages-list-direct",
-        "chat +messages-list-pin",
-        "chat +messages-list-unread-conversations",
-        "chat +messages-mget",
-        "chat +messages-query-send-status",
-        "chat +messages-read-status",
-        "chat +messages-send-by-webhook",
-        "chat +messages-update-card",
-        "chat +my-groups",
-        "chat +send-to-group",
-        "chat +unread-chats",
-        "contact +by-mobile",
-        "contact +dept-members",
-        "contact +list-dept-members",
-        "contact +list-followings",
-        "contact +list-role-members",
-        "contact +list-roles",
-        "contact +list-sub-depts",
-        "contact +lookup",
-        "contact +me",
-        "contact +org",
-        "contact +resolve-dept",
-        "contact +search-mobile",
-        "contact +search-user",
-        "contact +team",
-        "devapp +create",
-        "devapp +delete",
-        "devapp +disable",
-        "devapp +enable",
-        "devapp +event-list",
-        "devapp +get",
-        "devapp +list",
-        "devapp +member-add",
-        "devapp +member-list",
-        "devapp +member-remove",
-        "devapp +permission-list",
-        "devapp +robot-get",
-        "devapp +update",
-        "devapp +version-check-approval",
-        "devapp +version-get",
-        "devapp +version-list",
-        "devapp +version-status",
-        "devapp +webapp-config",
-        "devapp +webapp-get",
-        "ding +list",
-        "ding +recall-personal",
-        "ding +receiver-status",
-        "ding +send-personal",
-        "doc +comment-create",
-        "doc +comment-list",
-        "doc +comment-reply",
-        "doc +copy",
-        "doc +doc-append",
-        "doc +export-get",
-        "doc +export-submit",
-        "doc +find-doc",
-        "doc +list",
-        "doc +move",
-        "doc +search",
-        "doc +share-doc",
-        "doc +template-list",
-        "doc +template-search",
-        "doc +version-list",
-        "doc +version-revert",
-        "doc +version-save",
-        "drive +copy",
-        "drive +find-file",
-        "drive +info",
-        "drive +move",
-        "drive +recent",
-        "drive +search",
-        "drive +search-docs",
-        "mail +contact-list",
-        "mail +find-mail-user",
-        "mail +folder-list",
-        "mail +recent-mail",
-        "mail +search-mail",
-        "mail +tag-list",
-        "mail +template-list",
-        "mail +thread-list",
-        "mail +unread-mail",
-        "mail +user-search",
-        "minutes +detail",
-        "minutes +list-all",
-        "minutes +list-mine",
-        "minutes +list-shared",
-        "minutes +record-start",
-        "minutes +replace-batch",
-        "oa +list-cc",
-        "oa +list-executed",
-        "oa +list-forms",
-        "oa +list-pending",
-        "oa +list-submitted",
-        "oa +my-initiated",
-        "oa +search-forms",
-        "report +inbox-list",
-        "report +outbox-list",
-        "sheet +list-sheets",
-        "sheet +read",
-        "todo +assign",
-        "todo +assign-multi",
-        "todo +created-todos",
-        "todo +get",
-        "todo +get-my-tasks",
-        "todo +list-attachment",
-        "todo +list-comment",
-        "todo +list-sub",
-        "todo +overdue",
-        "todo +remind",
-        "todo +todo-done",
-        "wiki +space-search"
-      ]
-    },
     {
       "id": "agoal-out-of-surface",
       "reason": "The Agoal product remains executable for compatibility but is outside the currently reviewed open-source Agent command surface.",
@@ -316,12 +99,10 @@
         "chat group notice list",
         "chat group share-invite",
         "chat group update-alias",
-        "chat group update-nick",
         "chat hide",
         "chat list-all-conversations",
         "chat mark-read",
         "chat mark-unread",
-        "chat media upload",
         "chat message list-emotion-replies",
         "chat message set-top-msg",
         "chat message unset-top-msg",

internal/cli/schema_command_registry.go

@@ -34,7 +34,7 @@ var embeddedSchemaCommandRegistrySchemaJSON []byte
 var (
 	commandRegistryProductIDPattern = regexp.MustCompile(`^[A-Za-z0-9][A-Za-z0-9_-]*$`)
 	commandRegistryCanonicalPattern = regexp.MustCompile(`^[A-Za-z0-9][A-Za-z0-9_-]*\.[A-Za-z0-9][A-Za-z0-9_.-]*$`)
-	commandRegistryCLIPathToken     = regexp.MustCompile(`^[A-Za-z0-9][A-Za-z0-9._:-]*$`)
+	commandRegistryCLIPathToken     = regexp.MustCompile(`^(?:[A-Za-z0-9][A-Za-z0-9._:-]*|\+[A-Za-z0-9][A-Za-z0-9._:-]*)$`)
 )
 
 type schemaCommandRegistrySnapshot struct {

internal/cli/schema_command_registry.schema.json

@@ -36,8 +36,8 @@
     },
     "cliPath": {
       "type": "string",
-      "pattern": "^(?!dws(?:\\s|$))(?!.*(?:^|\\s)--)(?!.*[*?\\[\\]])[A-Za-z0-9][A-Za-z0-9._:-]*(?: [A-Za-z0-9][A-Za-z0-9._:-]*)*$",
-      "description": "Exact Cobra command path without the leading 'dws', flags, wildcard characters, or surrounding/repeated whitespace."
+      "pattern": "^(?!dws(?:\\s|$))(?!.*(?:^|\\s)--)(?!.*[*?\\[\\]])(?:[A-Za-z0-9][A-Za-z0-9._:-]*|\\+[A-Za-z0-9][A-Za-z0-9._:-]*)(?: (?:[A-Za-z0-9][A-Za-z0-9._:-]*|\\+[A-Za-z0-9][A-Za-z0-9._:-]*))*$",
+      "description": "Exact Cobra command path without the leading 'dws', flags, wildcard characters, or surrounding/repeated whitespace. A token may use the reviewed '+' shortcut prefix."
     },
     "product": {
       "type": "object",

internal/cli/schema_command_registry_contract_test.go

@@ -94,6 +94,14 @@ func TestDecodeCommandRegistryEnforcesReviewedSourceConstraints(t *testing.T) {
 	if got := registry.ByCanonical["sample.run"].Visibility; got != SchemaVisibilityPublic {
 		t.Fatalf("default visibility = %q, want public", got)
 	}
+
+	shortcut, err := decodeCommandRegistry([]byte(wrap(`[{"id":"sample","tools":[{"canonical_path":"sample.shortcut_run","cli_path":"sample +run"}]}]`)))
+	if err != nil {
+		t.Fatalf("decode shortcut registry path: %v", err)
+	}
+	if got := shortcut.ByCanonical["sample.shortcut_run"].PrimaryCLIPath; got != "sample +run" {
+		t.Fatalf("shortcut primary path = %q, want sample +run", got)
+	}
 }
 
 func TestCommandRegistryHashCoversEveryStableCommandField(t *testing.T) {

internal/cli/schema_dry_run_capabilities.go

@@ -35,6 +35,10 @@ var reviewedDryRunCapabilityGroups = []dryRunCapabilityGroup{
 		"doc.upload",
 		"drive.download_file",
 		"drive.upload",
+		"markdown.create",
+		"markdown.fetch",
+		"markdown.overwrite",
+		"markdown.patch",
 		"sheet.filter_view_get_criteria",
 		"sheet.filter_view_info",
 		"sheet.filter_view_list_criteria",

