fix(ci): enforce stable Schema compatibility

ci(schema): allow reviewed parameter type migrations

.github/reviewer-routing.js

@@ -0,0 +1,285 @@
+'use strict';
+
+// 评审归属是受保护分支上的声明式规则；未知路径不猜测，交给工作流负载均衡兜底。
+const REVIEWER_POOL = ['wxianfeng', 'typefield', 'haofeng0705', 'hlzjsong'];
+
+const PRODUCT_GROUPS = [
+  {
+    primary: 'wxianfeng',
+    backup: 'typefield',
+    products: ['chat', 'contact', 'ding', 'event', 'mail', 'live', 'conference', 'dev', 'devapp', 'mcp', 'aiapp'],
+  },
+  {
+    primary: 'typefield',
+    backup: 'wxianfeng',
+    products: ['doc', 'drive', 'wiki', 'markdown', 'docparse', 'aidesign', 'devdoc', 'blackboard', 'finance', 'law', 'credit'],
+  },
+  {
+    primary: 'haofeng0705',
+    backup: 'typefield',
+    products: ['minutes', 'sheet', 'aitable', 'calendar', 'todo', 'oa', 'attendance', 'report', 'agoal', 'aisearch', 'yida', 'hrbrain'],
+  },
+];
+
+const pathStartsWith = (prefixes) => (path) => prefixes.some((prefix) => path.startsWith(prefix));
+
+const MODULES = [
+  {
+    id: 'security',
+    label: '登录、认证、权限、安全',
+    primary: 'hlzjsong',
+    backup: 'typefield',
+    requiresSecondary: true,
+    matches: pathStartsWith([
+      'internal/auth/',
+      'internal/keychain/',
+      'internal/audit/',
+      'internal/pat/',
+      'internal/security/',
+      'internal/safety/',
+      'pkg/edition/',
+    ]),
+  },
+  {
+    id: 'delivery',
+    label: 'CI、测试、发布、安装',
+    primary: 'haofeng0705',
+    backup: 'wxianfeng',
+    requiresSecondary: true,
+    matches: (path) =>
+      path.startsWith('.github/') ||
+      path.startsWith('scripts/release/') ||
+      path.startsWith('scripts/policy/') ||
+      path.startsWith('scripts/dev/') ||
+      path.startsWith('scripts/install') ||
+      path.startsWith('Formula/') ||
+      path.startsWith('build/') ||
+      path.startsWith('internal/upgrade/') ||
+      path.startsWith('internal/app/upgrade') ||
+      path.startsWith('test/') ||
+      path.startsWith('verify/') ||
+      path.startsWith('.workflow/') ||
+      path === 'coverage.txt' ||
+      path === 'coverage-base.txt' ||
+      path === '.goreleaser.yaml' ||
+      path === 'package.json' ||
+      path === 'package-lock.json' ||
+      path === 'docs/releasing.md',
+  },
+  {
+    id: 'architecture',
+    label: 'DWS 架构、公共内核',
+    primary: 'wxianfeng',
+    backup: 'typefield',
+    requiresSecondary: true,
+    matches: pathStartsWith([
+      'cmd/',
+      'internal/apiclient/',
+      'internal/app/',
+      'internal/cli/',
+      'internal/cobracmd/',
+      'internal/corecmd/',
+      'internal/errors/',
+      'internal/executor/',
+      'internal/generator/',
+      'internal/i18n/',
+      'internal/interfacesnapshot/',
+      'internal/jsonutil/',
+      'internal/localio/',
+      'internal/logging/',
+      'internal/output/',
+      'internal/pipeline/',
+      'internal/plugin/',
+      'internal/profilectx/',
+      'internal/registry/',
+      'internal/syncdata/',
+      'internal/testseam/',
+      'internal/transport/',
+      'pkg/',
+    ]),
+  },
+  {
+    id: 'compatibility',
+    label: '兼容性',
+    primary: 'wxianfeng',
+    backup: 'typefield',
+    requiresSecondary: true,
+    matches: (path) =>
+      /(?:^|[/_.-])compat(?:ibility)?(?=$|[/_.-])/.test(path) ||
+      path.includes('schema_compat'),
+  },
+];
+
+function productMatches(path, product) {
+  const aliases = product === 'blackboard' ? ['blackboard', 'whiteboard'] : [product];
+  return aliases.some((alias) => new RegExp(`(?:^|[/_.-])${alias}(?=$|[/_.-])`).test(path));
+}
+
+const PRODUCT_MODULES = PRODUCT_GROUPS.flatMap((group) =>
+  group.products.map((product) => ({
+    id: `product:${product}`,
+    label: `产品：${product}`,
+    primary: group.primary,
+    backup: group.backup,
+    requiresSecondary: false,
+    matches: (path) => productMatches(path, product),
+  })),
+);
+
+const ALL_MODULES = [MODULES[0], MODULES[1], ...PRODUCT_MODULES, MODULES[2], MODULES[3]];
+
+function normalizedPaths(file) {
+  return [file?.filename, file?.previous_filename]
+    .filter((path) => typeof path === 'string' && path !== '')
+    .map((path) => path.toLowerCase());
+}
+
+function compareStats(left, right) {
+  return right.files - left.files || left.module.order - right.module.order || left.module.id.localeCompare(right.module.id);
+}
+
+function classifyFiles(files) {
+  const counts = new Map();
+  for (const file of files || []) {
+    const matchingModules = new Set();
+    for (const path of normalizedPaths(file)) {
+      const matches = ALL_MODULES.filter((module) => module.matches(path));
+      const securityOrDelivery = matches.filter(
+        (module) => module.id === 'security' || module.id === 'delivery',
+      );
+      const effectiveMatches = securityOrDelivery.length > 0
+        ? [...securityOrDelivery, ...matches.filter((module) => module.id === 'compatibility')]
+        : matches;
+      for (const match of effectiveMatches) {
+        matchingModules.add(match.id);
+      }
+      if (
+        effectiveMatches.length === 0 &&
+        (path.startsWith('internal/helpers/') || path.startsWith('internal/shortcut/'))
+      ) {
+        matchingModules.add('architecture');
+      }
+    }
+    for (const moduleID of matchingModules) {
+      counts.set(moduleID, (counts.get(moduleID) || 0) + 1);
+    }
+  }
+
+  return [...counts.entries()]
+    .map(([id, files]) => {
+      const index = ALL_MODULES.findIndex((module) => module.id === id);
+      return {module: {...ALL_MODULES[index], order: index}, files};
+    })
+    .sort(compareStats);
+}
+
+function chooseModuleReviewer(module, unavailable) {
+  return [module.primary, module.backup].find(
+    (reviewer) => REVIEWER_POOL.includes(reviewer) && !unavailable.has(reviewer),
+  );
+}
+
+function addReviewer(reviewers, reviewer) {
+  if (reviewer && !reviewers.includes(reviewer)) {
+    reviewers.push(reviewer);
+  }
+}
+
+function reviewerCandidates({preferredReviewers, fallbackReviewers, eligibleReviewers}) {
+  const eligible = new Set(eligibleReviewers.map((reviewer) => reviewer.toLowerCase()));
+  const candidates = [];
+  for (const reviewer of [...preferredReviewers, ...fallbackReviewers]) {
+    if (
+      eligible.has(reviewer.toLowerCase()) &&
+      !candidates.some((candidate) => candidate.toLowerCase() === reviewer.toLowerCase())
+    ) {
+      candidates.push(reviewer);
+    }
+  }
+  return candidates;
+}
+
+async function requestReviewersWithFallback({
+  candidates,
+  requiredReviewers,
+  satisfiedReviewers = [],
+  requestReviewer,
+  onFailure = () => {},
+}) {
+  const alreadySatisfied = new Set(
+    satisfiedReviewers.map((reviewer) => reviewer.toLowerCase()),
+  );
+  const satisfied = new Set();
+  const requested = [];
+
+  for (const reviewer of candidates) {
+    if (satisfied.size >= requiredReviewers) {
+      break;
+    }
+    const normalizedReviewer = reviewer.toLowerCase();
+    if (alreadySatisfied.has(normalizedReviewer)) {
+      satisfied.add(normalizedReviewer);
+      continue;
+    }
+
+    try {
+      const shouldContinue = await requestReviewer(reviewer);
+      if (shouldContinue === false) {
+        return {requested, satisfiedReviewers: [...satisfied], aborted: true};
+      }
+      requested.push(reviewer);
+      satisfied.add(normalizedReviewer);
+    } catch (error) {
+      onFailure(reviewer, error);
+    }
+  }
+
+  return {requested, satisfiedReviewers: [...satisfied], aborted: false};
+}
+
+function resolveReviewRouting({files, author, latestPusher, fallbackReviewers = REVIEWER_POOL}) {
+  const modules = classifyFiles(files);
+  const unavailable = new Set([author, latestPusher].filter(Boolean).map((login) => login.toLowerCase()));
+  const reviewers = [];
+  const primaryModule = modules[0];
+
+  if (!primaryModule) {
+    return {modules: [], reviewers, requiredReviewers: 1, reason: 'unknown_paths'};
+  }
+
+  addReviewer(reviewers, chooseModuleReviewer(primaryModule.module, unavailable));
+
+  const requiresSecondary =
+    modules.length > 1 || modules.some(({module}) => module.requiresSecondary);
+  const secondaryModule = modules.find(({module}) => module.id !== primaryModule.module.id) || primaryModule;
+  if (requiresSecondary) {
+    addReviewer(
+      reviewers,
+      chooseModuleReviewer(secondaryModule.module, new Set([...unavailable, ...reviewers])),
+    );
+  }
+
+  for (const reviewer of fallbackReviewers) {
+    if (reviewers.length >= (requiresSecondary ? 2 : 1)) {
+      break;
+    }
+    if (REVIEWER_POOL.includes(reviewer) && !unavailable.has(reviewer)) {
+      addReviewer(reviewers, reviewer);
+    }
+  }
+
+  return {
+    modules: modules.map(({module, files}) => ({id: module.id, label: module.label, files})),
+    reviewers,
+    requiredReviewers: requiresSecondary ? 2 : 1,
+    reason: requiresSecondary ? 'cross_or_sensitive' : 'single_module',
+  };
+}
+
+module.exports = {
+  REVIEWER_POOL,
+  classifyFiles,
+  requestReviewersWithFallback,
+  resolveReviewRouting,
+  reviewerCandidates,
+};

.github/reviewer-routing.test.js

@@ -0,0 +1,148 @@
+'use strict';
+
+const assert = require('node:assert/strict');
+const {
+  requestReviewersWithFallback,
+  resolveReviewRouting,
+  reviewerCandidates,
+} = require('./reviewer-routing');
+
+function route(files, author = 'author', latestPusher = author) {
+  return resolveReviewRouting({files: files.map((filename) => ({filename})), author, latestPusher});
+}
+
+{
+  const result = route(['internal/helpers/chat_toolbar.go']);
+  assert.deepEqual(result.reviewers, ['wxianfeng']);
+  assert.equal(result.requiredReviewers, 1);
+  assert.deepEqual(result.modules.map((module) => module.id), ['product:chat']);
+}
+
+{
+  const result = route(['internal/helpers/chat_toolbar.go', 'internal/helpers/doc_style.go']);
+  assert.deepEqual(result.reviewers, ['wxianfeng', 'typefield']);
+  assert.equal(result.requiredReviewers, 2);
+}
+
+{
+  const result = route(['.github/workflows/ci.yml']);
+  assert.deepEqual(result.reviewers, ['haofeng0705', 'wxianfeng']);
+  assert.equal(result.requiredReviewers, 2);
+  assert.equal(result.reason, 'cross_or_sensitive');
+}
+
+{
+  const result = route(['internal/auth/login.go'], 'hlzjsong');
+  assert.deepEqual(result.reviewers, ['typefield', 'wxianfeng']);
+  assert.equal(result.requiredReviewers, 2);
+}
+
+{
+  const result = route(['internal/upgrade/downloader.go']);
+  assert.deepEqual(result.reviewers, ['haofeng0705', 'wxianfeng']);
+  assert.equal(result.requiredReviewers, 2);
+}
+
+{
+  const result = route(['internal/app/upgrade.go', 'scripts/dev/test-release.sh']);
+  assert.deepEqual(result.reviewers, ['haofeng0705', 'wxianfeng']);
+  assert.equal(result.requiredReviewers, 2);
+}
+
+{
+  const result = route(['pkg/edition/edition.go']);
+  assert.deepEqual(result.reviewers, ['hlzjsong', 'typefield']);
+  assert.equal(result.requiredReviewers, 2);
+}
+
+{
+  const result = route(['internal/shortcut/chat/compatibility_coverage_test.go']);
+  assert.deepEqual(result.reviewers, ['wxianfeng', 'typefield']);
+  assert.equal(result.requiredReviewers, 2);
+  assert.deepEqual(result.modules.map((module) => module.id), ['product:chat', 'compatibility']);
+}
+
+{
+  const result = route(['internal/helpers/leaf_dispatch.go']);
+  assert.deepEqual(result.reviewers, ['wxianfeng', 'typefield']);
+  assert.equal(result.requiredReviewers, 2);
+}
+
+{
+  const result = route(['docs/unknown-area.md']);
+  assert.deepEqual(result.reviewers, []);
+  assert.equal(result.reason, 'unknown_paths');
+}
+
+async function testSingleReviewerFallback() {
+  const candidates = reviewerCandidates({
+    preferredReviewers: ['wxianfeng'],
+    fallbackReviewers: ['wxianfeng', 'typefield', 'haofeng0705'],
+    eligibleReviewers: ['wxianfeng', 'typefield', 'haofeng0705'],
+  });
+  const attempts = [];
+  const result = await requestReviewersWithFallback({
+    candidates,
+    requiredReviewers: 1,
+    requestReviewer: async (reviewer) => {
+      attempts.push(reviewer);
+      if (reviewer === 'wxianfeng') {
+        throw Object.assign(new Error('cannot request primary'), {status: 422});
+      }
+      return true;
+    },
+  });
+  assert.deepEqual(attempts, ['wxianfeng', 'typefield']);
+  assert.deepEqual(result.requested, ['typefield']);
+  assert.equal(result.satisfiedReviewers.length, 1);
+}
+
+async function testTwoReviewerFallback() {
+  const candidates = reviewerCandidates({
+    preferredReviewers: ['haofeng0705', 'wxianfeng'],
+    fallbackReviewers: ['haofeng0705', 'wxianfeng', 'typefield', 'hlzjsong'],
+    eligibleReviewers: ['haofeng0705', 'wxianfeng', 'typefield', 'hlzjsong'],
+  });
+  const attempts = [];
+  const result = await requestReviewersWithFallback({
+    candidates,
+    requiredReviewers: 2,
+    requestReviewer: async (reviewer) => {
+      attempts.push(reviewer);
+      if (reviewer === 'wxianfeng') {
+        throw Object.assign(new Error('temporary failure'), {status: 503});
+      }
+      return true;
+    },
+  });
+  assert.deepEqual(attempts, ['haofeng0705', 'wxianfeng', 'typefield']);
+  assert.deepEqual(result.requested, ['haofeng0705', 'typefield']);
+  assert.equal(result.satisfiedReviewers.length, 2);
+}
+
+async function testLowerPriorityExistingRequestDoesNotReplaceOwner() {
+  const attempts = [];
+  const result = await requestReviewersWithFallback({
+    candidates: ['wxianfeng', 'typefield'],
+    requiredReviewers: 1,
+    satisfiedReviewers: ['typefield'],
+    requestReviewer: async (reviewer) => {
+      attempts.push(reviewer);
+      return true;
+    },
+  });
+  assert.deepEqual(attempts, ['wxianfeng']);
+  assert.deepEqual(result.requested, ['wxianfeng']);
+  assert.deepEqual(result.satisfiedReviewers, ['wxianfeng']);
+}
+
+Promise.all([
+  testSingleReviewerFallback(),
+  testTwoReviewerFallback(),
+  testLowerPriorityExistingRequestDoesNotReplaceOwner(),
+])
+  .then(() => console.log('reviewer routing policy tests passed'))
+  .catch((error) => {
+    console.error(error);
+    process.exitCode = 1;
+  });