internal/cli/schema_hints/index.json

@@ -16,13 +16,17 @@
     "chat": "metadata/chat.json",
     "contact": "metadata/contact.json",
     "dev": "metadata/dev.json",
+    "devapp": "metadata/devapp.json",
     "devdoc": "metadata/devdoc.json",
     "ding": "metadata/ding.json",
     "doc": "metadata/doc.json",
     "drive": "metadata/drive.json",
     "event": "metadata/event.json",
+    "hrbrain": "metadata/hrbrain.json",
     "live": "metadata/live.json",
     "mail": "metadata/mail.json",
+    "markdown": "metadata/markdown.json",
+    "mcp": "metadata/mcp.json",
     "minutes": "metadata/minutes.json",
     "oa": "metadata/oa.json",
     "pat": "metadata/pat.json",
@@ -40,13 +44,17 @@
     "chat": "selection/chat.json",
     "contact": "selection/contact.json",
     "dev": "selection/dev.json",
+    "devapp": "selection/devapp.json",
     "devdoc": "selection/devdoc.json",
     "ding": "selection/ding.json",
     "doc": "selection/doc.json",
     "drive": "selection/drive.json",
     "event": "selection/event.json",
+    "hrbrain": "selection/hrbrain.json",
     "live": "selection/live.json",
     "mail": "selection/mail.json",
+    "markdown": "selection/markdown.json",
+    "mcp": "selection/mcp.json",
     "minutes": "selection/minutes.json",
     "oa": "selection/oa.json",
     "pat": "selection/pat.json",
@@ -434,10 +442,6 @@
       "status": "stale",
       "reason": "旧版 Skill 或说明性引用，当前公开命令面无等价 leaf，禁止词法模糊映射"
     },
-    "chat media upload": {
-      "status": "stale",
-      "reason": "旧版 Skill 或说明性引用，当前公开命令面无等价 leaf，禁止词法模糊映射"
-    },
     "chat message list-emotion-replies": {
       "status": "stale",
       "reason": "旧版 Skill 或说明性引用，当前公开命令面无等价 leaf，禁止词法模糊映射"

internal/cli/schema_hints/metadata/aitable.json

@@ -1056,6 +1056,383 @@
       "review_reason": "The command maps --base-id, --workflow-id and a decoded complete workflow-dsl/v1 --dsl object to aitable/update_workflow. This is full replacement rather than a partial patch; locale remains optional.",
       "cli_path": "aitable workflow update",
       "runtime_gate": "none"
+    },
+    "aitable.shortcut_base_list": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "aitable +base-list",
+      "runtime_gate": "none"
+    },
+    "aitable.shortcut_base_search": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "aitable +base-search",
+      "runtime_gate": "none"
+    },
+    "aitable.shortcut_base_get": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "aitable +base-get",
+      "runtime_gate": "none"
+    },
+    "aitable.shortcut_table_get": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "aitable +table-get",
+      "runtime_gate": "none"
+    },
+    "aitable.shortcut_field_get": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "aitable +field-get",
+      "runtime_gate": "none"
+    },
+    "aitable.shortcut_record_query": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "aitable +record-query",
+      "runtime_gate": "none"
+    },
+    "aitable.shortcut_record_query_empty": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "aitable +record-query-empty",
+      "runtime_gate": "none"
+    },
+    "aitable.shortcut_record_history_list": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "aitable +record-history-list",
+      "runtime_gate": "none"
+    },
+    "aitable.shortcut_record_share_url": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "aitable +record-share-url",
+      "runtime_gate": "none"
+    },
+    "aitable.shortcut_template_search": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "aitable +template-search",
+      "runtime_gate": "none"
+    },
+    "aitable.shortcut_view_get": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "aitable +view-get",
+      "runtime_gate": "none"
+    },
+    "aitable.shortcut_view_get_lock": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "aitable +view-get-lock",
+      "runtime_gate": "none"
+    },
+    "aitable.shortcut_view_get_frozen_cols": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "aitable +view-get-frozen-cols",
+      "runtime_gate": "none"
+    },
+    "aitable.shortcut_view_get_row_height": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "aitable +view-get-row-height",
+      "runtime_gate": "none"
+    },
+    "aitable.shortcut_form_list": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "aitable +form-list",
+      "runtime_gate": "none"
+    },
+    "aitable.shortcut_form_field_list": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "aitable +form-field-list",
+      "runtime_gate": "none"
+    },
+    "aitable.shortcut_form_share_get": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "aitable +form-share-get",
+      "runtime_gate": "none"
+    },
+    "aitable.shortcut_dashboard_config_example": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "aitable +dashboard-config-example",
+      "runtime_gate": "none"
+    },
+    "aitable.shortcut_dashboard_get": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "aitable +dashboard-get",
+      "runtime_gate": "none"
+    },
+    "aitable.shortcut_chart_widgets_example": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "aitable +chart-widgets-example",
+      "runtime_gate": "none"
+    },
+    "aitable.shortcut_chart_get": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "aitable +chart-get",
+      "runtime_gate": "none"
+    },
+    "aitable.shortcut_role_list": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "aitable +role-list",
+      "runtime_gate": "none"
+    },
+    "aitable.shortcut_section_list_empty": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "aitable +section-list-empty",
+      "runtime_gate": "none"
+    },
+    "aitable.shortcut_section_list_nodes": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "aitable +section-list-nodes",
+      "runtime_gate": "none"
+    },
+    "aitable.shortcut_find_record": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "aitable +find-record",
+      "runtime_gate": "none"
+    },
+    "aitable.shortcut_list_tables": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "aitable +list-tables",
+      "runtime_gate": "none"
+    },
+    "aitable.shortcut_record_share_links": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "aitable +record-share-links",
+      "runtime_gate": "none"
+    },
+    "aitable.shortcut_resolve_base": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "aitable +resolve-base",
+      "runtime_gate": "none"
+    },
+    "aitable.shortcut_resolve_table": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "aitable +resolve-table",
+      "runtime_gate": "none"
     }
   }
 }

internal/cli/schema_hints/metadata/attendance.json

@@ -428,6 +428,253 @@
       "interface_reason": "Reviewed unpinned remote adapter: this executable CLI wrapper calls a remote helper that is absent from the pinned MCP metadata snapshot; no single pinned semantically equivalent interface_ref can represent the command.",
       "reviewed": true,
       "runtime_gate": "typed_yes"
+    },
+    "attendance.shortcut_check_result": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "attendance +check-result",
+      "runtime_gate": "none"
+    },
+    "attendance.shortcut_check_record": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "attendance +check-record",
+      "runtime_gate": "none"
+    },
+    "attendance.shortcut_list_approve": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "attendance +list-approve",
+      "runtime_gate": "none"
+    },
+    "attendance.shortcut_get_approve_template": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "attendance +get-approve-template",
+      "runtime_gate": "none"
+    },
+    "attendance.shortcut_get_schedule": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "attendance +get-schedule",
+      "runtime_gate": "none"
+    },
+    "attendance.shortcut_search_class": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "attendance +search-class",
+      "runtime_gate": "none"
+    },
+    "attendance.shortcut_get_adjustment_rule": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "attendance +get-adjustment-rule",
+      "runtime_gate": "none"
+    },
+    "attendance.shortcut_search_adjustment_rule": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "attendance +search-adjustment-rule",
+      "runtime_gate": "none"
+    },
+    "attendance.shortcut_get_overtime_rule": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "attendance +get-overtime-rule",
+      "runtime_gate": "none"
+    },
+    "attendance.shortcut_search_overtime_rule": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "attendance +search-overtime-rule",
+      "runtime_gate": "none"
+    },
+    "attendance.shortcut_search_group": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "attendance +search-group",
+      "runtime_gate": "none"
+    },
+    "attendance.shortcut_get_summary": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "attendance +get-summary",
+      "runtime_gate": "none"
+    },
+    "attendance.shortcut_get_self_setting": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "attendance +get-self-setting",
+      "runtime_gate": "none"
+    },
+    "attendance.shortcut_query_report_data": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "attendance +query-report-data",
+      "runtime_gate": "none"
+    },
+    "attendance.shortcut_list_leave_types": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "attendance +list-leave-types",
+      "runtime_gate": "none"
+    },
+    "attendance.shortcut_get_leave_records": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "attendance +get-leave-records",
+      "runtime_gate": "none"
+    },
+    "attendance.shortcut_get_checkin_record": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "attendance +get-checkin-record",
+      "runtime_gate": "none"
+    },
+    "attendance.shortcut_my_attendance": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "attendance +my-attendance",
+      "runtime_gate": "none"
+    },
+    "attendance.shortcut_this_month": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "attendance +this-month",
+      "runtime_gate": "none"
     }
   }
 }

internal/cli/schema_hints/metadata/calendar.json

@@ -119,6 +119,266 @@
       "interface_mode": "mcp",
       "availability": "available",
       "reviewed": true
+    },
+    "calendar.shortcut_agenda": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "calendar +agenda",
+      "runtime_gate": "none"
+    },
+    "calendar.shortcut_attendee_list": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "calendar +attendee-list",
+      "runtime_gate": "none"
+    },
+    "calendar.shortcut_room_search": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "calendar +room-search",
+      "runtime_gate": "none"
+    },
+    "calendar.shortcut_room_groups": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "calendar +room-groups",
+      "runtime_gate": "none"
+    },
+    "calendar.shortcut_freebusy": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "calendar +freebusy",
+      "runtime_gate": "none"
+    },
+    "calendar.shortcut_book_list": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "calendar +book-list",
+      "runtime_gate": "none"
+    },
+    "calendar.shortcut_book_search": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "calendar +book-search",
+      "runtime_gate": "none"
+    },
+    "calendar.shortcut_book": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "calendar +book",
+      "runtime_gate": "typed_yes"
+    },
+    "calendar.shortcut_cancel_event": {
+      "effect": "destructive",
+      "risk": "high",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "calendar +cancel-event",
+      "runtime_gate": "typed_yes"
+    },
+    "calendar.shortcut_conflicts": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "calendar +conflicts",
+      "runtime_gate": "none"
+    },
+    "calendar.shortcut_free_slots": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "calendar +free-slots",
+      "runtime_gate": "none"
+    },
+    "calendar.shortcut_free": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "calendar +free",
+      "runtime_gate": "none"
+    },
+    "calendar.shortcut_invite": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "calendar +invite",
+      "runtime_gate": "typed_yes"
+    },
+    "calendar.shortcut_my_free": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "calendar +my-free",
+      "runtime_gate": "none"
+    },
+    "calendar.shortcut_next_event": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "calendar +next-event",
+      "runtime_gate": "none"
+    },
+    "calendar.shortcut_reschedule": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "calendar +reschedule",
+      "runtime_gate": "typed_yes"
+    },
+    "calendar.shortcut_suggest_time": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "calendar +suggest-time",
+      "runtime_gate": "none"
+    },
+    "calendar.shortcut_today": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "calendar +today",
+      "runtime_gate": "none"
+    },
+    "calendar.shortcut_tomorrow": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "calendar +tomorrow",
+      "runtime_gate": "none"
+    },
+    "calendar.shortcut_week": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "calendar +week",
+      "runtime_gate": "none"
     }
   }
 }

internal/cli/schema_hints/metadata/chat.json

@@ -649,6 +649,703 @@
       "interface_mode": "mcp",
       "availability": "available",
       "reviewed": true
+    },
+    "chat.edit_message": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "not_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed unpinned remote adapter: the executable CLI builds or accepts message content and calls im/edit_message, which is absent from the pinned MCP metadata snapshot.",
+      "reviewed": true,
+      "parameters": {
+        "conversation-id": {
+          "property": "openConversationId",
+          "required": true
+        },
+        "group": {
+          "property": "openConversationId",
+          "required": false
+        },
+        "id": {
+          "property": "openConversationId",
+          "required": false
+        },
+        "chat": {
+          "property": "openConversationId",
+          "required": false
+        },
+        "msg-id": {
+          "property": "openMessageId",
+          "required": true
+        },
+        "text": {
+          "property": "text",
+          "required": false
+        },
+        "title": {
+          "property": "title",
+          "required": false
+        },
+        "content": {
+          "property": "content",
+          "required": false
+        },
+        "at-all": {
+          "property": "atAll",
+          "required": false,
+          "interface_type": "boolean"
+        },
+        "at-open-dingtalk-ids": {
+          "property": "atOpenDingTalkIds",
+          "required": false,
+          "interface_type": "array"
+        }
+      },
+      "review_reason": "The reviewed handler requires one conversation locator, a message ID, and exactly one of text or raw content; it optionally derives a title and maps mention controls before invoking edit_message.",
+      "cli_path": "chat message edit",
+      "runtime_gate": "none"
+    },
+    "chat.get_conv_categories_info": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed unpinned remote adapter: the executable CLI parses category IDs and calls im/get_conv_categories_info, which is absent from the pinned MCP metadata snapshot.",
+      "reviewed": true,
+      "parameters": {
+        "category-ids": {
+          "property": "categoryIds",
+          "required": true,
+          "interface_type": "array"
+        }
+      },
+      "review_reason": "The reviewed handler requires a comma-separated category-ID list, parses it to integers, and invokes get_conv_categories_info without a mutation or confirmation gate.",
+      "cli_path": "chat category batch-info",
+      "runtime_gate": "none"
+    },
+    "chat.list_conv_categories_by_conv": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed unpinned remote adapter: the executable CLI maps a conversation locator to im/list_conv_categories_by_conv, which is absent from the pinned MCP metadata snapshot.",
+      "reviewed": true,
+      "parameters": {
+        "group": {
+          "property": "openConversationId",
+          "required": true
+        },
+        "conversation-id": {
+          "property": "openConversationId",
+          "required": false
+        },
+        "id": {
+          "property": "openConversationId",
+          "required": false
+        }
+      },
+      "review_reason": "The reviewed handler accepts one of three conversation locator aliases and invokes list_conv_categories_by_conv without a mutation or confirmation gate.",
+      "cli_path": "chat category list-by-conv",
+      "runtime_gate": "none"
+    },
+    "chat.update_group_nick": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed unpinned remote adapter: the executable CLI maps nickname update or clear semantics to im/update_group_nick, which is absent from the pinned MCP metadata snapshot.",
+      "reviewed": true,
+      "parameters": {
+        "group": {
+          "property": "openConversationId",
+          "required": true
+        },
+        "nick": {
+          "property": "nick",
+          "required": false
+        }
+      },
+      "review_reason": "The reviewed handler requires the target conversation while intentionally allowing omitted --nick to send an empty nickname and clear the current user's group nickname.",
+      "cli_path": "chat group update-nick",
+      "runtime_gate": "none"
+    },
+    "chat.upgrade_group_to_external": {
+      "effect": "destructive",
+      "risk": "high",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed unpinned remote adapter: the executable CLI validates an optional string map and calls im/upgrade_group_to_external, which is absent from the pinned MCP metadata snapshot.",
+      "reviewed": true,
+      "parameters": {
+        "group": {
+          "property": "openConversationId",
+          "required": true
+        },
+        "extension": {
+          "property": "extension",
+          "required": false,
+          "interface_type": "object"
+        }
+      },
+      "review_reason": "The reviewed handler validates the target group and optional string-valued extension object, permits caller-authoritative dry-run, and requires explicit --yes before the irreversible upgrade.",
+      "cli_path": "chat group upgrade-to-external",
+      "runtime_gate": "confirm_dangerous"
+    },
+    "chat.shortcut_bot_search": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +bot-search",
+      "runtime_gate": "none"
+    },
+    "chat.shortcut_bot_find": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +bot-find",
+      "runtime_gate": "none"
+    },
+    "chat.shortcut_conversation_info": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +conversation-info",
+      "runtime_gate": "none"
+    },
+    "chat.shortcut_conversation_clear_all_red_point": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +conversation-clear-all-red-point",
+      "runtime_gate": "typed_yes"
+    },
+    "chat.shortcut_conversation_list": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +conversation-list",
+      "runtime_gate": "none"
+    },
+    "chat.shortcut_conversation_list_top": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +conversation-list-top",
+      "runtime_gate": "none"
+    },
+    "chat.shortcut_category_list": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +category-list",
+      "runtime_gate": "none"
+    },
+    "chat.shortcut_category_create": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +category-create",
+      "runtime_gate": "typed_yes"
+    },
+    "chat.shortcut_category_delete": {
+      "effect": "destructive",
+      "risk": "high",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +category-delete",
+      "runtime_gate": "typed_yes"
+    },
+    "chat.shortcut_category_rename": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +category-rename",
+      "runtime_gate": "typed_yes"
+    },
+    "chat.shortcut_chat_search": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +chat-search",
+      "runtime_gate": "none"
+    },
+    "chat.shortcut_chat_invite_url": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +chat-invite-url",
+      "runtime_gate": "none"
+    },
+    "chat.shortcut_chat_dismiss": {
+      "effect": "destructive",
+      "risk": "high",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +chat-dismiss",
+      "runtime_gate": "typed_yes"
+    },
+    "chat.shortcut_chat_set_history": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +chat-set-history",
+      "runtime_gate": "typed_yes"
+    },
+    "chat.shortcut_chat_update_nick": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +chat-update-nick",
+      "runtime_gate": "typed_yes"
+    },
+    "chat.shortcut_chat_update_alias": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +chat-update-alias",
+      "runtime_gate": "typed_yes"
+    },
+    "chat.shortcut_chat_list_mine": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +chat-list-mine",
+      "runtime_gate": "none"
+    },
+    "chat.shortcut_chat_list_all": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +chat-list-all",
+      "runtime_gate": "none"
+    },
+    "chat.shortcut_chat_list_join_requests": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +chat-list-join-requests",
+      "runtime_gate": "none"
+    },
+    "chat.shortcut_chat_bots": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +chat-bots",
+      "runtime_gate": "none"
+    },
+    "chat.shortcut_chat_set_admin": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +chat-set-admin",
+      "runtime_gate": "typed_yes"
+    },
+    "chat.shortcut_chat_mute": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +chat-mute",
+      "runtime_gate": "typed_yes"
+    },
+    "chat.shortcut_chat_role_list": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +chat-role-list",
+      "runtime_gate": "none"
+    },
+    "chat.shortcut_chat_role_add": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +chat-role-add",
+      "runtime_gate": "typed_yes"
+    },
+    "chat.shortcut_chat_role_update": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +chat-role-update",
+      "runtime_gate": "typed_yes"
+    },
+    "chat.shortcut_chat_role_set_user": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +chat-role-set-user",
+      "runtime_gate": "typed_yes"
+    },
+    "chat.shortcut_chat_role_query_user": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +chat-role-query-user",
+      "runtime_gate": "none"
+    },
+    "chat.shortcut_messages_send_by_webhook": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +messages-send-by-webhook",
+      "runtime_gate": "typed_yes"
+    },
+    "chat.shortcut_messages_list_direct": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +messages-list-direct",
+      "runtime_gate": "none"
+    },
+    "chat.shortcut_messages_list_unread_conversations": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +messages-list-unread-conversations",
+      "runtime_gate": "none"
+    },
+    "chat.shortcut_messages_mget": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +messages-mget",
+      "runtime_gate": "none"
+    },
+    "chat.shortcut_messages_query_send_status": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +messages-query-send-status",
+      "runtime_gate": "none"
+    },
+    "chat.shortcut_messages_read_status": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +messages-read-status",
+      "runtime_gate": "none"
+    },
+    "chat.shortcut_messages_update_card": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +messages-update-card",
+      "runtime_gate": "typed_yes"
+    },
+    "chat.shortcut_messages_list_pin": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +messages-list-pin",
+      "runtime_gate": "none"
+    },
+    "chat.shortcut_at_me": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +at-me",
+      "runtime_gate": "none"
+    },
+    "chat.shortcut_broadcast": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +broadcast",
+      "runtime_gate": "typed_yes"
+    },
+    "chat.shortcut_dm": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +dm",
+      "runtime_gate": "typed_yes"
+    },
+    "chat.shortcut_group_members": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +group-members",
+      "runtime_gate": "none"
+    },
+    "chat.shortcut_my_groups": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +my-groups",
+      "runtime_gate": "none"
+    },
+    "chat.shortcut_send_to_group": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +send-to-group",
+      "runtime_gate": "typed_yes"
+    },
+    "chat.shortcut_unread_chats": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "chat +unread-chats",
+      "runtime_gate": "none"
     }
   }
 }