.github/workflows/ci.yml

@@ -433,6 +433,10 @@ jobs:
         if: steps.classify.outputs.changelog_only != 'true' && steps.classify.outputs.docs_only != 'true'
         run: go run github.com/rhysd/actionlint/cmd/actionlint@v1.7.12
 
+      - name: Test reviewer routing policy
+        if: steps.classify.outputs.changelog_only != 'true' && steps.classify.outputs.docs_only != 'true'
+        run: node .github/reviewer-routing.test.js
+
   test-focused:
     name: Test (changed packages)
     needs: lint
@@ -1366,36 +1370,47 @@ jobs:
           if [ -z "$base_ref" ] || [ "$base_ref" = "0000000000000000000000000000000000000000" ]; then
             base_ref="$(git rev-parse HEAD^)"
           fi
+          candidate_ref="$(git rev-parse 'HEAD^{commit}')"
+          if [ "$GITHUB_EVENT_NAME" = "pull_request" ] && [ "$candidate_ref" != "$PR_HEAD_SHA" ]; then
+            echo "Compatibility checkout $candidate_ref does not match PR head $PR_HEAD_SHA" >&2
+            exit 1
+          fi
           git rev-parse --verify "${base_ref}^{commit}" >/dev/null
-          stable_ref="$(git tag --merged "$base_ref" --list 'v[0-9]*' --sort=-version:refname | awk 'index($0, "-") == 0 { print; exit }')"
+          . ./scripts/release/release-lib.sh
+          stable_ref=""
+          for tag in $(git tag --merged "$base_ref" --list 'v*' --sort=-version:refname); do
+            release_is_stable_version "$tag" || continue
+            if git rev-parse --verify --quiet "refs/tags/withdrawn/$tag" >/dev/null; then
+              continue
+            fi
+            stable_ref="$tag"
+            break
+          done
           if [ -z "$stable_ref" ]; then
             echo "No stable release tag is reachable from compatibility base $base_ref" >&2
             exit 1
           fi
           git rev-parse --verify "${stable_ref}^{commit}" >/dev/null
-          echo "COMPATIBILITY_BASE_REF=$base_ref" >> "$GITHUB_ENV"
-          echo "COMPATIBILITY_STABLE_REF=$stable_ref" >> "$GITHUB_ENV"
+          printf '%s\n' \
+            "COMPATIBILITY_BASE_REF=$base_ref" \
+            "COMPATIBILITY_STABLE_REF=$stable_ref" \
+            "COMPATIBILITY_CANDIDATE_REF=$candidate_ref" >> "$GITHUB_ENV"
 
-      - name: Check historical commands and help compatibility
+      - name: Check historical commands, help, and complete CLI compatibility
         if: ${{ needs.lint.outputs.changelog_only != 'true' && needs.lint.outputs.docs_only != 'true' && (needs.lint.outputs.full_suite == 'true' || needs.lint.outputs.interface_sensitive == 'true') }}
         run: |
           make authoritative-interface-integrity \
-            BASE_REF="$COMPATIBILITY_BASE_REF"
-          if [ "$(git rev-parse "${COMPATIBILITY_BASE_REF}^{commit}")" != "$(git rev-parse "${COMPATIBILITY_STABLE_REF}^{commit}")" ]; then
-            make authoritative-interface-integrity \
-              BASE_REF="$COMPATIBILITY_STABLE_REF"
-          fi
-
-      - name: Check complete CLI command compatibility
-        if: ${{ needs.lint.outputs.changelog_only != 'true' && needs.lint.outputs.docs_only != 'true' && (needs.lint.outputs.full_suite == 'true' || needs.lint.outputs.interface_sensitive == 'true') }}
-        run: |
-          ./scripts/policy/check-command-compatibility.sh \
-            --base-ref "$COMPATIBILITY_BASE_REF" \
-            --stable-ref "$COMPATIBILITY_STABLE_REF"
+            BASE_REF="$COMPATIBILITY_BASE_REF" \
+            STABLE_REF="$COMPATIBILITY_STABLE_REF" \
+            CANDIDATE_REF="$COMPATIBILITY_CANDIDATE_REF"
 
       - name: Check complete Schema compatibility
         if: ${{ needs.lint.outputs.changelog_only != 'true' && needs.lint.outputs.docs_only != 'true' && (needs.lint.outputs.full_suite == 'true' || needs.lint.outputs.interface_sensitive == 'true') }}
-        run: make schema-compatibility BASE_REF="$COMPATIBILITY_BASE_REF"
+        run: |
+          make schema-compatibility \
+            BASE_REF="$COMPATIBILITY_BASE_REF" \
+            STABLE_REF="$COMPATIBILITY_STABLE_REF" \
+            CANDIDATE_REF="$COMPATIBILITY_CANDIDATE_REF"
 
       - name: Check skill command references
         if: ${{ needs.lint.outputs.changelog_only != 'true' && needs.lint.outputs.docs_only != 'true' && (needs.lint.outputs.full_suite == 'true' || needs.lint.outputs.interface_sensitive == 'true') }}

.github/workflows/reviewer-router.yml

@@ -21,6 +21,11 @@ jobs:
     runs-on: ubuntu-latest
     timeout-minutes: 5
     steps:
+      - name: Check out trusted routing policy
+        uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          ref: ${{ github.event.pull_request.base.sha }}
+          persist-credentials: false
       - name: Route review and enable auto-merge
         uses: actions/github-script@f28e40c7f34bde8b3046d885e986cb6290c5673b # v7.1.0
         with:
@@ -29,12 +34,13 @@ jobs:
             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',
-            ];
+            const {
+              REVIEWER_POOL,
+              requestReviewersWithFallback,
+              resolveReviewRouting,
+              reviewerCandidates,
+            } = require('./.github/reviewer-routing.js');
+            const reviewerPool = REVIEWER_POOL;
 
             async function getReadyEventPull(phase) {
               const {data: currentPull} = await github.rest.pulls.get({
@@ -67,6 +73,21 @@ jobs:
                 : author;
 
             async function routeReview() {
+              let changedFiles;
+              try {
+                changedFiles = await github.paginate(github.rest.pulls.listFiles, {
+                  owner,
+                  repo,
+                  pull_number: pullNumber,
+                  per_page: 100,
+                });
+              } catch (error) {
+                core.warning(
+                  `Could not inspect changed files for PR #${pullNumber}; using load-balanced fallback (${error.status || 'unknown status'}).`,
+                );
+                changedFiles = [];
+              }
+
               const eligible = reviewerPool.filter(
                 reviewer =>
                   reviewer.toLowerCase() !== author &&
@@ -77,13 +98,9 @@ jobs:
                 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;
-              }
+              const existingRequestedReviewers = new Set(
+                (pullRequest.requested_reviewers || []).map(({login}) => login.toLowerCase()),
+              );
 
               let reviews;
               try {
@@ -116,22 +133,18 @@ jobs:
                   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),
+              const currentHeadReviewers = new Set(
+                [...latestDecisionByLogin.values()]
+                  .filter(
+                    review =>
+                      review.commit_id === headSha &&
+                      eligible.some(
+                        reviewer => reviewer.toLowerCase() === review.user.login.toLowerCase(),
+                      ) &&
+                      ['APPROVED', 'CHANGES_REQUESTED'].includes(review.state),
+                  )
+                  .map(review => review.user.login.toLowerCase()),
               );
-              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 {
@@ -178,20 +191,42 @@ jobs:
                   tieOrder.get(left) - tieOrder.get(right),
               );
 
-              for (const reviewer of ranked) {
-                try {
+              const routing = resolveReviewRouting({
+                files: changedFiles,
+                author,
+                latestPusher,
+                fallbackReviewers: ranked,
+              });
+              const candidates = reviewerCandidates({
+                preferredReviewers: routing.reviewers,
+                fallbackReviewers: ranked,
+                eligibleReviewers: eligible,
+              });
+              const desiredReviewers = candidates.slice(0, routing.requiredReviewers);
+              if (routing.reason === 'unknown_paths' && currentHeadReviewers.size > 0) {
+                core.info(
+                  `PR #${pullNumber} has a current-head review for unknown paths; leaving manual ownership unchanged.`,
+                );
+                return;
+              }
+              core.info(
+                `PR #${pullNumber} routing: ${routing.reason}; modules=${routing.modules.map(module => module.id).join(',') || 'unknown'}; reviewers=${desiredReviewers.join(',') || 'load-balanced fallback'}.`,
+              );
+
+              const satisfiedReviewers = new Set([
+                ...currentHeadReviewers,
+                ...[...existingRequestedReviewers].filter((reviewer) =>
+                  candidates.some((candidate) => candidate.toLowerCase() === reviewer),
+                ),
+              ]);
+              const requestResult = await requestReviewersWithFallback({
+                candidates,
+                requiredReviewers: routing.requiredReviewers,
+                satisfiedReviewers: [...satisfiedReviewers],
+                requestReviewer: async (reviewer) => {
                   const currentPull = await getReadyEventPull('review request');
                   if (!currentPull) {
-                    return;
-                  }
-                  if (
-                    (currentPull.requested_reviewers || []).length > 0 ||
-                    (currentPull.requested_teams || []).length > 0
-                  ) {
-                    core.info(
-                      `PR #${pullNumber} received a reviewer while routing; leaving it unchanged.`,
-                    );
-                    return;
+                    return false;
                   }
                   await github.rest.pulls.requestReviewers({
                     owner,
@@ -202,15 +237,23 @@ jobs:
                   core.info(
                     `Requested @${reviewer} for PR #${pullNumber} (open request load: ${loads.get(reviewer)}).`,
                   );
-                  return;
-                } catch (error) {
+                  return true;
+                },
+                onFailure: (reviewer, error) => {
                   core.warning(
                     `Could not request @${reviewer} for PR #${pullNumber}; trying the next candidate (${error.status || 'unknown status'}).`,
                   );
-                }
+                },
+              });
+              if (requestResult.aborted) {
+                return;
               }
 
-              core.warning(`No reviewer request could be created for PR #${pullNumber}.`);
+              if (requestResult.satisfiedReviewers.length < routing.requiredReviewers) {
+                core.warning(
+                  `Only ${requestResult.satisfiedReviewers.length} of ${routing.requiredReviewers} required reviewers could be satisfied for PR #${pullNumber}.`,
+                );
+              }
             }
 
             async function enableAutoMerge() {

CHANGELOG.md

@@ -6,15 +6,93 @@ The format is inspired by [Keep a Changelog](https://keepachangelog.com/) and th
 
 ## [Unreleased]
 
+### Added
+
+- **Streaming-card mentions** — `chat +messages-send-card` now accepts
+  `--at-open-dingtalk-ids` and `--at-all` for group cards, passing mention
+  targets to the initial card-creation request and prepending its returned
+  `atTag` to the automatic streaming update.
+
+## [1.0.58-beta.2] - 2026-08-10
+
+### Added
+
+- **`dws sheet create-with-data`（新命令）** — 建表并写入初始数据与样式：`--values`（二维数组写默认表）/ `--sheets`（typed table 多工作表，二者必须给一个）/ `--styles`（`cell_styles` / `row_sizes` / `col_sizes` / `cell_merges`，顶层键对齐飞书 snake_case、列表项内字段兼容 camelCase）。所有结构、字段类型与枚举在创建文档之前校验，非法配置不会留下白建的空文档：`--sheets` 按 `table_put` 的输入契约逐字段校验（`columns` 必填且列名非空不重复、`data` 为二维且行宽与 `columns` 一致、单元格仅限字符串/数字/布尔/null、`dtypes`/`formats` 的键须是列名、`mode`/`header`/`allowOverwrite`/`startCell` 类型与取值、单表 30000 单元格上限），并拒绝未知键、snake_case 变体、`{"sheets":"bad"}` 这类畸形包装与 `sheetId`（服务端会静默丢弃写错的键，导致"只写了表头却报成功"的静默丢数据）；`--values` 校验单元格为标量并受 30000 单元格 / 2000000 字符上限约束；`--styles` 的顶层键与列表项内字段同样拒绝未知键，避免样式只应用一半。写入后回读校验按 `startCell` / `header` / `mode` 推算的首个预期非空单元格，而非固定 A1。该命令是多步编排（建文档 → 探活 → 定位默认工作表 → 写数据 → 回读 → 可选样式），因此如实声明为独立叶子 `sheet.create_with_data` + `interface_mode: composite`（附评审 reason，按契约不带 `interface_ref`）；**`dws sheet create` 保持原样不变**——仍是一次 `create_workspace_sheet` 直连（`interface_mode: mcp`），不新增 flag，避免让 Schema 消费者把编排步骤的参数误当成该 RPC 的入参。
+- **`dws sheet export-csv`（新命令）** — 同步导出单个工作表为 RFC4180 CSV，支持 `--sheet-id` 选表、`--range` 限定范围、`--value-render-option` 选取值模式；`--output` 落盘（为目录时按 `sheet-export.csv` 命名，落盘走 `AtomicWrite` 原子替换，写入失败时已有文件保持原样；父目录不存在按错误处理，不会自动创建），不传则把纯 CSV 打到 stdout（警告只走 stderr）。数据超出单次读取上限时默认报错、既不输出也不写文件，需 `--allow-truncated` 显式接受不完整结果。响应缺 `csv` 字段或类型不对一律报错，不会用 0 字节覆盖已有文件。该分支读的是 `get_range_as_csv`、与 xlsx 的异步导出任务毫无关系，因此独立成叶子 `sheet.export_csv` 并如实声明 `interface_mode: mcp` + `interface_ref: get_range_as_csv`；**`dws sheet export` 保持原样不变**——仍只导 xlsx（`interface_ref: submit_export_job`），flag 面仍是 `--node` / `--output`，csv 专属 flag 不会出现在它上面（此前挂在同一条命令上时，漏写 `--export-format csv` 会让 `--range` 被静默丢弃而导出整篇工作簿）。
+- **`sheet update-dimension --size-type`** — `pixel` / `standard`（恢复默认行高列宽）/ `auto`（按内容自适应行高，仅 ROWS）。
+- **`sheet replace --match-formula`** — 在公式文本中查找替换。
+- **`sheet range set-style` 扩展样式维度** — 新增 `--font-style`（斜体）/ `--font-line`（下划线、删除线）/ `--font-family` / `--border-styles-json`（四边边框；每条边只接受 `style` / `color`，未知键与非字符串 `color` 直接报错，不再静默忽略而画出无颜色的边框，`set-style` / `batch-set-style` / `create-with-data --styles` 三条路径同源校验）。
+- **`sheet range batch-set-style --ranges`** — 一组样式刷多个带工作表前缀的区域，组装为一次原子 `batch_update`。
+
 ### Changed
 
+- **Chat message post-send ID handoff** (#897) — CLI Help and bundled Skills
+  now document the `send` → `query-send-status` → `edit`/`recall` workflow,
+  so callers can reuse returned task, message, and conversation IDs instead
+  of searching message history by content.
+- **Sheet mono/multi Skill alignment** — replaces the oversized mono Sheet
+  reference with the progressive routing layout, aligns all 20 Sheet topic
+  references across the mono and multi bundles, and adds a content-policy guard
+  that prevents the paired topic trees from drifting again.
+- **`sheet range set-style` 后端切换为 `set_cell_range`** — 样式统一走 cellStyles 路径（仅设样式、保留原值），这是斜体/下划线删除线/字体族/边框唯一可用的通道。`interface_ref` 由 `update_range` 变为 `set_cell_range`，12 个样式 flag 改为 reviewed mapping exclusion。CLI 用法向后兼容、无 flag 删除；schema-compatibility 经 reviewed 豁免判定为兼容（0 changed fields）。
+- **`sheet range batch-set-style` 改为单次原子提交** — 由本地循环多次 `update_range` 改为一次 `batch_update`，任一项失败默认整批回滚；`--continue-on-error` 由本地控制改为透传服务端。新增批量上限：最多 100 个区域且累计不超过 200000 个单元格。
+- **`sheet range batch-clear` / `batch-set-style` 的 `--ranges` 拒绝空白工作表前缀**（用户可见行为变更）— 此前只按原始串里 `!` 的位置判断，`" !A1:B2"` 修剪后工作表名成了空串，操作却照样带着 `sheetId: ""` 提交：服务端要么让整批 `batch_update` 失败，要么更糟——落到默认工作表而不是用户指定的那张表，且命令报成功。现在工作表名与范围都必须在修剪之后仍非空，否则在发起任何请求之前报错。`batch-set-style --batch` 的纯空白 `sheetId` / `range` 同样拒绝（此前只挡空字符串）；`--batch` 下发仍用原值不替用户修剪，因为 `sheetId` 可以是允许带首尾空格的工作表**名**。两条 `--ranges` 路径现在共用同一个拆分器。
+- **`sheet insert-dimension` / `delete-dimension` / `update-dimension` 的 `--length` 严格校验**（用户可见行为变更）— 解析由 `fmt.Sscanf("%d")` 改为 `strconv.Atoi`。此前只消费前缀数字，`--length 2x` / `3foo` 会被静默当成 `2` / `3` 并对错误的行列数执行操作（删除方向不可回滚）；现在整个值必须是合法正整数，否则报错「`--length` 必须为正整数（>= 1）」且不发起任何请求。**升级影响**：原先依赖这种宽松解析、在传畸形 `--length` 的脚本会开始报错，请把参数修正为纯数字。合法数字值行为不变，上限仍为 5000。`add-dimension` 的 `--length` 是 `Int` 类型 flag，一直由 cobra 严格校验，不受影响。
+- **CLI 接口兼容门禁支持 reviewed flag 类型豁免**（无用户可见变更）— `authoritative-interface-integrity` 与 `check-command-compatibility.sh` 此前一律拒绝历史命令的 flag 类型变更，即使新类型只是把同一套校验从 RunE 前移到解析期，也没有任何评审通道。现在两道门禁各带一张精确豁免表：命令路径 + flag 名 + 旧类型 → 新类型四元组全等才命中、方向敏感（`string`→`int` 与 `int`→`string` 是两个不同的键，只有被评审的方向可用），且仅当该 flag 的其他契约（shorthand / required / hidden / no-opt / scope）纹丝不动时才放行，因此豁免夹带不了别的破坏。首条也是目前唯一一条登记的是 `dws minutes permission apply --policy` 的 `string` → `int`（配合 #912）：旧实现在 RunE 里做 `strconv.ParseInt(v, 10, 64)` 再校验 `[2,4]`，新实现由 pflag 以 `strconv.ParseInt(s, 0, 64)` 解析后仍校验 `[2,4]`，**历史上能成功的调用集是新调用集的子集**（base 0 额外接受 `0x3` 这类写法，只放宽不收紧），非法值依然失败、只是报错文案与时机前移；flag 默认值由 `""` 变 `"0"` 是类型的必然结果，两道门禁都不比较默认值，且该 flag 必须显式给出、默认值不可达。两张表必须逐字一致并有守卫测试锚定漂移——重复是被迫的而非选择：`check-authoritative-interface-baselines.sh` 会把整个 `scripts/policy/interface-baseline` 目录复制进检出历史版本的 worktree 再编译，那份拷贝不能 import 本分支新增的包。
+- **Schema 兼容门禁支持 reviewed 参数类型豁免**（无用户可见变更）— 接上一条。`schema-compatibility` 是同一个 `Interface Integrity` job 里排在两道 CLI 接口门禁之后的第三道检查，此前也一律拒绝已发布参数的 `type` 变更。由于前两道先失败、`set -e` 让它从未在 CI 上暴露，上一条豁免只解决了三分之二。现在 `checkParameterCompatibility` 也带一张精确豁免表：`<product>/<tool id>` + 参数名 + 旧类型 + 新类型四元组全等才命中、方向敏感，且仅当该参数**除 `type` 外的全部已发布字段逐字段相等**时才放行。这里刻意用相等性比较而非「没有产生其他兼容性错误」：放宽 `required` / `cli_required`、清空 `required_when`、扩宽 `enum`、清空 `interface_type`、经 reviewed mapping exclusion 清空 `property`——这些变化单独看都是兼容的、根本不产生错误，若以错误列表代替相等性检查，它们就能搭着一次已评审的类型迁移一起蒙混过关。结构体整体比较还意味着将来给 `parameterSchema` 新增字段时会自动纳入守卫，而不是悄悄放宽每一条既有条目。唯一条目是 `minutes/minutes.apply_minutes_permission` 的 `policy` 由 `"string"` 迁移到 `"integer"`（配合 #912）：该 `type` 由 Cobra flag 类型投影而来（provenance `cobra_flag_type`），描述的是 CLI 如何接受取值；消费方据此拼装的是命令行，而 `--policy 4` 在两种声明下是同一个 argv，加引号的 `--policy "4"` 到 pflag 仍是 4，RunE 也仍校验 `[2,4]`——而且该参数映射的 property `policyId` 一直以数字上报，新声明比旧声明更贴近真实请求。表里的类型值必须是 `schemaType` 实际产出的带引号形态（`"string"` 而非裸 `string`），守卫测试用 `schemaType` 复算并校验类型名属于 JSON Schema 的封闭取值集合——`reviewedInterfaceRefRedirect` 曾因键的书写形态错误两次静默失效，这里不重犯。
+
+## [1.0.58-beta.1] - 2026-08-07
+
+### Added
+
+- **Robot image and file messages** (#867) — `dws chat message send-by-bot`
+  now supports image URLs and local-file uploads through explicit message
+  types, while retaining Markdown as the default and preserving its existing
+  title and text requirements.
+- **Conversation shortcut-bar management** (#877) — adds `dws chat toolbar`
+  commands to list, add, hide, sort, and manage custom conversation shortcuts,
+  with validation and confirmation for destructive removal.
+- **Complete AI Table Shortcut surface** (#901) — makes all 92 supported
+  AI Table Shortcuts discoverable through Runtime Schema and adds reliable
+  Base, table, record, attachment, view, dashboard, and workflow operations
+  with explicit confirmation and result-verification semantics for writes.
+
+### Changed
+
+- **Doc import upload fallback** — `dws doc import` no longer fails on file
+  formats outside the conversion whitelist (html, pdf, zip, extensionless,
+  and any future format): it now hands the file to the document-space upload
+  chain (the same primitive as `dws drive upload --workspace`), stores the
+  original file at the requested `--folder`/`--workspace` target, and prints
+  an explicit stderr notice with the supported-format list and the
+  convert-to-md alternative. The fallback shares the import file checks
+  (20MB cap, empty-file guard), keeps `--format json` / `--dry-run` output as
+  a single JSON document, and marks the machine-readable result with
+  `fallback: "upload"` and `converted: false` so agents never mistake the
+  stored file for a converted online document. The fallback fails closed
+  unless the commit response parses as JSON and carries a file identity
+  (exposed as `dentry_id`); empty or unverifiable responses surface as
+  errors instead of fabricated success. Importable formats and
+  `dws sheet import` validation are unchanged.
+- **IM natural-target and history alignment** — Chat shortcuts can resolve natural user/group targets before execution, and message-history workflows expose bounded time ranges, ordering, explicit all-page controls, continuation ledgers, safe local export, and thread-reply pagination without treating empty or incomplete reads as successful results. Bundled mono/multi Skills and intent routing now describe the same executable surface.
 - **Sheet CSV formula writes** — `dws sheet csv-put` and batch `csv-put` now expose the service contract that CSV fields beginning with `=` are written as formulas. Prefix the field with an apostrophe to write literal text beginning with `=`; CSV content continues to pass through unchanged.
+- **Release-equivalent PR compatibility gate** (#889) — pull-request
+  admission now runs command-surface compatibility checks against the current
+  release baseline before code reaches `main`.
+- **Reviewer routing governance** (#903) — updates the Reviewer Router pool
+  used for new ready PRs while retaining the existing current-head review and
+  required-check gates.
 
 ### Fixed
 
+- **Fail-closed IM pagination and audit evidence** — `+chat-messages`, `+search-msg`, `+thread-replies`, `+at-me`, `+my-groups`, conversation lists, and favorites preserve partial-read failures, reject missing or stalled continuation state, deduplicate page boundaries, and publish completion evidence. The live-audit regression suite now rejects empty projections and incomplete reads instead of promoting them to passing results.
 - **Sheet formula verification** (#873) — `dws sheet formula-verify` now calls
   the registered remote tool name `verify_formula`; the previous
   `formula_verify` name failed at gateway dispatch.
+- **CLI and parameter recovery boundaries** (#864) — command and parameter
+  recovery now fail closed when an Agent-provided path or flag cannot be
+  reconciled with the executable CLI surface, reducing unsafe hallucinated
+  retries.
 