internal/cli/schema_hints/metadata/contact.json

@@ -171,6 +171,405 @@
       "interface_mode": "mcp",
       "availability": "available",
       "reviewed": true
+    },
+    "contact.department_create": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "non_idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed unpinned remote adapter: the executable CLI maps department creation flags to contact/department_create, which is absent from the pinned MCP metadata snapshot.",
+      "reviewed": true,
+      "parameters": {
+        "name": {
+          "property": "deptName",
+          "required": true
+        },
+        "dept-name": {
+          "property": "deptName",
+          "required": false
+        },
+        "parent": {
+          "property": "superDeptId",
+          "required": false,
+          "interface_type": "integer"
+        },
+        "super-dept-id": {
+          "property": "superDeptId",
+          "required": false,
+          "interface_type": "integer"
+        },
+        "super-dept": {
+          "property": "superDeptId",
+          "required": false,
+          "interface_type": "integer"
+        },
+        "create-dept-group": {
+          "property": "createDeptGroup",
+          "required": true,
+          "interface_type": "boolean"
+        }
+      },
+      "review_reason": "The reviewed handler requires one department-name spelling and an explicit boolean group choice, optionally parses a parent department ID, confirms, then invokes department_create.",
+      "cli_path": "contact dept create",
+      "runtime_gate": "confirm_dangerous"
+    },
+    "contact.department_update": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed unpinned remote adapter: the executable CLI maps department update flags to contact/department_update, which is absent from the pinned MCP metadata snapshot.",
+      "reviewed": true,
+      "parameters": {
+        "dept": {
+          "property": "deptId",
+          "required": true,
+          "interface_type": "integer"
+        },
+        "id": {
+          "property": "deptId",
+          "required": false,
+          "interface_type": "integer"
+        },
+        "ids": {
+          "property": "deptId",
+          "required": false,
+          "interface_type": "integer"
+        },
+        "dept-id": {
+          "property": "deptId",
+          "required": false,
+          "interface_type": "integer"
+        },
+        "dept-ids": {
+          "property": "deptId",
+          "required": false,
+          "interface_type": "integer"
+        },
+        "name": {
+          "property": "deptName",
+          "required": true
+        },
+        "dept-name": {
+          "property": "deptName",
+          "required": false
+        },
+        "parent": {
+          "property": "superDeptId",
+          "required": false,
+          "interface_type": "integer"
+        },
+        "super-dept-id": {
+          "property": "superDeptId",
+          "required": false,
+          "interface_type": "integer"
+        },
+        "super-dept": {
+          "property": "superDeptId",
+          "required": false,
+          "interface_type": "integer"
+        }
+      },
+      "review_reason": "The reviewed handler requires one department-ID spelling and one department-name spelling, optionally parses a parent ID, confirms, then invokes department_update.",
+      "cli_path": "contact dept update",
+      "runtime_gate": "confirm_dangerous"
+    },
+    "contact.employee_update": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed unpinned remote adapter: the executable CLI maps employee update flags to contact/employee_update, which is absent from the pinned MCP metadata snapshot.",
+      "reviewed": true,
+      "parameters": {
+        "user-id": {
+          "property": "userId",
+          "required": true
+        },
+        "id": {
+          "property": "userId",
+          "required": false
+        },
+        "userid": {
+          "property": "userId",
+          "required": false
+        },
+        "org-user-name": {
+          "property": "orgUserName",
+          "required": false
+        },
+        "depts": {
+          "property": "depts",
+          "required": false,
+          "interface_type": "array"
+        },
+        "master-user-id": {
+          "property": "masterUserId",
+          "required": false
+        }
+      },
+      "review_reason": "The reviewed handler requires one employee-ID spelling plus at least one update field, decodes optional department JSON, confirms, then invokes employee_update.",
+      "cli_path": "contact user update",
+      "runtime_gate": "confirm_dangerous"
+    },
+    "contact.exclusive_account_user_update": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed unpinned remote adapter: the executable CLI maps enterprise-account update flags to contact/exclusive_account_user_update, which is absent from the pinned MCP metadata snapshot.",
+      "reviewed": true,
+      "parameters": {
+        "user-id": {
+          "property": "userId",
+          "required": true
+        },
+        "id": {
+          "property": "userId",
+          "required": false
+        },
+        "userid": {
+          "property": "userId",
+          "required": false
+        },
+        "org-user-name": {
+          "property": "orgUserName",
+          "required": false
+        },
+        "depts": {
+          "property": "depts",
+          "required": false,
+          "interface_type": "array"
+        },
+        "master-user-id": {
+          "property": "masterUserId",
+          "required": false
+        },
+        "nick": {
+          "property": "nick",
+          "required": false
+        },
+        "avatar-file-id": {
+          "property": "avatarFileId",
+          "required": false
+        }
+      },
+      "review_reason": "The reviewed handler requires one enterprise-account user ID plus at least one profile or organization update field, confirms, then invokes exclusive_account_user_update.",
+      "cli_path": "contact account update",
+      "runtime_gate": "confirm_dangerous"
+    },
+    "contact.self_user_profile_update": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed unpinned remote adapter: the executable CLI maps self-profile update flags to contact/self_user_profile_update, which is absent from the pinned MCP metadata snapshot.",
+      "reviewed": true,
+      "parameters": {
+        "nick": {
+          "property": "nick",
+          "required": false
+        },
+        "avatar-file-id": {
+          "property": "avatarFileId",
+          "required": false
+        }
+      },
+      "review_reason": "The reviewed handler requires at least one of nickname or avatar file ID, confirms, then invokes self_user_profile_update for the current user.",
+      "cli_path": "contact user update-self",
+      "runtime_gate": "confirm_dangerous"
+    },
+    "contact.shortcut_list_followings": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "contact +list-followings",
+      "runtime_gate": "none"
+    },
+    "contact.shortcut_search_user": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "contact +search-user",
+      "runtime_gate": "none"
+    },
+    "contact.shortcut_search_mobile": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "contact +search-mobile",
+      "runtime_gate": "none"
+    },
+    "contact.shortcut_list_roles": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "contact +list-roles",
+      "runtime_gate": "none"
+    },
+    "contact.shortcut_list_role_members": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "contact +list-role-members",
+      "runtime_gate": "none"
+    },
+    "contact.shortcut_list_sub_depts": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "contact +list-sub-depts",
+      "runtime_gate": "none"
+    },
+    "contact.shortcut_list_dept_members": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "contact +list-dept-members",
+      "runtime_gate": "none"
+    },
+    "contact.shortcut_by_mobile": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "contact +by-mobile",
+      "runtime_gate": "none"
+    },
+    "contact.shortcut_dept_members": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "contact +dept-members",
+      "runtime_gate": "none"
+    },
+    "contact.shortcut_lookup": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "contact +lookup",
+      "runtime_gate": "none"
+    },
+    "contact.shortcut_org": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "contact +org",
+      "runtime_gate": "none"
+    },
+    "contact.shortcut_resolve_dept": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "contact +resolve-dept",
+      "runtime_gate": "none"
+    },
+    "contact.shortcut_team": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "contact +team",
+      "runtime_gate": "none"
+    },
+    "contact.shortcut_me": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "contact +me",
+      "runtime_gate": "none"
     }
   }
 }