 ## [1.0.57-beta.4] - 2026-08-06
 
@@ -149,6 +227,7 @@ and compatibility and CI reliability fixes.
 
 ### Fixed
 
+- **Fail-closed IM pagination and audit evidence** — `+chat-messages`, `+search-msg`, `+thread-replies`, `+at-me`, `+my-groups`, conversation lists, and favorites preserve partial-read failures, reject missing or stalled continuation state, deduplicate page boundaries, and publish completion evidence. The live-audit regression suite now rejects empty projections and incomplete reads instead of promoting them to passing results.
 - **Unified command safety and Shortcut runtime (H0)** — Shortcut leaves now execute through `corecmd.New`, sharing the same typed Safety confirmation gate as Leaf commands. EOF / closed stdin returns `confirmation_required`, and interactive `no` returns the existing non-zero cancellation validation error instead of reporting success for an operation that did not run. Pass `--yes` or `--dry-run` to skip the prompt.
 - **Constraint "provided" for `at_least_one` / `exactly_one` (H0)** — a flag set to an empty string (`--flag ""`) no longer counts as provided; previously bare Cobra `Changed` satisfied the constraint. Pass a non-blank value for a member of the group.
 - **Chat media download JSON compatibility** — `dws chat message download-media --format json` once again returns a clean `{success, downloadUrl, output}` result after the file is saved, preserving the temporary URL and resolved local path without progress text corrupting JSON stdout.

CONTRIBUTING.md

@@ -68,9 +68,17 @@ coverage is additionally selected for platform-sensitive code.
 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.
+   paths/flags change. CI resolves the exact merge-base, latest reachable
+   non-withdrawn stable GA tag, and committed candidate SHA, then enters the single compatibility
+   decision seam through
+   `make authoritative-interface-integrity BASE_REF=<merge-base> STABLE_REF=<latest-GA-tag> CANDIDATE_REF=<candidate-sha>`.
+   The Make target delegates to the authoritative wrapper; CI does not invoke a
+   second comparator or the legacy fixture checker. See
+   [CLI flag compatibility migration governance](docs/cli-interface-flag-migrations.md)
+   for the reviewed two-stage `pending` → `consumed` lifecycle.
+   Agent-visible flag migrations must also run
+   `make schema-compatibility BASE_REF=<merge-base> STABLE_REF=<latest-GA-tag> CANDIDATE_REF=<candidate-sha>`;
+   it consumes the same base-owned ledger rather than a second exception list.
 5. Run `./scripts/policy/check-generated-drift.sh` when generated artifacts may
    change.
 6. Run `./scripts/release/verify-package-managers.sh` when packaging or

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 "1.0.57-beta.4"
+  version "1.0.58-beta.2"
   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/v1.0.57-beta.4/dws-darwin-arm64.tar.gz"
-      sha256 "ff363e258d463732e4dc02aa71ac1b5b1f05c25107c32ddcc5ae11d14931a782"
+      url "https://github.com/DingTalk-Real-AI/dingtalk-workspace-cli/releases/download/v1.0.58-beta.2/dws-darwin-arm64.tar.gz"
+      sha256 "1b2b6953f7f1ae1ca6ecb0702424ac0e1a976a6a5ff91e8ffc3b5ae495d98c7c"
     else
-      url "https://github.com/DingTalk-Real-AI/dingtalk-workspace-cli/releases/download/v1.0.57-beta.4/dws-darwin-amd64.tar.gz"
-      sha256 "3eafcc4c27931b8457f3611a12ce4ae727f7bd65356fbef80958699defa09acf"
+      url "https://github.com/DingTalk-Real-AI/dingtalk-workspace-cli/releases/download/v1.0.58-beta.2/dws-darwin-amd64.tar.gz"
+      sha256 "a1c1b3c58b48e04c0ae520062f9d6ab0dc961eddb635497bdb9b4345316e45f6"
     end
   end
 
   on_linux do
     if Hardware::CPU.arm?
-      url "https://github.com/DingTalk-Real-AI/dingtalk-workspace-cli/releases/download/v1.0.57-beta.4/dws-linux-arm64.tar.gz"
-      sha256 "49eabb4d2c419d8d1fb0fcf71d9c318ef49cf3d198431df1bcbaa71589e11383"
+      url "https://github.com/DingTalk-Real-AI/dingtalk-workspace-cli/releases/download/v1.0.58-beta.2/dws-linux-arm64.tar.gz"
+      sha256 "7f35e3c4734f17b125a8c32f3c95e05d1410f683cf6956be857ee9349f8e4d36"
     else
-      url "https://github.com/DingTalk-Real-AI/dingtalk-workspace-cli/releases/download/v1.0.57-beta.4/dws-linux-amd64.tar.gz"
-      sha256 "fb3903485fe494e1fadb67ec87b2fef19cf575c72b7df25f0b93f80f8b36273f"
+      url "https://github.com/DingTalk-Real-AI/dingtalk-workspace-cli/releases/download/v1.0.58-beta.2/dws-linux-amd64.tar.gz"
+      sha256 "37beb9e39790563cf0584ac23376f713bf2eb2c50cff4222965e831ac9adbb0e"
     end
   end
 
   resource "skills" do
-    url "https://github.com/DingTalk-Real-AI/dingtalk-workspace-cli/releases/download/v1.0.57-beta.4/dws-skills.zip"
-    sha256 "6bd11363dbd2ce79627f49bc80b8a92f9c79fca083a6bd3ff9bc5c4833594fb9"
+    url "https://github.com/DingTalk-Real-AI/dingtalk-workspace-cli/releases/download/v1.0.58-beta.2/dws-skills.zip"
+    sha256 "7e10fead4192059c98d596c5b1886f77fd550526de5cd18c425cdad6fd64cd3a"
   end
 
   def install

Makefile

@@ -24,13 +24,13 @@ help:
 	@printf "  make format-check  - Check all repository Go source files with gofmt\n"
 	@printf "  make fmt           - Format all repository Go source files\n"
 	@printf "  make policy        - Check the built dws plus open-source and Schema policies\n"
-	@printf "  make interface-integrity - Check historical commands and help contracts still work\n"
-	@printf "  make authoritative-interface-integrity BASE_REF=<ref> - Check the Git-owned PR merge-base\n"
+	@printf "  make interface-integrity [BASE_REF=<ref>] [STABLE_REF=<tag>] [CANDIDATE_REF=<ref>] - Check authoritative CLI history\n"
+	@printf "  make authoritative-interface-integrity BASE_REF=<ref> [STABLE_REF=<tag>] [CANDIDATE_REF=<ref>] - Check Git-owned CLI history\n"
 	@printf "  make coverage-gate BASE_REF=<ref> - Enforce overall non-regression and 100%% changed-code coverage\n"
 	@printf "  make coverage-gate-platform BASE_REF=<ref> PROFILE=<file> - Enforce 100%% native changed-code coverage\n"
-	@printf "  make update-interface-baseline - Add new CLI contracts without removing history\n"
-	@printf "  make reset-interface-baseline - DANGEROUS: replace all CLI compatibility history\n"
-	@printf "  make schema-compatibility BASE_REF=<ref> - Check the complete Schema contract against the PR merge-base\n"
+	@printf "  make update-interface-baseline - Update the non-authoritative CLI smoke fixture\n"
+	@printf "  make reset-interface-baseline - DANGEROUS: replace the non-authoritative CLI smoke fixture\n"
+	@printf "  make schema-compatibility BASE_REF=<ref> [STABLE_REF=<tag>] [CANDIDATE_REF=<ref>] - Check the authoritative Schema history\n"
 	@printf "  make skill-command-integrity - Check dws commands referenced by skills exist\n"
 	@printf "  make skill-context-budget - Check generated Skill drift and common-path context budgets\n"
 	@printf "  make multi-im-skill-chain-integrity - Check reviewed IM intents keep one default Skill route\n"
@@ -89,6 +89,7 @@ policy: test-auth-legacy-compat
 	@$(POLICY_ENV) ./scripts/policy/check-open-source-assets.sh
 	@$(POLICY_ENV) ./scripts/policy/check-skill-context-budget.sh
 	@$(POLICY_ENV) ./scripts/policy/check-multi-im-skill-chain.sh
+	@python3 scripts/run_chat_shortcut_live_audit_test.py
 	@$(POLICY_ENV) ./scripts/policy/check-command-surface.sh --strict
 	@$(POLICY_ENV) ./scripts/policy/check-generated-drift.sh
 	@$(POLICY_ENV) ./scripts/policy/check-param-concepts.sh
@@ -102,10 +103,22 @@ edition-test:
 	$(GO) test -v -count=1 ./pkg/editiontest/...
 
 interface-integrity:
-	@./scripts/policy/check-interface-baseline.sh
+	@base_ref="$(BASE_REF)"; \
+	candidate_ref="$(CANDIDATE_REF)"; \
+	if [ -z "$$base_ref" ]; then base_ref="origin/main"; fi; \
+	if [ -z "$$candidate_ref" ]; then candidate_ref="HEAD"; fi; \
+	./scripts/policy/check-authoritative-interface-baselines.sh \
+		--base-ref "$$base_ref" \
+		--stable-ref "$(STABLE_REF)" \
+		--candidate-ref "$$candidate_ref"
 
 authoritative-interface-integrity:
-	@./scripts/policy/check-authoritative-interface-baselines.sh --base-ref "$(BASE_REF)"
+	@candidate_ref="$(CANDIDATE_REF)"; \
+	if [ -z "$$candidate_ref" ]; then candidate_ref="HEAD"; fi; \
+	./scripts/policy/check-authoritative-interface-baselines.sh \
+		--base-ref "$(BASE_REF)" \
+		--stable-ref "$(STABLE_REF)" \
+		--candidate-ref "$$candidate_ref"
 
 coverage-gate:
 	@./scripts/policy/check-coverage-gate.sh --base-ref "$(BASE_REF)" --scope-buildable
@@ -120,7 +133,12 @@ reset-interface-baseline:
 	@./scripts/policy/check-interface-baseline.sh --reset
 
 schema-compatibility:
-	@./scripts/policy/check-authoritative-schema-compatibility.sh --base-ref "$(BASE_REF)"
+	@candidate_ref="$(CANDIDATE_REF)"; \
+	if [ -z "$$candidate_ref" ]; then candidate_ref="HEAD"; fi; \
+	./scripts/policy/check-authoritative-schema-compatibility.sh \
+		--base-ref "$(BASE_REF)" \
+		--stable-ref "$(STABLE_REF)" \
+		--candidate-ref "$$candidate_ref"
 
 skill-command-integrity:
 	@./scripts/policy/check-skill-commands.sh

cmd/interface-snapshot/main.go

@@ -17,18 +17,23 @@
 package main
 
 import (
+	"bytes"
 	"encoding/json"
 	"flag"
 	"fmt"
 	"io"
 	"os"
 	"path/filepath"
+	"strings"
 
 	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/app"
 	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/i18n"
 	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/interfacesnapshot"
+	"github.com/spf13/cobra"
 )
 
+var newRootCommand = func() *cobra.Command { return app.NewRootCommand() }
+
 func main() {
 	os.Exit(run(os.Args[1:], os.Stdout, os.Stderr))
 }
@@ -112,7 +117,11 @@ func runGenerate(args []string, stdout, stderr io.Writer) error {
 	defer i18n.SetLang(previousLang)
 	i18n.SetLang("en")
 
-	snapshot := interfacesnapshot.Capture(app.NewRootCommand())
+	root := newRootCommand()
+	snapshot := interfacesnapshot.Capture(root)
+	if err := validateHelpRendering(root, snapshot); err != nil {
+		return err
+	}
 	if *output == "-" {
 		return interfacesnapshot.Write(stdout, snapshot)
 	}
@@ -138,6 +147,16 @@ func runCompare(args []string, stdout, stderr io.Writer) (bool, error) {
 	currentPath := flags.String("current", "", "candidate snapshot path")
 	basePath := flags.String("base", "", "target main/development baseline snapshot path")
 	stablePath := flags.String("stable", "", "latest stable GA snapshot path")
+	approvedMigrationsPath := flags.String(
+		"approved-flag-migrations",
+		"",
+		"merge-base-owned approved flag migration manifest",
+	)
+	candidateMigrationsPath := flags.String(
+		"candidate-flag-migrations",
+		"",
+		"candidate flag migration manifest",
+	)
 	if err := flags.Parse(args); err != nil {
 		return false, err
 	}
@@ -150,6 +169,14 @@ func runCompare(args []string, stdout, stderr io.Writer) (bool, error) {
 	if *basePath == "" && *stablePath == "" {
 		return false, fmt.Errorf("compare requires --base, --stable, or both")
 	}
+	if (*approvedMigrationsPath == "") != (*candidateMigrationsPath == "") {
+		return false, fmt.Errorf(
+			"--approved-flag-migrations and --candidate-flag-migrations must be provided together",
+		)
+	}
+	if *approvedMigrationsPath != "" && (*basePath == "" || *stablePath == "") {
+		return false, fmt.Errorf("flag migration compare requires both --base and --stable")
+	}
 
 	current, err := readSnapshot(*currentPath)
 	if err != nil {
@@ -170,6 +197,25 @@ func runCompare(args []string, stdout, stderr io.Writer) (bool, error) {
 	}
 
 	report := interfacesnapshot.CompareAll(current, references)
+	if *approvedMigrationsPath != "" {
+		approved, readErr := readFlagMigrationManifest(*approvedMigrationsPath)
+		if readErr != nil {
+			return false, fmt.Errorf("read approved flag migrations: %w", readErr)
+		}
+		candidate, readErr := readFlagMigrationManifest(*candidateMigrationsPath)
+		if readErr != nil {
+			return false, fmt.Errorf("read candidate flag migrations: %w", readErr)
+		}
+		report, err = interfacesnapshot.CompareAllWithFlagMigrations(
+			current,
+			references,
+			approved,
+			candidate,
+		)
+		if err != nil {
+			return false, fmt.Errorf("validate flag migration lifecycle: %w", err)
+		}
+	}
 	encoder := json.NewEncoder(stdout)
 	encoder.SetEscapeHTML(false)
 	encoder.SetIndent("", "  ")
@@ -179,6 +225,49 @@ func runCompare(args []string, stdout, stderr io.Writer) (bool, error) {
 	return report.Compatible, nil
 }
 
+func readFlagMigrationManifest(path string) (interfacesnapshot.FlagMigrationManifest, error) {
+	file, err := os.Open(filepath.Clean(path))
+	if err != nil {
+		return interfacesnapshot.FlagMigrationManifest{}, err
+	}
+	defer file.Close()
+	return interfacesnapshot.ReadFlagMigrationManifest(file)
+}
+
+func validateHelpRendering(root *cobra.Command, snapshot interfacesnapshot.Snapshot) error {
+	for _, command := range snapshot.Commands {
+		path := strings.TrimPrefix(command.Path, "dws")
+		resolved, remaining, err := root.Find(strings.Fields(path))
+		if err != nil || len(remaining) != 0 || resolved == nil {
+			return fmt.Errorf("resolve %q before help rendering: remaining=%v error=%v", command.Path, remaining, err)
+		}
+		if err := renderCommandHelp(resolved); err != nil {
+			return fmt.Errorf("render %q help: %w", command.Path, err)
+		}
+	}
+	return nil
+}
+
+func renderCommandHelp(command *cobra.Command) (err error) {
+	var stdout, stderr bytes.Buffer
+	command.InitDefaultHelpFlag()
+	command.SetOut(&stdout)
+	command.SetErr(&stderr)
+	defer func() {
+		if recovered := recover(); recovered != nil {
+			err = fmt.Errorf("help renderer panicked: %v", recovered)
+		}
+	}()
+	command.HelpFunc()(command, []string{})
+	if stderr.Len() > 0 {
+		return fmt.Errorf("help renderer wrote an error: %s", strings.TrimSpace(stderr.String()))
+	}
+	if stdout.Len() == 0 {
+		return fmt.Errorf("help renderer produced empty output")
+	}
+	return nil
+}
+
 func readSnapshot(path string) (interfacesnapshot.Snapshot, error) {
 	file, err := os.Open(filepath.Clean(path))
 	if err != nil {
@@ -191,5 +280,5 @@ func readSnapshot(path string) (interfacesnapshot.Snapshot, error) {
 func printUsage(w io.Writer) {
 	fmt.Fprintln(w, "usage:")
 	fmt.Fprintln(w, "  interface-snapshot generate [--output FILE]")
-	fmt.Fprintln(w, "  interface-snapshot compare --current FILE [--base FILE] [--stable FILE]")
+	fmt.Fprintln(w, "  interface-snapshot compare --current FILE [--base FILE] [--stable FILE] [--approved-flag-migrations FILE --candidate-flag-migrations FILE]")
 }

cmd/interface-snapshot/main_test.go

@@ -15,11 +15,16 @@ package main
 
 import (
 	"bytes"
+	"errors"
+	"io"
 	"os"
 	"path/filepath"
+	"strings"
 	"testing"
 
 	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/interfacesnapshot"
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/testseam"
+	"github.com/spf13/cobra"
 )
 
 func TestCrossPlatformCoverageRunGenerateCapturesActualRootOffline(t *testing.T) {
@@ -56,6 +61,24 @@ func TestCrossPlatformCoverageRunGenerateCapturesActualRootOffline(t *testing.T)
 	}
 }
 
+func TestCrossPlatformCoverageRunGenerateRejectsHelpRenderingFailure(t *testing.T) {
+	testseam.Swap(t, &newRootCommand, func() *cobra.Command {
+		root := &cobra.Command{Use: "dws"}
+		root.SetHelpFunc(func(command *cobra.Command, _ []string) {
+			_, _ = io.WriteString(command.ErrOrStderr(), "injected help failure")
+		})
+		return root
+	})
+
+	var stdout, stderr bytes.Buffer
+	if exitCode := run([]string{"generate"}, &stdout, &stderr); exitCode != 2 {
+		t.Fatalf("run(generate) exit=%d stderr=%s", exitCode, stderr.String())
+	}
+	if !strings.Contains(stderr.String(), "injected help failure") {
+		t.Fatalf("run(generate) stderr=%q", stderr.String())
+	}
+}
+
 func TestCrossPlatformCoverageRunCompareUsesBothSnapshotInputsAndExitCode(t *testing.T) {
 	current := commandSnapshot("dws")
 	mergeBase := commandSnapshot("dws")
@@ -83,6 +106,368 @@ func TestCrossPlatformCoverageRunCompareUsesBothSnapshotInputsAndExitCode(t *tes
 	}
 }
 
+func TestCrossPlatformCoverageRunCompareEnforcesBaseOwnedFlagMigrationLifecycle(t *testing.T) {
+	dir := t.TempDir()
+	before := flagMigrationSnapshot(false)
+	after := flagMigrationSnapshot(true)
+	currentPath := writeSnapshot(t, dir, "current.json", after)
+	basePath := writeSnapshot(t, dir, "base.json", before)
+	stablePath := writeSnapshot(t, dir, "stable.json", before)
+	approvedPath := writeManifest(t, dir, "approved.json", flagMigrationManifestJSON("pending"))
+	candidatePath := writeManifest(t, dir, "candidate.json", flagMigrationManifestJSON("consumed"))
+
+	var stdout, stderr bytes.Buffer
+	exitCode := run([]string{
+		"compare",
+		"--current", currentPath,
+		"--base", basePath,
+		"--stable", stablePath,
+		"--approved-flag-migrations", approvedPath,
+		"--candidate-flag-migrations", candidatePath,
+	}, &stdout, &stderr)
+	if exitCode != 0 {
+		t.Fatalf("exact base-owned migration exit=%d stderr=%s", exitCode, stderr.String())
+	}
+	if !bytes.Contains(stdout.Bytes(), []byte(`"compatible": true`)) {
+		t.Fatalf("exact migration report is not compatible:\n%s", stdout.String())
+	}
+
+	stdout.Reset()
+	stderr.Reset()
+	emptyApproved := writeManifest(t, dir, "empty-approved.json", `{"version":1,"migrations":[]}`)
+	exitCode = run([]string{
+		"compare",
+		"--current", currentPath,
+		"--base", basePath,
+		"--stable", stablePath,
+		"--approved-flag-migrations", emptyApproved,
+		"--candidate-flag-migrations", candidatePath,
+	}, &stdout, &stderr)
+	if exitCode != 2 || !strings.Contains(stderr.String(), "must start pending") {
+		t.Fatalf("candidate self-approval exit=%d stdout=%s stderr=%s", exitCode, stdout.String(), stderr.String())
+	}
+}
+
+func TestCrossPlatformCoverageRunCompareRequiresBothFlagMigrationInputs(t *testing.T) {
+	dir := t.TempDir()
+	currentPath := writeSnapshot(t, dir, "current.json", commandSnapshot("dws"))
+	basePath := writeSnapshot(t, dir, "base.json", commandSnapshot("dws"))
+	approvedPath := writeManifest(t, dir, "approved.json", `{"version":1,"migrations":[]}`)
+
+	var stdout, stderr bytes.Buffer
+	exitCode := run([]string{
+		"compare",
+		"--current", currentPath,
+		"--base", basePath,
+		"--approved-flag-migrations", approvedPath,
+	}, &stdout, &stderr)
+	if exitCode != 2 || !strings.Contains(stderr.String(), "must be provided together") {
+		t.Fatalf("one-sided migration input exit=%d stdout=%s stderr=%s", exitCode, stdout.String(), stderr.String())
+	}
+}
+
+func TestCrossPlatformCoverageRunCompareRequiresBothReferencesForFlagMigrations(t *testing.T) {
+	dir := t.TempDir()
+	currentPath := writeSnapshot(t, dir, "current.json", commandSnapshot("dws"))
+	basePath := writeSnapshot(t, dir, "base.json", commandSnapshot("dws"))
+	stablePath := writeSnapshot(t, dir, "stable.json", commandSnapshot("dws"))
+	approvedPath := writeManifest(t, dir, "approved.json", `{"version":1,"migrations":[]}`)
+	candidatePath := writeManifest(t, dir, "candidate.json", `{"version":1,"migrations":[]}`)
+
+	tests := []struct {
+		name string
+		args []string
+	}{
+		{
+			name: "missing stable",
+			args: []string{"--base", basePath},
+		},
+		{
+			name: "missing base",
+			args: []string{"--stable", stablePath},
+		},
+	}
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			args := []string{"compare", "--current", currentPath}
+			args = append(args, test.args...)
+			args = append(args,
+				"--approved-flag-migrations", approvedPath,
+				"--candidate-flag-migrations", candidatePath,
+			)
+			var stdout, stderr bytes.Buffer
+			exitCode := run(args, &stdout, &stderr)
+			if exitCode != 2 || !strings.Contains(stderr.String(), "requires both --base and --stable") {
+				t.Fatalf("one-reference migration compare exit=%d stdout=%s stderr=%s", exitCode, stdout.String(), stderr.String())
+			}
+		})
+	}
+}
+
+func TestCrossPlatformCoverageRunPrintsUsageForMissingAndUnknownCommands(t *testing.T) {
+	tests := []struct {
+		name       string
+		args       []string
+		wantStderr []string
+	}{
+		{
+			name:       "missing command",
+			args:       nil,
+			wantStderr: []string{"usage:", "interface-snapshot generate", "--approved-flag-migrations"},
+		},
+		{
+			name:       "unknown command",
+			args:       []string{"unknown"},
+			wantStderr: []string{`unknown command "unknown"`, "usage:", "interface-snapshot compare"},
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			var stdout, stderr bytes.Buffer
+			if exitCode := run(test.args, &stdout, &stderr); exitCode != 2 {
+				t.Fatalf("run(%v) exit=%d, want 2", test.args, exitCode)
+			}
+			if stdout.Len() != 0 {
+				t.Fatalf("run(%v) unexpectedly wrote stdout: %s", test.args, stdout.String())
+			}
+			for _, want := range test.wantStderr {
+				if !strings.Contains(stderr.String(), want) {
+					t.Errorf("run(%v) stderr missing %q:\n%s", test.args, want, stderr.String())
+				}
+			}
+		})
+	}
+}
+
+func TestCrossPlatformCoverageRunRejectsInvalidSubcommandArguments(t *testing.T) {
+	tests := []struct {
+		name string
+		args []string
+		want string
+	}{
+		{name: "generate unknown flag", args: []string{"generate", "--unknown"}, want: "flag provided but not defined"},
+		{name: "generate positional", args: []string{"generate", "unexpected"}, want: "generate accepts no positional arguments"},
+		{name: "compare unknown flag", args: []string{"compare", "--unknown"}, want: "flag provided but not defined"},
+		{name: "compare positional", args: []string{"compare", "unexpected"}, want: "compare accepts no positional arguments"},
+		{name: "compare missing current", args: []string{"compare", "--base", "base.json"}, want: "compare requires --current"},
+		{name: "compare missing reference", args: []string{"compare", "--current", "current.json"}, want: "compare requires --base, --stable, or both"},
+	}
+
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			var stdout, stderr bytes.Buffer
+			if exitCode := run(test.args, &stdout, &stderr); exitCode != 2 {
+				t.Fatalf("run(%v) exit=%d, want 2", test.args, exitCode)
+			}
+			if !strings.Contains(stderr.String(), test.want) {
+				t.Fatalf("run(%v) stderr missing %q:\n%s", test.args, test.want, stderr.String())
+			}
+		})
+	}
+}
+
+func TestCrossPlatformCoverageRunGenerateRejectsUnsafeOutputPath(t *testing.T) {
+	outputDirectory := t.TempDir()
+	var stdout, stderr bytes.Buffer
+	exitCode := run([]string{"generate", "--output", outputDirectory}, &stdout, &stderr)
+	if exitCode != 2 || !strings.Contains(stderr.String(), "create snapshot") {
+		t.Fatalf("directory output exit=%d stdout=%s stderr=%s", exitCode, stdout.String(), stderr.String())
+	}
+}
+
+func TestCrossPlatformCoverageRunGenerateReportsTemporaryDirectoryFailure(t *testing.T) {
+	missingTempRoot := filepath.Join(t.TempDir(), "missing")
+	for _, name := range []string{"TMPDIR", "TMP", "TEMP"} {
+		t.Setenv(name, missingTempRoot)
+	}
+
+	var stdout, stderr bytes.Buffer
+	exitCode := run([]string{"generate"}, &stdout, &stderr)
+	if exitCode != 2 || !strings.Contains(stderr.String(), "create isolated home") {
+		t.Fatalf("invalid temporary root exit=%d stdout=%s stderr=%s", exitCode, stdout.String(), stderr.String())
+	}
+}
+
+func TestCrossPlatformCoverageRunCompareReportsSnapshotReadFailures(t *testing.T) {
+	dir := t.TempDir()
+	validPath := writeSnapshot(t, dir, "valid.json", commandSnapshot("dws"))
+	missingPath := filepath.Join(dir, "missing.json")
+	tests := []struct {
+		name string
+		args []string
+		want string
+	}{
+		{
+			name: "current",
+			args: []string{"compare", "--current", missingPath, "--base", validPath},
+			want: "read current snapshot",
+		},
+		{
+			name: "main",
+			args: []string{"compare", "--current", validPath, "--base", missingPath},
+			want: "read main/development baseline snapshot",
+		},
+		{
+			name: "stable",
+			args: []string{"compare", "--current", validPath, "--stable", missingPath},
+			want: "read stable snapshot",
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			var stdout, stderr bytes.Buffer
+			if exitCode := run(test.args, &stdout, &stderr); exitCode != 2 {
+				t.Fatalf("run(compare) exit=%d, want 2", exitCode)
+			}
+			if !strings.Contains(stderr.String(), test.want) {
+				t.Fatalf("stderr missing %q:\n%s", test.want, stderr.String())
+			}
+		})
+	}
+}
+
+func TestCrossPlatformCoverageRunCompareReportsEachManifestReadFailure(t *testing.T) {
+	dir := t.TempDir()
+	snapshotPath := writeSnapshot(t, dir, "snapshot.json", commandSnapshot("dws"))
+	validManifest := writeManifest(t, dir, "valid-manifest.json", `{"version":1,"migrations":[]}`)
+	invalidManifest := writeManifest(t, dir, "invalid-manifest.json", `{`)
+	tests := []struct {
+		name      string
+		approved  string
+		candidate string
+		want      string
+	}{
+		{
+			name:      "approved manifest",
+			approved:  invalidManifest,
+			candidate: validManifest,
+			want:      "read approved flag migrations",
+		},
+		{
+			name:      "candidate manifest",
+			approved:  validManifest,
+			candidate: invalidManifest,
+			want:      "read candidate flag migrations",
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			var stdout, stderr bytes.Buffer
+			exitCode := run([]string{
+				"compare",
+				"--current", snapshotPath,
+				"--base", snapshotPath,
+				"--stable", snapshotPath,
+				"--approved-flag-migrations", test.approved,
+				"--candidate-flag-migrations", test.candidate,
+			}, &stdout, &stderr)
+			if exitCode != 2 || !strings.Contains(stderr.String(), test.want) {
+				t.Fatalf("%s exit=%d stdout=%s stderr=%s", test.name, exitCode, stdout.String(), stderr.String())
+			}
+		})
+	}
+}
+
+func TestCrossPlatformCoverageRunCompareReportsOutputFailure(t *testing.T) {
+	dir := t.TempDir()
+	snapshotPath := writeSnapshot(t, dir, "snapshot.json", commandSnapshot("dws"))
+	var stderr bytes.Buffer
+	exitCode := run([]string{
+		"compare",
+		"--current", snapshotPath,
+		"--base", snapshotPath,
+	}, failingWriter{}, &stderr)
+	if exitCode != 2 || !strings.Contains(stderr.String(), "write comparison report") {
+		t.Fatalf("comparison output failure exit=%d stderr=%s", exitCode, stderr.String())
+	}
+}
+
+func TestCrossPlatformCoverageReadHelpersRejectMissingAndInvalidInputs(t *testing.T) {
+	dir := t.TempDir()
+	missingPath := filepath.Join(dir, "missing.json")
+	invalidPath := filepath.Join(dir, "invalid.json")
+	if err := os.WriteFile(invalidPath, []byte(`{`), 0o600); err != nil {
+		t.Fatalf("write invalid fixture: %v", err)
+	}
+
+	if _, err := readSnapshot(missingPath); err == nil {
+		t.Fatal("readSnapshot(missing) unexpectedly succeeded")
+	}
+	if _, err := readSnapshot(invalidPath); err == nil {
+		t.Fatal("readSnapshot(invalid) unexpectedly succeeded")
+	}
+	if _, err := readFlagMigrationManifest(missingPath); err == nil {
+		t.Fatal("readFlagMigrationManifest(missing) unexpectedly succeeded")
+	}
+	if _, err := readFlagMigrationManifest(invalidPath); err == nil {
+		t.Fatal("readFlagMigrationManifest(invalid) unexpectedly succeeded")
+	}
+}
+
+func TestCrossPlatformCoverageValidateHelpRenderingReportsResolveError(t *testing.T) {
+	root := &cobra.Command{Use: "dws"}
+	err := validateHelpRendering(root, commandSnapshot("dws missing"))
+	if err == nil || !strings.Contains(err.Error(), `resolve "dws missing" before help rendering`) {
+		t.Fatalf("validateHelpRendering resolve error = %v", err)
+	}
+}
+
+func TestCrossPlatformCoverageValidateHelpRenderingReportsTemplateError(t *testing.T) {
+	root := &cobra.Command{Use: "dws"}
+	root.SetHelpTemplate(`{{index .Commands 99}}`)
+	err := validateHelpRendering(root, commandSnapshot("dws"))
+	if err == nil || !strings.Contains(err.Error(), `render "dws" help`) {
+		t.Fatalf("validateHelpRendering template error = %v", err)
+	}
+}
+
+func TestCrossPlatformCoverageValidateHelpRenderingRecoversTemplatePanic(t *testing.T) {
+	root := &cobra.Command{Use: "dws"}
+	root.SetHelpTemplate("{{")
+	err := validateHelpRendering(root, commandSnapshot("dws"))
+	if err == nil || !strings.Contains(err.Error(), `render "dws" help`) {
+		t.Fatalf("validateHelpRendering template panic = %v", err)
+	}
+}
+
+func TestCrossPlatformCoverageValidateHelpRenderingRejectsCustomHelpStderr(t *testing.T) {
+	root := &cobra.Command{Use: "dws"}
+	root.SetHelpFunc(func(command *cobra.Command, _ []string) {
+		_, _ = io.WriteString(command.ErrOrStderr(), "injected help failure")
+	})
+	err := validateHelpRendering(root, commandSnapshot("dws"))
+	if err == nil || !strings.Contains(err.Error(), "injected help failure") {
+		t.Fatalf("validateHelpRendering custom stderr = %v", err)
+	}
+}
+
+func TestCrossPlatformCoverageValidateHelpRenderingRejectsEmptyOutput(t *testing.T) {
+	root := &cobra.Command{Use: "dws"}
+	root.SetHelpFunc(func(*cobra.Command, []string) {})
+	err := validateHelpRendering(root, commandSnapshot("dws"))
+	if err == nil || !strings.Contains(err.Error(), "empty") {
+		t.Fatalf("validateHelpRendering empty output = %v", err)
+	}
+}
+
+func TestCrossPlatformCoverageValidateHelpRenderingAcceptsNormalOutput(t *testing.T) {
+	root := &cobra.Command{Use: "dws", Short: "root command"}
+	if err := validateHelpRendering(root, commandSnapshot("dws")); err != nil {
+		t.Fatalf("validateHelpRendering normal output: %v", err)
+	}
+}
+
+type failingWriter struct{}
+
+func (failingWriter) Write([]byte) (int, error) {
+	return 0, errors.New("injected write failure")
+}
+
+var _ io.Writer = failingWriter{}
+
 func commandSnapshot(paths ...string) interfacesnapshot.Snapshot {
 	commands := make([]interfacesnapshot.Command, 0, len(paths))
 	for _, path := range paths {
@@ -120,6 +505,68 @@ func writeSnapshot(t *testing.T, dir, name string, snapshot interfacesnapshot.Sn
 	return path
 }
 