internal/cli/schema_hints/metadata/devapp.json

@@ -0,0 +1,259 @@
+{
+  "version": 1,
+  "source": {
+    "kind": "explicit",
+    "name": "dws-tool-metadata/devapp",
+    "repository": "DingTalk-Real-AI/dingtalk-workspace-cli",
+    "channel": "open-source",
+    "reviewed": true
+  },
+  "tools": {
+    "devapp.shortcut_list": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "devapp +list",
+      "runtime_gate": "none"
+    },
+    "devapp.shortcut_get": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "devapp +get",
+      "runtime_gate": "none"
+    },
+    "devapp.shortcut_create": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "devapp +create",
+      "runtime_gate": "typed_yes"
+    },
+    "devapp.shortcut_update": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "devapp +update",
+      "runtime_gate": "typed_yes"
+    },
+    "devapp.shortcut_delete": {
+      "effect": "destructive",
+      "risk": "high",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "devapp +delete",
+      "runtime_gate": "typed_yes"
+    },
+    "devapp.shortcut_enable": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "devapp +enable",
+      "runtime_gate": "typed_yes"
+    },
+    "devapp.shortcut_disable": {
+      "effect": "destructive",
+      "risk": "high",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "devapp +disable",
+      "runtime_gate": "typed_yes"
+    },
+    "devapp.shortcut_webapp_get": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "devapp +webapp-get",
+      "runtime_gate": "none"
+    },
+    "devapp.shortcut_webapp_config": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "devapp +webapp-config",
+      "runtime_gate": "typed_yes"
+    },
+    "devapp.shortcut_permission_list": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "devapp +permission-list",
+      "runtime_gate": "none"
+    },
+    "devapp.shortcut_member_list": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "devapp +member-list",
+      "runtime_gate": "none"
+    },
+    "devapp.shortcut_member_add": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "devapp +member-add",
+      "runtime_gate": "typed_yes"
+    },
+    "devapp.shortcut_member_remove": {
+      "effect": "destructive",
+      "risk": "high",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "devapp +member-remove",
+      "runtime_gate": "typed_yes"
+    },
+    "devapp.shortcut_robot_get": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "devapp +robot-get",
+      "runtime_gate": "none"
+    },
+    "devapp.shortcut_event_list": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "devapp +event-list",
+      "runtime_gate": "none"
+    },
+    "devapp.shortcut_version_list": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "devapp +version-list",
+      "runtime_gate": "none"
+    },
+    "devapp.shortcut_version_get": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "devapp +version-get",
+      "runtime_gate": "none"
+    },
+    "devapp.shortcut_version_check_approval": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "devapp +version-check-approval",
+      "runtime_gate": "none"
+    },
+    "devapp.shortcut_version_status": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "devapp +version-status",
+      "runtime_gate": "none"
+    }
+  }
+}

internal/cli/schema_hints/metadata/ding.json

@@ -21,6 +21,58 @@
       "interface_mode": "mcp",
       "availability": "available",
       "reviewed": true
+    },
+    "ding.shortcut_list": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "ding +list",
+      "runtime_gate": "none"
+    },
+    "ding.shortcut_receiver_status": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "ding +receiver-status",
+      "runtime_gate": "none"
+    },
+    "ding.shortcut_send_personal": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "ding +send-personal",
+      "runtime_gate": "typed_yes"
+    },
+    "ding.shortcut_recall_personal": {
+      "effect": "destructive",
+      "risk": "high",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "ding +recall-personal",
+      "runtime_gate": "typed_yes"
     }
   }
 }