+func writeManifest(t *testing.T, dir, name, contents string) string {
+	t.Helper()
+	path := filepath.Join(dir, name)
+	if err := os.WriteFile(path, []byte(contents), 0o600); err != nil {
+		t.Fatalf("write %s: %v", path, err)
+	}
+	return path
+}
+
+func flagMigrationSnapshot(after bool) interfacesnapshot.Snapshot {
+	legacy := interfacesnapshot.Flag{
+		Name:      "legacy-id",
+		Shorthand: "l",
+		Type:      "string",
+		Default:   "",
+		NoOpt:     "auto",
+		Required:  true,
+	}
+	flags := []interfacesnapshot.Flag{legacy}
+	if after {
+		legacy.Required = false
+		legacy.Hidden = true
+		legacy.AliasOf = "message-id"
+		flags = []interfacesnapshot.Flag{
+			legacy,
+			{Name: "message-id", Type: "string", Default: "", Required: true},
+		}
+	}
+	return interfacesnapshot.Snapshot{
+		SchemaVersion: interfacesnapshot.SchemaVersion,
+		Rules: interfacesnapshot.Rules{
+			ExcludedCommandSubtrees: []string{},
+			ExcludedFlags:           []string{},
+		},
+		Commands: []interfacesnapshot.Command{
+			{Path: "dws", Runnable: true, Aliases: []string{}, LocalFlags: []interfacesnapshot.Flag{}, InheritedFlags: []interfacesnapshot.Flag{}},
+			{Path: "dws chat send", Runnable: true, Aliases: []string{}, LocalFlags: flags, InheritedFlags: []interfacesnapshot.Flag{}},
+		},
+	}
+}
+
+func flagMigrationManifestJSON(state string) string {
+	return strings.Replace(`{
+  "version": 1,
+  "migrations": [{
+    "command": "dws chat send",
+    "legacy": {
+      "name": "legacy-id",
+      "before": {"present": true, "type": "string", "required": true, "shorthand": "l", "no_opt": "auto", "scope": "local"},
+      "after": {"present": true, "type": "string", "hidden": true, "shorthand": "l", "no_opt": "auto", "scope": "local", "alias_of": "message-id"}
+    },
+    "canonical": {
+      "name": "message-id",
+      "before": {"present": false},
+      "after": {"present": true, "type": "string", "required": true, "scope": "local"}
+    },
+    "state": "STATE",
+    "reason": "reviewed exact migration"
+  }]
+}`, "STATE", state, 1)
+}
+
 func hasFlag(flags []interfacesnapshot.Flag, name, flagType string) bool {
 	for _, flag := range flags {
 		if flag.Name == name && flag.Type == flagType {

docs/ci-pr-gates.md

@@ -157,15 +157,33 @@ expected to repeat every CI job locally:
 ```sh
 make build
 make policy
-make interface-integrity
-make authoritative-interface-integrity BASE_REF=<merge-base>
-make schema-compatibility BASE_REF=<merge-base>
+make interface-integrity BASE_REF=<merge-base> STABLE_REF=<stable-tag> CANDIDATE_REF=<candidate-sha>
+make schema-compatibility BASE_REF=<merge-base> STABLE_REF=<stable-tag> CANDIDATE_REF=<candidate-sha>
 make skill-command-integrity
 make cli-smoke
 make mock-mcp-smoke
 go test -v -count=1 ./pkg/editiontest/...
 ```
 