internal/cli/schema_hints/metadata/doc.json

@@ -19,13 +19,26 @@
       "reviewed": true
     },
     "doc.create_comment": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "not_required",
+      "idempotency": "unknown",
       "interface_ref": {
         "product_id": "doc-comment",
         "rpc_name": "create_comment"
       },
       "interface_mode": "mcp",
       "availability": "available",
-      "reviewed": true
+      "reviewed": true,
+      "parameters": {
+        "mentioned-open-conversation-id": {
+          "required": false,
+          "interface_type": "array"
+        }
+      },
+      "review_reason": "The reviewed handler normalizes repeated or comma-separated group conversation IDs into a stable de-duplicated array while preserving existing user mentions.",
+      "cli_path": "doc comment create",
+      "runtime_gate": "none"
     },
     "doc.create_document": {
       "interface_mode": "mcp",
@@ -91,9 +104,44 @@
       "reviewed": true
     },
     "doc.get_document_content": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_ref": {
+        "product_id": "doc",
+        "rpc_name": "get_document_content"
+      },
       "interface_mode": "mcp",
       "availability": "available",
-      "reviewed": true
+      "reviewed": true,
+      "parameters": {
+        "content-format": {
+          "property": "format",
+          "required": false
+        },
+        "scope": {
+          "required": false
+        },
+        "tags": {
+          "required": false,
+          "required_when": "--scope=tags"
+        },
+        "max-depth": {
+          "required": false,
+          "interface_type": "integer"
+        },
+        "start-block-id": {
+          "required": false,
+          "required_when": "--scope=range or --scope=section"
+        },
+        "end-block-id": {
+          "required": false
+        }
+      },
+      "review_reason": "The reviewed handler preserves full Markdown/JSONML reads and adds validated outline, range, section, and tag-scoped JSONML fragment reads with optional file output.",
+      "cli_path": "doc read",
+      "runtime_gate": "none"
     },
     "doc.get_document_info": {
       "interface_mode": "mcp",
@@ -173,13 +221,26 @@
       "reviewed": true
     },
     "doc.reply_comment": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "not_required",
+      "idempotency": "unknown",
       "interface_ref": {
         "product_id": "doc-comment",
         "rpc_name": "reply_comment"
       },
       "interface_mode": "mcp",
       "availability": "available",
-      "reviewed": true
+      "reviewed": true,
+      "parameters": {
+        "mentioned-open-conversation-id": {
+          "required": false,
+          "interface_type": "array"
+        }
+      },
+      "review_reason": "The reviewed handler normalizes group conversation IDs into a stable de-duplicated array and rejects group mentions for emoji replies, which the remote operation does not support.",
+      "cli_path": "doc comment reply",
+      "runtime_gate": "none"
     },
     "doc.search_documents": {
       "interface_mode": "mcp",
@@ -220,6 +281,10 @@
       "review_reason": "migrated to reviewed metadata block"
     },
     "doc.update_comment": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "not_required",
+      "idempotency": "unknown",
       "interface_mode": "composite",
       "availability": "available",
       "interface_reason": "Reviewed unpinned remote adapter: this executable CLI wrapper calls a remote helper that is absent from the pinned MCP metadata snapshot; no single pinned semantically equivalent interface_ref can represent the command.",
@@ -227,12 +292,16 @@
       "parameters": {
         "mention": {
           "interface_type": "array"
+        },
+        "mentioned-open-conversation-id": {
+          "property": "mentionedOpenConversationIds",
+          "required": false,
+          "interface_type": "array"
         }
       },
-      "review_reason": "Preserve the reviewed CLI-to-interface conversion: the comma-separated --mention string is normalized to the mentionedUserIds array.",
+      "review_reason": "Preserve the reviewed user-mention conversion and normalize repeated or comma-separated group conversation IDs into mentionedOpenConversationIds.",
       "cli_path": "doc comment update",
-      "runtime_gate": "none",
-      "confirmation": "not_required"
+      "runtime_gate": "none"
     },
     "doc.update_document": {
       "effect": "write",
@@ -297,6 +366,227 @@
       "interface_reason": "Reviewed unpinned remote adapter: this executable CLI wrapper calls a remote helper that is absent from the pinned MCP metadata snapshot; no single pinned semantically equivalent interface_ref can represent the command.",
       "reviewed": true,
       "review_reason": "migrated to reviewed metadata block"
+    },
+    "doc.shortcut_search": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "doc +search",
+      "runtime_gate": "none"
+    },
+    "doc.shortcut_list": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "doc +list",
+      "runtime_gate": "none"
+    },
+    "doc.shortcut_copy": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "doc +copy",
+      "runtime_gate": "typed_yes"
+    },
+    "doc.shortcut_move": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "doc +move",
+      "runtime_gate": "typed_yes"
+    },
+    "doc.shortcut_comment_list": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "doc +comment-list",
+      "runtime_gate": "none"
+    },
+    "doc.shortcut_comment_create": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "doc +comment-create",
+      "runtime_gate": "typed_yes"
+    },
+    "doc.shortcut_comment_reply": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "doc +comment-reply",
+      "runtime_gate": "typed_yes"
+    },
+    "doc.shortcut_export_submit": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "doc +export-submit",
+      "runtime_gate": "none"
+    },
+    "doc.shortcut_export_get": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "doc +export-get",
+      "runtime_gate": "none"
+    },
+    "doc.shortcut_version_save": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "doc +version-save",
+      "runtime_gate": "typed_yes"
+    },
+    "doc.shortcut_version_list": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "doc +version-list",
+      "runtime_gate": "none"
+    },
+    "doc.shortcut_version_revert": {
+      "effect": "destructive",
+      "risk": "high",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "doc +version-revert",
+      "runtime_gate": "typed_yes"
+    },
+    "doc.shortcut_template_list": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "doc +template-list",
+      "runtime_gate": "none"
+    },
+    "doc.shortcut_template_search": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "doc +template-search",
+      "runtime_gate": "none"
+    },
+    "doc.shortcut_doc_append": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "doc +doc-append",
+      "runtime_gate": "typed_yes"
+    },
+    "doc.shortcut_find_doc": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "doc +find-doc",
+      "runtime_gate": "none"
+    },
+    "doc.shortcut_share_doc": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "doc +share-doc",
+      "runtime_gate": "typed_yes"
     }
   }
 }

internal/cli/schema_hints/metadata/drive.json

@@ -213,10 +213,112 @@
     "drive.upload": {
       "effect": "write",
       "risk": "medium",
+      "confirmation": "not_required",
+      "idempotency": "unknown",
       "interface_mode": "composite",
       "availability": "available",
       "interface_reason": "命令包含多个 RPC、条件分派或本地 HTTP/文件步骤，不能绑定为单一 interface_ref",
-      "reviewed": true
+      "reviewed": true,
+      "parameters": {
+        "node": {
+          "property": "nodeId",
+          "required": false
+        }
+      },
+      "review_reason": "The reviewed helper preserves the stable command-level Schema contract for ordinary uploads; --node switches to overwrite mode and the runtime still requires explicit confirmation unless --yes is supplied.",
+      "cli_path": "drive upload",
+      "runtime_gate": "none"
+    },
+    "drive.shortcut_info": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "drive +info",
+      "runtime_gate": "none"
+    },
+    "drive.shortcut_search": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "drive +search",
+      "runtime_gate": "none"
+    },
+    "drive.shortcut_search_docs": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "drive +search-docs",
+      "runtime_gate": "none"
+    },
+    "drive.shortcut_copy": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "drive +copy",
+      "runtime_gate": "typed_yes"
+    },
+    "drive.shortcut_move": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "drive +move",
+      "runtime_gate": "typed_yes"
+    },
+    "drive.shortcut_recent": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "drive +recent",
+      "runtime_gate": "none"
+    },
+    "drive.shortcut_find_file": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "drive +find-file",
+      "runtime_gate": "none"
     }
   }
 }