+CI 先解析并核对精确的 merge-base、最近可达且未撤回的 stable GA tag 和已提交的 candidate
+SHA，再调用 `make authoritative-interface-integrity`。本地 `make interface-integrity`
+与该 CI target 都只委托给同一个 modern authoritative wrapper，不存在第二个比较
+入口。省略 `BASE_REF` 时本地 target 默认比较 `origin/main`，省略 `STABLE_REF` 时自动
+选择该 base 可达且未撤回的最近 stable GA tag，省略 `CANDIDATE_REF` 时比较已提交的 `HEAD`。
+需要逐字复现某次 CI 时，应显式传入该次运行记录的 merge-base、stable tag 和
+candidate SHA。
+
+`make update-interface-baseline` / `make reset-interface-baseline` 只维护
+`test/fixtures/cli-interface-baseline.txt` 这一份非权威 CLI Smoke fixture。底层旧
+`check-interface-baseline.sh` 不再作为本地或 CI 的兼容性审批入口，也不能用于批准
+flag 迁移。
+
+Schema compatibility 使用同一组 base、stable、candidate refs 和同一份 base-owned flag
+migration ledger。merge-base-owned checker 分别规范化 merge-base 与 stable 的完整
+Schema，并让 candidate 对两份历史 contract 独立执行检查；它只把已通过 Interface
+lifecycle 的 exact rename 规范化到当前历史副本，不会维护第二份 allowlist，也不会
+放宽其他 Schema 历史字段。
+
 For an exact CHANGELOG-only branch:
 
 ```sh
@@ -186,11 +204,24 @@ 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
-a breaking change by editing a fixture. Schema additions are allowed;
-historical products, tools, parameters, mappings, positional execution fields,
-constraints, and safety semantics remain protected.
+CLI 兼容检查只使用 modern Interface Snapshot 这一处权威比较 seam，并从 PR
+merge-base 和最近的可达 stable release 生成权威快照。本治理机制合入后，
+merge-base 拥有生成器、比较器和已审批迁移清单，因此 candidate 不能通过修改
+helper、fixture 或在同一 PR 新增 self-approval 记录来放行 breaking change。首次
+bootstrap 仍由 merge-base 已有的 modern helper 做无豁免比较，并只接受 candidate
+提交中的规范空清单；完整边界见下方治理文档。
+
+精确的两阶段 flag 迁移生命周期见
+[CLI flag 兼容迁移治理](cli-interface-flag-migrations.md)。治理 PR 只能在
+surface 未变化时新增 `pending`；后续产品 PR 达到审批的精确 surface 后，才能
+消费 base-owned 记录并改为 `consumed`。在 main 与 stable 都达到 after 状态前
+必须保留该回执，之后再由单独 PR 清理。机制只放行记录中的 legacy
+visible-to-hidden，以及 canonical required 新增或提升；删除、type、scope、
+shorthand、no-opt 和任何无关漂移仍然阻塞。Schema 可以新增；历史 product、
+tool、parameter、mapping、positional execution、constraint 与 safety 语义继续
+受保护。`alias_of` 只是一项由 `FlagSpec.Aliases` 产生的框架关系证据，不是 payload
+等价证明；产品 PR 仍须证明 canonical 与 legacy 的最终运行 payload 等价并在 transport
+前拒绝冲突输入。当前迁移清单为空，不授权 PR #904。
 
 ## Required GitHub repository settings
 

docs/cli-interface-flag-migrations.md

@@ -0,0 +1,176 @@
+# CLI flag 兼容迁移治理
+
+本文定义一种受控迁移：保留旧 flag 的可执行兼容性，但把它从 Help 与 Agent Schema 中隐藏，并将新的规范 flag 提升为必填。它只解决这一种精确变更，不是通用 breaking-change 豁免。
+
+同名 flag 的精确类型迁移属于另一类评审机制，只能进入
+`internal/interfacesnapshot/reviewed.go` 与 legacy smoke helper 的镜像表；flag rename
+只能进入本文的 JSON lifecycle ledger。一项迁移不得跨两种机制组合授权。
+
+## 唯一比较入口与信任边界
+
+PR 与本地兼容性审批的唯一权威比较入口是 modern Interface Snapshot：
+
+- `cmd/interface-snapshot` 生成和比较快照；
+- `internal/interfacesnapshot` 实现兼容规则和迁移生命周期；
+- `scripts/policy/check-command-compatibility.sh` 组装 candidate、PR merge-base 和最近可达且未撤回的 stable GA 三份快照；
+- `scripts/policy/check-authoritative-interface-baselines.sh` 只保留为 Makefile 的兼容包装，不再维护第二套判断逻辑。
+
+紧随其后的 Schema compatibility 不是第二份审批清单。它从同一 merge-base-owned
+ledger 和同一组三方 Interface Snapshot 取得已经完成 lifecycle 校验的
+authorization。merge-base-owned checker 会分别规范化 merge-base 与 stable 的完整
+Schema，并让 candidate 对两份历史 contract 独立执行检查；授权的 flag rename 只会
+精确投影到当前被检查的历史副本。candidate 不能为 CLI 与 Schema 分别提供两套例外。
+
+`make interface-integrity` 也调用上述 authoritative wrapper；默认 base 为
+`origin/main`，stable 可由包装脚本自动解析，candidate 默认为已提交的 `HEAD`。旧
+`scripts/policy/check-interface-baseline.sh` 只供
+`make update-interface-baseline` / `make reset-interface-baseline` 维护非权威 CLI
+Smoke fixture，不参与迁移审批。
+
+直接调用 `interface-snapshot compare` 时，只要提供 migration manifest 参数，就必须
+同时提供 `--base` 与 `--stable`；核心 lifecycle 也拒绝缺失 stable 的非空清单，避免
+调用方因漏传历史参考而提前清理 consumed receipt。
+
+PR merge-base 同时拥有快照生成器、比较器和已审批清单。门禁用这套 base-owned helper 检查同一个已提交 candidate revision、merge-base 与 stable，candidate 不能通过修改自己的 Go 比较 helper 来放宽规则。candidate 中的清单只参与迁移状态流转，不能批准同一个 PR 引入的接口变化。首次引入本机制时，merge-base 尚无迁移解析器；bootstrap 会用 merge-base 已有的 modern Interface Snapshot 做不带豁免的普通比较，并只接受 candidate 中逐字匹配的空清单，不会让 candidate 新增的 comparator 决定本 PR 是否兼容。bootstrap 无法让旧 helper 证明新治理实现本身正确，因此本治理 PR 的新 parser、lifecycle、launcher 与 hostile tests 仍是必须由真人评审的受保护策略变更；它们合入后才成为后续 PR 的 base-owned authority。
+
+这条边界保护比较规则和审批数据，不是任意代码沙箱。GitHub workflow / launcher 的变更仍由仓库保护规则和真人评审负责；candidate Cobra 构建也会执行 candidate 代码，因此对同一 runner 上的主动恶意代码，需要独立进程或文件系统隔离，不能把本门禁描述成已经解决。
+
+已审批清单固定为：
+
+```text
+scripts/policy/interface-migrations/approved-flag-migrations-v1.json
+```
+
+清单使用严格 JSON 解析：版本、字段名大小写、JSON 值类型、命令路径和 flag 名都必须精确；拒绝重复键、未知键、scalar `null` 与尾随 JSON 值，`reason` 不能为空；禁止 `*`、`?`、前缀规则或其他 wildcard。当前清单为空，因此本治理 PR **不授权 PR #904 或任何产品接口变化**。
+
+## 两阶段迁移与回执清理
+
+每条迁移以 `(command, legacy flag, canonical flag)` 为唯一精确键，并经历以下生命周期：
+
+| 阶段 | PR 可以做什么 | 必须满足的快照状态 |
+|---|---|---|
+| 1. 治理审批 | 新增 `state: pending` 的精确记录；不得在同一个 PR 修改产品 surface | candidate 和 merge-base 都与记录中的 `before` 完全一致；该记录不改变 stable 的判断 |
+| 2. 产品迁移 | merge-base 已拥有 `pending` 后，按记录一次性切到精确 `after`，并把记录改为 `state: consumed` | legacy 仍存在但由 visible 变 hidden，且声明 `alias_of`；canonical 达到记录的必填状态 |
+| 3. 保留回执 | 产品 PR 合入后，如果 stable 仍是 `before`，继续保留 `consumed` | merge-base 或 stable 仍有任一份尚未达到 `after` |
+| 4. 单独清理 | 当 merge-base 和 stable 都已经是 `after`，在后续 PR 删除该记录 | 两份参考快照均精确匹配 `after`；继续保留过期回执会被门禁拒绝 |
+
+因此，新增 `pending` 和修改产品 surface 不能发生在同一个 PR；candidate 自己新增的记录不能 self-approve。迁移也不能部分执行：legacy、canonical、`alias_of` 或状态只要有一项不匹配，门禁即失败。
+
+下面只是清单结构示例，不代表已审批命令；实际字段必须从 Interface Snapshot 核对：
+
+```json
+{
+  "version": 1,
+  "migrations": [
+    {
+      "command": "dws chat message recall",
+      "legacy": {
+        "name": "msg-id",
+        "before": {
+          "present": true,
+          "type": "string",
+          "required": true,
+          "scope": "local"
+        },
+        "after": {
+          "present": true,
+          "type": "string",
+          "hidden": true,
+          "scope": "local",
+          "alias_of": "message-id"
+        }
+      },
+      "canonical": {
+        "name": "message-id",
+        "before": { "present": false },
+        "after": {
+          "present": true,
+          "type": "string",
+          "required": true,
+          "scope": "local"
+        }
+      },
+      "state": "pending",
+      "reason": "保留旧 argv 兼容性，并将规范 flag 设为唯一可见入口"
+    }
+  ]
+}
+```
+
+产品迁移 PR 必须保持同一条记录的命令、flag、before/after 和 reason 不变，只把 `pending` 改成 `consumed`。
+
+## `alias_of` 是框架来源的受评审关系证据
+
+`alias_of` 不是 Schema 同义词、参数概念词典或任意文字声明。它只能由 `FlagSpec.Aliases` 写入，并与内部 origin `corecmd.flag_spec_aliases.v1` 成对出现；每次 Interface Integrity 都会在已提交的 detached candidate 上执行源码门禁，禁止其他生产文件写入或复刻这些 evidence token。Interface Snapshot 会验证：
+
+- legacy 与 canonical 位于同一个可执行命令；
+- canonical flag 确实存在；
+- legacy 与 canonical 类型一致；
+- legacy 不是指向自身，也不存在 alias chain；
+- legacy 的 after 状态精确指向该记录中的 canonical flag。
+
+通过命令框架声明 `FlagSpec.Aliases` 时，框架会自动注册隐藏的兼容 flag，并写入两项 relation annotation；仅手写 `alias_of`、伪造 origin、重复值或不精确值都会让快照生成失败。不要用 Schema overlay、迁移清单或手写 Cobra annotation 伪造关系。
+
+这项关系证据只证明 legacy/canonical 经过受控框架路径建立关系，不证明最终 transport payload 等价，也不会替产品代码实现命令特有的值同步。当前框架还禁止把
+`MarkRequired` 与 `FlagSpec.Aliases` 直接组合，因为 Cobra 的 hard-required
+校验只识别 canonical spelling。若产品迁移同时需要 canonical 的 Cobra required
+标记和 legacy spelling，产品 PR 必须提供明确的运行时方案，并通过 canonical / legacy
+最终 payload 等价、同值输入一致、冲突输入在 transport 前失败、legacy 仍可调用但 Help 隐藏等测试；迁移清单和 relation evidence 都不能替代这些证明。
+
+## 豁免边界
+
+一条 base-owned、状态正确且前后快照精确匹配的记录，只会从普通兼容报告中移除以下两类预期 finding：
+
+1. legacy flag 的 `flag_became_hidden`（visible → hidden）；
+2. canonical flag 的 `required_flag_added`（新增时即必填）或 `flag_became_required`（已有 flag 从可选变必填）。
+
+以下变化仍按普通兼容规则阻塞，不能被迁移记录掩盖：
+
+- 删除 legacy、canonical、命令或其他 flag；
+- flag 类型或迁移记录中的 scope、shorthand、`no_opt` 漂移；
+- `alias_of` 缺失、指向变化或 alias chain；
+- 命令路径及任何无关的阻塞性接口变化；
+- 不精确、部分完成、超出记录范围的 surface 变化。
+
+## Schema 投影边界
+
+Agent-visible command 会把 visible Cobra flag 投影为 Schema parameter，因此合法的
+legacy hidden 迁移会同时表现为历史 parameter 消失，constraint member 也可能从
+legacy 名改为 canonical 名。Schema adapter 只接受已经由三方 Interface Snapshot
+判定为 authorized 的迁移，并按 tool 的精确 `primary_cli_path` 绑定：
+
+- reference 仍处于 `before` 且该 flag 有 Schema surface 时，baseline legacy parameter
+  必须存在，candidate legacy parameter 必须消失，candidate canonical parameter 必须存在；
+  如果 baseline 只有 canonical、没有 legacy，则 adapter 不得借 CLI ledger 提升
+  `required` / `cli_required` 或重写 constraint；
+- rename 前后的 `type`、`property`、`interface_type`、default、format、enum 与
+  `required_when` 必须完全一致；
+- `required` / `cli_required` 只能保持不变或按审批从 `false` 提升为 `true`，禁止降低；
+- constraint 只允许在同一 tool 内按已枚举的 legacy → canonical map 做 member 替换、
+  排序与去重；group kind、非迁移 member 或 group 增删仍然阻塞；
+- 多个 legacy 指向同一 canonical 时，所有历史 parameter signature 必须一致，否则
+  fail closed。
+
+adapter 先构造经过上述验证的历史 contract 副本，再调用原 Schema checker；它不会按
+错误字符串删除 finding。这样既能处理纯 rename，也能阻止“旧 required 参数改名后意外
+变为 optional”或 property 漂移等伪兼容。`consumed` 回执在 merge-base Schema 已经处于
+canonical-only `after` 状态时不需要再次投影；adapter 保持 baseline 不变，由原 checker
+验证 candidate 是否仍与该 canonical contract 兼容。
+
+## 本地验证
+
+先确保 merge-base 和 stable tag 已在本地，然后运行与 CI 相同的权威门禁：
+
+```sh
+make interface-integrity \
+  BASE_REF=<merge-base> \
+  STABLE_REF=<stable-tag> \
+  CANDIDATE_REF=<candidate-sha>
+
+make schema-compatibility \
+  BASE_REF=<merge-base> \
+  STABLE_REF=<stable-tag> \
+  CANDIDATE_REF=<candidate-sha>
+```
+
+`STABLE_REF` 必须解析到从该 merge-base 可达的最高未撤回 stable GA tag；primary checker 会按 release contract 独立核对，不能用任意 after commit 或已撤回版本提前清理回执。包装脚本可以在省略时自动解析。`CANDIDATE_REF` 省略时固定为命令启动时的已提交 `HEAD`；评审和复现 CI 时应显式传入 candidate SHA，避免 surface 与清单来自不同 revision。

docs/shortcut-lark-alignment.md

@@ -70,16 +70,17 @@
 
 > 已落地：minutes `+detail`（✅ smart `+detail`）、minutes `+word-replace`（✅ smart `+replace-batch`，批量+去重）、base `+record-share-link-create`（✅ smart `+record-share-links`，>20 去重+分片+合并）、im `+threads-messages-list`（✅ smart `chat +thread-replies`，list_topic_replies + 投影）、im `+chat-list`（✅ smart `chat +chat-list`）、task `+get-related-tasks`（✅ smart `todo +related-tasks`，三角色并集+去重+投影）。
 
-### im → chat（5）
+### im → chat（7）
 
 | lark 命令 | risk | 保真度差距（钉钉有 tool，缺什么智能） |
 |---|---|---|
-| `+chat-list` ✅ | read | **已建 smart `chat +chat-list`**：`list_all_conversations` + 默认仅群聊 + `--types group/p2p` 当前页过滤 + `--exclude-muted` + page-size/page-token 别名 + openConversationId/name/conversationType 投影。剩余未做：sort/sort-type、bot 身份 p2p 剥离（DWS 无对应身份模型） |
-| `+chat-messages-list` ✅ | read | **已建 smart `chat +chat-messages`**：群/单聊 list_conversation_message_v2 / list_individual_chat_message 互斥 + sender/text/time 投影。剩余未做：reactions 富化、资源下载 |
-| `+chat-search` | read | dws 无群名模糊搜索v2对应 tool(search_common_groups/find 语义不同)，缺 query规范化、mode映射、mute过滤、meta投影 |
+| `+chat-list` ✅ | read | **已建 smart `chat +chat-list`**：`list_all_conversations` + 默认仅群聊 + `--types group/p2p` + `--exclude-muted` + page-size/page-token 别名 + `--page-all/--page-limit` 数字 cursor 自动翻页、跨页去重、合并后类型过滤和完整性 ledger。剩余未做：sort/sort-type、bot 身份 p2p 剥离（DWS 无对应身份模型） |
+| `+chat-messages-list` ✅ | read | **已建 smart `chat +chat-messages`**：群/单聊互斥解析、时间范围、asc/desc、时间边界全量翻页、reaction、资源下载与完整性 ledger |
+| `+chat-search` ✅ | read | **已建 smart `chat +chat-search`**：真实 `search_groups` 关键词搜索 + page-size/page-token 别名 + `--page-all/--page-limit` 不透明 cursor 自动翻页、跨页去重和完整性 ledger。Lark v2 的 member/type/mode/manager/sort 过滤没有可验证的钉钉对应参数，未伪造 |
+| `+flag-list` ✅ | read | **已建 smart `chat +flag-list`**：真实 `list_message_favorites` 的 `items + hasMore + 数字 nextCursor`，支持 page-size/page-token、`--page-all/--page-limit`、跨页去重和完整性 ledger；仅对齐 message favorite，不模拟 Lark Feed thread flag |
 | `+messages-resources-download` | write | dws download-media 走 get_resource_download_url 拿URL，缺分片Range下载/重试/扩展名推断/安全落盘路径校验 |
-| `+messages-search` | read | dws 有 search_messages_by_keyword/by_time_range/by_sender/at_me 多个原子 tool，但各自单点，缺统一多维filter编排+mget+chat上下文富化+跨字段Validate |
-| `+threads-messages-list` ✅ | read | **已建 smart `chat +thread-replies`**：list_topic_replies + sender/text/time 投影。剩余未做：reactions 富化、资源下载 |
+| `+messages-search` ✅ | read | **已建 smart `chat +search-msg`**：统一多维过滤、精确时间范围、asc/desc、cursor 全量翻页、mget 富化、reaction、资源下载与完整性 ledger。剩余差异是 Lark chat 上下文和部分 sender/attachment 类型过滤 |
+| `+threads-messages-list` ✅ | read | **已建 smart `chat +thread-replies`**：支持主消息 ID 自动只读解析 conversation/thread，也支持显式 group + thread/topic ID；list_topic_replies + sender/text/time/reaction/resource 投影 + 下层毫秒级 nextCursor 有界自动翻页、跨页去重、完整性 ledger，以及全量结果 asc/desc。与 Lark 的剩余差异是钉钉底层没有服务端 asc 单页，因此 DWS 的 asc 明确要求 `--page-all`，避免伪全局排序 |
 
 ### task → todo（3）
 

docs/shortcut-public-catalog.json

@@ -1,209 +1,926 @@
 {
-  "generated_at": "2026-08-05T22:43:52.497190",
-  "count": 294,
+  "generated_at": "2026-08-06T21:08:26.697858",