internal/cli/schema_hints/metadata/hrbrain.json

@@ -0,0 +1,289 @@
+{
+  "version": 1,
+  "source": {
+    "kind": "explicit",
+    "name": "dws-tool-metadata/hrbrain",
+    "repository": "DingTalk-Real-AI/dingtalk-workspace-cli",
+    "channel": "open-source",
+    "reviewed": true
+  },
+  "tools": {
+    "hrbrain.list_talent_pools": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "runtime_gate": "none",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed unpinned remote adapter: this executable CLI wrapper calls the hrbrain MCP tool list_talent_pools, which is absent from the pinned MCP metadata snapshot; no single pinned interface_ref can represent the command.",
+      "reviewed": true,
+      "cli_path": "hrbrain talent-pool list",
+      "review_reason": "Reviewed unpinned remote adapter: flags map 1:1 to the list_talent_pools MCP call arguments observed in internal/helpers/hrbrain.go.",
+      "parameters": {
+        "keyword": {
+          "property": "keyword",
+          "required": false
+        },
+        "pool-type": {
+          "property": "poolType",
+          "required": false
+        },
+        "creator": {
+          "property": "creator",
+          "required": false
+        },
+        "labels": {
+          "property": "labels",
+          "required": false
+        },
+        "page": {
+          "property": "currentPage",
+          "required": false
+        },
+        "page-size": {
+          "property": "pageSize",
+          "required": false
+        }
+      }
+    },
+    "hrbrain.get_talent_pool_detail": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "runtime_gate": "none",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed unpinned remote adapter: this executable CLI wrapper calls the hrbrain MCP tool get_talent_pool_detail, which is absent from the pinned MCP metadata snapshot; no single pinned interface_ref can represent the command.",
+      "reviewed": true,
+      "cli_path": "hrbrain talent-pool detail",
+      "review_reason": "Reviewed unpinned remote adapter: --pool-code maps 1:1 to the poolCode MCP call argument observed in internal/helpers/hrbrain.go.",
+      "parameters": {
+        "pool-code": {
+          "property": "poolCode",
+          "required": true
+        }
+      }
+    },
+    "hrbrain.list_pool_employees": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "runtime_gate": "none",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed unpinned remote adapter: this executable CLI wrapper calls the hrbrain MCP tool list_pool_employees, which is absent from the pinned MCP metadata snapshot; no single pinned interface_ref can represent the command.",
+      "reviewed": true,
+      "cli_path": "hrbrain talent-pool employees",
+      "review_reason": "Reviewed unpinned remote adapter: flags map 1:1 to the list_pool_employees MCP call arguments observed in internal/helpers/hrbrain.go.",
+      "parameters": {
+        "pool-code": {
+          "property": "poolCode",
+          "required": true
+        },
+        "page": {
+          "property": "currentPage",
+          "required": false
+        },
+        "page-size": {
+          "property": "pageSize",
+          "required": false
+        }
+      }
+    },
+    "hrbrain.get_profile_metadata": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "runtime_gate": "none",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed unpinned remote adapter: this executable CLI wrapper calls the hrbrain MCP tool get_profile_metadata, which is absent from the pinned MCP metadata snapshot; no single pinned interface_ref can represent the command.",
+      "reviewed": true,
+      "cli_path": "hrbrain profile metadata",
+      "review_reason": "Reviewed unpinned remote adapter: --work-no maps 1:1 to the workNo MCP call argument observed in internal/helpers/hrbrain.go.",
+      "parameters": {
+        "work-no": {
+          "property": "workNo",
+          "required": true
+        }
+      }
+    },
+    "hrbrain.query_profile_data": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "runtime_gate": "none",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed unpinned remote adapter: this executable CLI wrapper calls the hrbrain MCP tool query_profile_data, which is absent from the pinned MCP metadata snapshot; no single pinned interface_ref can represent the command.",
+      "reviewed": true,
+      "cli_path": "hrbrain profile query",
+      "review_reason": "Reviewed unpinned remote adapter: flags map 1:1 to the query_profile_data MCP call arguments observed in internal/helpers/hrbrain.go.",
+      "parameters": {
+        "work-no": {
+          "property": "workNo",
+          "required": true
+        },
+        "data-queries": {
+          "property": "dataQueries",
+          "required": true
+        }
+      }
+    },
+    "hrbrain.get_profile_label": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "runtime_gate": "none",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed unpinned remote adapter: this executable CLI wrapper calls the hrbrain MCP tool get_profile_label, which is absent from the pinned MCP metadata snapshot; no single pinned interface_ref can represent the command.",
+      "reviewed": true,
+      "cli_path": "hrbrain profile labels",
+      "review_reason": "Reviewed unpinned remote adapter: flags map 1:1 to the get_profile_label MCP call arguments observed in internal/helpers/hrbrain.go.",
+      "parameters": {
+        "staff-ids": {
+          "property": "staffIds",
+          "required": true
+        },
+        "all-label": {
+          "property": "allLabel",
+          "required": false
+        }
+      }
+    },
+    "hrbrain.get_employee_career": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "runtime_gate": "none",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed unpinned remote adapter: this executable CLI wrapper calls the hrbrain MCP tool get_employee_career, which is absent from the pinned MCP metadata snapshot; no single pinned interface_ref can represent the command.",
+      "reviewed": true,
+      "cli_path": "hrbrain profile career",
+      "review_reason": "Reviewed unpinned remote adapter: --work-no maps 1:1 to the workNo MCP call argument observed in internal/helpers/hrbrain.go.",
+      "parameters": {
+        "work-no": {
+          "property": "workNo",
+          "required": true
+        }
+      }
+    },
+    "hrbrain.get_employee_performance": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "runtime_gate": "none",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed unpinned remote adapter: this executable CLI wrapper calls the hrbrain MCP tool get_employee_performance, which is absent from the pinned MCP metadata snapshot; no single pinned interface_ref can represent the command.",
+      "reviewed": true,
+      "cli_path": "hrbrain profile performance",
+      "review_reason": "Reviewed unpinned remote adapter: --work-no maps 1:1 to the workNo MCP call argument observed in internal/helpers/hrbrain.go.",
+      "parameters": {
+        "work-no": {
+          "property": "workNo",
+          "required": true
+        }
+      }
+    },
+    "hrbrain.search_employees": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "runtime_gate": "none",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed unpinned remote adapter: this executable CLI wrapper calls the hrbrain MCP tool search_employees, which is absent from the pinned MCP metadata snapshot; no single pinned interface_ref can represent the command.",
+      "reviewed": true,
+      "cli_path": "hrbrain search employees",
+      "review_reason": "Reviewed unpinned remote adapter: flags map 1:1 to the search_employees MCP call arguments observed in internal/helpers/hrbrain.go.",
+      "parameters": {
+        "keyword": {
+          "property": "keyword",
+          "required": false
+        },
+        "dept-name": {
+          "property": "deptName",
+          "required": false
+        },
+        "position-name": {
+          "property": "positionName",
+          "required": false
+        },
+        "job-level": {
+          "property": "jobLevel",
+          "required": false
+        },
+        "pool-code": {
+          "property": "poolCode",
+          "required": false
+        },
+        "page": {
+          "property": "currentPage",
+          "required": false
+        },
+        "page-size": {
+          "property": "pageSize",
+          "required": false
+        }
+      }
+    },
+    "hrbrain.search_employees_structured": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "runtime_gate": "none",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed unpinned remote adapter: this executable CLI wrapper calls the hrbrain MCP tool search_employees_structured, which is absent from the pinned MCP metadata snapshot; no single pinned interface_ref can represent the command.",
+      "reviewed": true,
+      "cli_path": "hrbrain search employees-structured",
+      "review_reason": "Reviewed unpinned remote adapter: flags map 1:1 to the search_employees_structured MCP call arguments observed in internal/helpers/hrbrain.go.",
+      "parameters": {
+        "origin-json": {
+          "property": "originJson",
+          "required": true
+        },
+        "fields": {
+          "property": "fields",
+          "required": true
+        },
+        "order-by": {
+          "property": "orderByClauses",
+          "required": false
+        },
+        "page": {
+          "property": "currentPage",
+          "required": false
+        },
+        "page-size": {
+          "property": "pageSize",
+          "required": false
+        }
+      }
+    },
+    "hrbrain.get_search_fields": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "runtime_gate": "none",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed unpinned remote adapter: this executable CLI wrapper calls the hrbrain MCP tool get_search_fields, which is absent from the pinned MCP metadata snapshot; no single pinned interface_ref can represent the command.",
+      "reviewed": true
+    }
+  }
+}

internal/cli/schema_hints/metadata/mail.json

@@ -218,6 +218,136 @@
       "cli_path": "mail template update",
       "runtime_gate": "none",
       "confirmation": "not_required"
+    },
+    "mail.shortcut_thread_list": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "mail +thread-list",
+      "runtime_gate": "none"
+    },
+    "mail.shortcut_folder_list": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "mail +folder-list",
+      "runtime_gate": "none"
+    },
+    "mail.shortcut_tag_list": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "mail +tag-list",
+      "runtime_gate": "none"
+    },
+    "mail.shortcut_user_search": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "mail +user-search",
+      "runtime_gate": "none"
+    },
+    "mail.shortcut_template_list": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "mail +template-list",
+      "runtime_gate": "none"
+    },
+    "mail.shortcut_contact_list": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "mail +contact-list",
+      "runtime_gate": "none"
+    },
+    "mail.shortcut_find_mail_user": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "mail +find-mail-user",
+      "runtime_gate": "none"
+    },
+    "mail.shortcut_recent_mail": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "mail +recent-mail",
+      "runtime_gate": "none"
+    },
+    "mail.shortcut_search_mail": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "mail +search-mail",
+      "runtime_gate": "none"
+    },
+    "mail.shortcut_unread_mail": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "mail +unread-mail",
+      "runtime_gate": "none"
     }
   }
 }

internal/cli/schema_hints/metadata/markdown.json

@@ -0,0 +1,176 @@
+{
+  "version": 1,
+  "source": {
+    "kind": "explicit",
+    "name": "dws-tool-metadata/markdown",
+    "repository": "DingTalk-Real-AI/dingtalk-workspace-cli",
+    "channel": "open-source",
+    "reviewed": true
+  },
+  "tools": {
+    "markdown.create": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "not_required",
+      "idempotency": "non_idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed cross-product adapter: this local workflow resolves content, validates a native .md file, and uploads through either Drive or Doc space; no single MCP interface represents the command.",
+      "reviewed": true,
+      "parameters": {
+        "name": {
+          "property": "fileName",
+          "required": false,
+          "required_when": "--content is used"
+        },
+        "content": {
+          "property": "content",
+          "required": false
+        },
+        "file": {
+          "property": "filePath",
+          "required": false
+        },
+        "folder": {
+          "property": "folderId",
+          "required": false
+        },
+        "workspace": {
+          "property": "workspaceId",
+          "required": false
+        },
+        "space-id": {
+          "property": "spaceId",
+          "required": false
+        }
+      },
+      "review_reason": "The reviewed workflow requires exactly one content source, requires an .md name for literal content, keeps workspace and Drive space mutually exclusive, and routes the upload without a confirmation gate.",
+      "cli_path": "markdown create",
+      "runtime_gate": "none"
+    },
+    "markdown.fetch": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed cross-product adapter: this local workflow resolves the file domain, downloads through Drive or Doc space, and optionally writes a sanitized local output path; no single MCP interface represents the command.",
+      "reviewed": true,
+      "parameters": {
+        "node": {
+          "property": "nodeId",
+          "required": true
+        },
+        "id": {
+          "property": "nodeId",
+          "required": false
+        },
+        "space-id": {
+          "property": "spaceId",
+          "required": false
+        },
+        "workspace": {
+          "property": "workspaceId",
+          "required": false
+        },
+        "output": {
+          "property": "output",
+          "required": false
+        }
+      },
+      "review_reason": "The reviewed workflow requires one node locator, keeps explicit Drive and Doc routes mutually exclusive, treats remote content as untrusted data, and protects local output paths.",
+      "cli_path": "markdown fetch",
+      "runtime_gate": "none"
+    },
+    "markdown.overwrite": {
+      "effect": "destructive",
+      "risk": "high",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed cross-product adapter: this local workflow resolves and previews existing content, then replaces a Drive or Doc-space native .md file; no single MCP interface represents the command.",
+      "reviewed": true,
+      "parameters": {
+        "node": {
+          "property": "nodeId",
+          "required": true
+        },
+        "content": {
+          "property": "content",
+          "required": false
+        },
+        "file": {
+          "property": "filePath",
+          "required": false
+        },
+        "name": {
+          "property": "fileName",
+          "required": false
+        },
+        "space-id": {
+          "property": "spaceId",
+          "required": false
+        },
+        "workspace": {
+          "property": "workspaceId",
+          "required": false
+        },
+        "dry-run": {
+          "property": "dryRun",
+          "required": false,
+          "interface_type": "boolean"
+        }
+      },
+      "review_reason": "The reviewed workflow requires a target and exactly one content source, supports a command-owned diff preview, and confirms before replacing the remote file.",
+      "cli_path": "markdown overwrite",
+      "runtime_gate": "confirm_dangerous"
+    },
+    "markdown.patch": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed cross-product adapter: this local workflow downloads a Drive or Doc-space native .md file, applies literal or RE2 replacement, and reuploads it; no single MCP interface represents the command.",
+      "reviewed": true,
+      "parameters": {
+        "node": {
+          "property": "nodeId",
+          "required": true
+        },
+        "pattern": {
+          "property": "pattern",
+          "required": true
+        },
+        "content": {
+          "property": "replacement",
+          "required": true
+        },
+        "regex": {
+          "property": "regex",
+          "required": false,
+          "interface_type": "boolean"
+        },
+        "space-id": {
+          "property": "spaceId",
+          "required": false
+        },
+        "workspace": {
+          "property": "workspaceId",
+          "required": false
+        },
+        "dry-run": {
+          "property": "dryRun",
+          "required": false,
+          "interface_type": "boolean"
+        }
+      },
+      "review_reason": "The reviewed workflow requires a target, pattern, and replacement; it blocks zero-hit and empty-result writes, supports a command-owned diff preview, and confirms before reupload.",
+      "cli_path": "markdown patch",
+      "runtime_gate": "confirm_dangerous"
+    }
+  }
+}

internal/cli/schema_hints/metadata/mcp.json

@@ -0,0 +1,24 @@
+{
+  "version": 1,
+  "source": {
+    "kind": "explicit",
+    "name": "dws-tool-metadata/mcp",
+    "repository": "DingTalk-Real-AI/dingtalk-workspace-cli",
+    "channel": "open-source",
+    "reviewed": true
+  },
+  "tools": {
+    "mcp.url_get": {
+      "effect": "read",
+      "risk": "medium",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed unpinned remote adapter: the public CLI wrapper calls the helper-only mcp-meta/get_mcp_server_url endpoint, which is intentionally absent from the public product catalog and pinned MCP metadata.",
+      "runtime_gate": "none",
+      "reviewed": true,
+      "review_reason": "Expose the reviewed MCP URL resolver through the public CLI and Agent Schema while preserving the helper endpoint as a non-product supplemental server. The returned URL contains user- and organization-scoped credentials and must be treated as sensitive output."
+    }
+  }
+}

internal/cli/schema_hints/metadata/minutes.json

@@ -194,6 +194,89 @@
       "interface_mode": "mcp",
       "availability": "available",
       "reviewed": true
+    },
+    "minutes.shortcut_list_mine": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "minutes +list-mine",
+      "runtime_gate": "none"
+    },
+    "minutes.shortcut_list_shared": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "minutes +list-shared",
+      "runtime_gate": "none"
+    },
+    "minutes.shortcut_list_all": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "minutes +list-all",
+      "runtime_gate": "none"
+    },
+    "minutes.shortcut_record_start": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "minutes +record-start",
+      "runtime_gate": "typed_yes"
+    },
+    "minutes.shortcut_detail": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "minutes +detail",
+      "runtime_gate": "none"
+    },
+    "minutes.shortcut_replace_batch": {
+      "effect": "write",
+      "risk": "medium",
+      "confirmation": "user_required",
+      "idempotency": "unknown",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "parameters": {
+        "pair": {
+          "description": "替换规则，格式 \"原文=>替换\"，可重复传多组（必填）；每组原文不能为空且不能重复"
+        }
+      },
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate, composite interface, and the custom --pair format/uniqueness validation without inventing a direct MCP identity.",
+      "cli_path": "minutes +replace-batch",
+      "runtime_gate": "typed_yes"
     }
   }
 }

internal/cli/schema_hints/metadata/oa.json

@@ -90,6 +90,97 @@
       "availability": "available",
       "reviewed": true,
       "runtime_gate": "confirm_delete"
+    },
+    "oa.shortcut_list_pending": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "oa +list-pending",
+      "runtime_gate": "none"
+    },
+    "oa.shortcut_list_forms": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "oa +list-forms",
+      "runtime_gate": "none"
+    },
+    "oa.shortcut_search_forms": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "oa +search-forms",
+      "runtime_gate": "none"
+    },
+    "oa.shortcut_list_executed": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "oa +list-executed",
+      "runtime_gate": "none"
+    },
+    "oa.shortcut_list_submitted": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "oa +list-submitted",
+      "runtime_gate": "none"
+    },
+    "oa.shortcut_list_cc": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "oa +list-cc",
+      "runtime_gate": "none"
+    },
+    "oa.shortcut_my_initiated": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "oa +my-initiated",
+      "runtime_gate": "none"
     }
   }
 }

internal/cli/schema_hints/metadata/report.json

@@ -51,6 +51,32 @@
       "interface_mode": "mcp",
       "availability": "available",
       "reviewed": true
+    },
+    "report.shortcut_inbox_list": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "report +inbox-list",
+      "runtime_gate": "none"
+    },
+    "report.shortcut_outbox_list": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "report +outbox-list",
+      "runtime_gate": "none"
     }
   }
 }

internal/cli/schema_hints/metadata/sheet.json

@@ -708,6 +708,32 @@
       "interface_mode": "mcp",
       "availability": "available",
       "reviewed": true
+    },
+    "sheet.shortcut_list_sheets": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "sheet +list-sheets",
+      "runtime_gate": "none"
+    },
+    "sheet.shortcut_read": {
+      "effect": "read",
+      "risk": "low",
+      "confirmation": "not_required",
+      "idempotency": "idempotent",
+      "interface_mode": "composite",
+      "availability": "available",
+      "interface_reason": "Reviewed built-in shortcut adapter: the executable CLI owns validation, optional multi-step orchestration, output projection, and confirmation; the complete command contract is not represented by one pinned MCP interface_ref.",
+      "reviewed": true,
+      "review_reason": "Reviewed against the built-in Shortcut registry and mounted Cobra command: publishes the executable risk gate and composite interface without inventing a direct MCP identity.",
+      "cli_path": "sheet +read",
+      "runtime_gate": "none"
     }
   }
 }