+  "count": 357,
   "results": [
     {
-      "suite": "read",
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+advperm-disable",
+      "risk": "high-risk-write",
+      "status": "reviewed_available",
+      "disposition": "schema_leaf",
+      "semantic_delta": "关闭高级权限会使自定义角色失效，显式发布高风险事实。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+advperm-enable",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "schema_leaf",
+      "semantic_delta": "开启 Base 高级权限总开关的一对一入口。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+attachment-put",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "primary_smart",
+      "semantic_delta": "完整执行 prepare、HTTP PUT、attachment cell 覆盖和读回验证；空写回包只有最终附件可证明时成功。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+attachment-remove",
+      "risk": "high-risk-write",
+      "status": "reviewed_available",
+      "disposition": "primary_smart",
+      "semantic_delta": "清空或按文件名移除附件；若读回缺剩余 fileToken，会在写前明确停止而不破坏其他附件。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+attachment-upload",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "schema_leaf",
+      "semantic_delta": "申请附件 OSS 直传地址和 fileToken；只负责上传准备，不伪造文件已上传。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+base-bootstrap",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "primary_smart",
+      "semantic_delta": "声明式创建 Base、表和字段，字段自动分片，逐层读回验证并报告部分副作用。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+base-copy",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "统一完整复制与仅结构复制意图，并把目标目录映射为可执行参数。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+base-delete",
+      "risk": "high-risk-write",
+      "status": "reviewed_available",
+      "disposition": "schema_leaf",
+      "semantic_delta": "不可逆删除整个 Base 的兼容入口，显式发布高风险确认事实。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
       "service": "aitable",
       "command": "+base-get",
       "risk": "read",
-      "status": "real-ok"
+      "status": "reviewed_available",
+      "disposition": "schema_leaf",
+      "semantic_delta": "保留 Base 目录读取的 Shortcut 兼容入口；底层 leaf 已提供同等读取能力。",
+      "availability": "available"
     },
     {
-      "suite": "read",
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+base-get-primary-doc-id",
+      "risk": "read",
+      "status": "reviewed_available",
+      "disposition": "schema_leaf",
+      "semantic_delta": "已知 base/table/record 时读取主键文档 dentryUuid 的一对一入口。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
       "service": "aitable",
       "command": "+base-list",
       "risk": "read",
-      "status": "real-ok"
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "把最近访问 Base 响应稳定投影为 baseId/baseName 列表，作为未知 baseId 时的浏览入口。",
+      "availability": "available"
     },
     {
-      "suite": "read",
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+base-schema-snapshot",
+      "risk": "read",
+      "status": "reviewed_available",
+      "disposition": "primary_smart",
+      "semantic_delta": "完整读取 Base、tables、fields、views，逐层区分合法空集合与缺失响应契约。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
       "service": "aitable",
       "command": "+base-search",
       "risk": "read",
-      "status": "real-ok"
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "按名称关键词定位 Base，并稳定投影可继续操作的 baseId/baseName。",
+      "availability": "available"
     },
     {
-      "suite": "read",
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+base-update",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "schema_leaf",
+      "semantic_delta": "更新 Base 名称和备注的一对一写入，复用统一确认门。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+chart-delete",
+      "risk": "high-risk-write",
+      "status": "reviewed_available",
+      "disposition": "schema_leaf",
+      "semantic_delta": "删除图表及布局项，显式发布不可逆高风险事实。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
       "service": "aitable",
       "command": "+chart-get",
       "risk": "read",
-      "status": "real-ok"
+      "status": "reviewed_available",
+      "disposition": "schema_leaf",
+      "semantic_delta": "读取指定图表完整配置的一对一入口。",
+      "availability": "available"
     },
     {
-      "suite": "read",
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+chart-share-get",
+      "risk": "read",
+      "status": "reviewed_available",
+      "disposition": "schema_leaf",
+      "semantic_delta": "读取图表分享配置的一对一入口。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+chart-share-update",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "统一开启、关闭和设置图表分享类型。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+chart-update",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "统一更新图表配置或布局，并要求显式 config 输入。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
       "service": "aitable",
       "command": "+chart-widgets-example",
       "risk": "read",
-      "status": "real-ok"
+      "status": "reviewed_available",
+      "disposition": "schema_leaf",
+      "semantic_delta": "返回所有图表类型的 widget config 示例，供构造合法图表配置。",
+      "availability": "available"
     },
     {
-      "suite": "read",
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+dashboard-arrange",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "触发服务端智能布局重排并保留统一写入确认。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
       "service": "aitable",
       "command": "+dashboard-config-example",
       "risk": "read",
-      "status": "real-ok"
+      "status": "reviewed_available",
+      "disposition": "schema_leaf",
+      "semantic_delta": "返回 dashboard config 的服务端结构示例，供后续写入构造合法参数。",
+      "availability": "available"
     },
     {
-      "suite": "read",
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+dashboard-delete",
+      "risk": "high-risk-write",
+      "status": "reviewed_available",
+      "disposition": "schema_leaf",
+      "semantic_delta": "级联删除仪表盘及图表，显式发布不可逆高风险事实。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
       "service": "aitable",
       "command": "+dashboard-get",
       "risk": "read",
-      "status": "real-ok"
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "读取仪表盘详情及 charts summary，作为图表操作导航入口。",
+      "availability": "available"
     },
     {
-      "suite": "read",
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+dashboard-share-get",
+      "risk": "read",
+      "status": "reviewed_available",
+      "disposition": "schema_leaf",
+      "semantic_delta": "读取仪表盘分享配置的一对一入口。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+dashboard-share-update",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "统一开启、关闭和设置仪表盘分享类型。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+dashboard-update",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "schema_leaf",
+      "semantic_delta": "更新仪表盘配置的一对一入口。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+export-data",
+      "risk": "read",
+      "status": "reviewed_available",
+      "disposition": "primary_smart",
+      "semantic_delta": "统一创建导出任务与按 taskId 续等，直到得到可下载的导出结果或明确失败。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+field-delete",
+      "risk": "high-risk-write",
+      "status": "reviewed_available",
+      "disposition": "schema_leaf",
+      "semantic_delta": "不可逆删除字段的一对一入口，显式发布数据丢失风险。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
       "service": "aitable",
       "command": "+field-get",
       "risk": "read",
-      "status": "real-ok"
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "批量获取字段完整类型配置，避免仅返回名称的损失投影。",
+      "availability": "available"
     },
     {
-      "suite": "read",
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+field-update",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "统一字段名称、类型配置和 AI 配置的增量更新输入。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
       "service": "aitable",
       "command": "+find-record",
       "risk": "read",
-      "status": "real-ok"
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "将关键词查询投影为稳定的 recordId/cells 结果，适合先定位再写入。",
+      "availability": "available"
     },
     {
-      "suite": "read",
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+form-delete",
+      "risk": "high-risk-write",
+      "status": "reviewed_available",
+      "disposition": "schema_leaf",
+      "semantic_delta": "不可逆删除表单视图的一对一入口。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+form-field-hide",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "用隐藏/显示布尔语义切换表单字段可见性。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
       "service": "aitable",
       "command": "+form-field-list",
       "risk": "read",
-      "status": "real-ok"
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "列出表单当前字段顺序、可见性、必填和描述配置。",
+      "availability": "available"
     },
     {
-      "suite": "read",
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+form-field-update",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "统一更新表单字段必填状态与描述。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
       "service": "aitable",
       "command": "+form-list",
       "risk": "read",
-      "status": "real-ok"
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "从表的视图中稳定筛出表单视图并返回表单目录。",
+      "availability": "available"
     },
     {
-      "suite": "read",
+      "suite": "semantic",
       "service": "aitable",
       "command": "+form-share-get",
       "risk": "read",
-      "status": "real-ok"
+      "status": "reviewed_available",
+      "disposition": "schema_leaf",
+      "semantic_delta": "读取当前视图分享表单配置的一对一入口。",
+      "availability": "available"
     },
     {
-      "suite": "read",
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+form-share-update",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "统一开启或关闭指定视图的分享表单。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+form-update",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "统一更新表单标题和描述。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+import-data",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "统一将已上传文件导入新表或追加到已有表的目标语义。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+import-upload",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "schema_leaf",
+      "semantic_delta": "申请导入 OSS 直传地址和 importId；只负责上传准备，不伪造导入完成。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
       "service": "aitable",
       "command": "+list-tables",
       "risk": "read",
-      "status": "real-ok"
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "把 Base 目录稳定投影为 tableId/tableName 列表，隐藏非表节点。",
+      "availability": "available"
     },
     {
-      "suite": "read",
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+record-bulk-patch",
+      "risk": "high-risk-write",
+      "status": "reviewed_available",
+      "disposition": "primary_smart",
+      "semantic_delta": "先完整解析显式选择范围和 max-matches，再将同一 cells patch 自动分片写入并逐条读回验证。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+record-delete",
+      "risk": "high-risk-write",
+      "status": "reviewed_available",
+      "disposition": "primary_smart",
+      "semantic_delta": "自动按 100 条分片删除；每批仅在所有目标 recordId 读回均不存在时成功，失败提供 nextOffset。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
       "service": "aitable",
       "command": "+record-history-list",
       "risk": "read",
-      "status": "real-ok"
+      "status": "reviewed_available",
+      "disposition": "schema_leaf",
+      "semantic_delta": "按 recordId 查询变更历史的一对一只读入口。",
+      "availability": "available"
     },
     {
-      "suite": "read",
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+record-primary-doc-create",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "以幂等意图为 primaryDoc 字段创建关联文档。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+record-primary-doc-get",
+      "risk": "read",
+      "status": "reviewed_available",
+      "disposition": "schema_leaf",
+      "semantic_delta": "读取记录主键文档 nodeId 的一对一入口。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
       "service": "aitable",
       "command": "+record-query",
       "risk": "read",
-      "status": "real-ok"
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "统一 ID、筛选、关键词和分页查询，并投影稳定的 records/count/cursor 结果。",
+      "availability": "available"
     },
     {
-      "suite": "read",
+      "suite": "semantic",
       "service": "aitable",
       "command": "+record-query-empty",
       "risk": "read",
-      "status": "real-ok"
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "扫描记录并按用户字段为空的语义过滤空行，避免调用方手工判断 cells。",
+      "availability": "available"
     },
     {
-      "suite": "read",
+      "suite": "semantic",
       "service": "aitable",
       "command": "+record-share-links",
       "risk": "read",
-      "status": "real-ok"
+      "status": "reviewed_available",
+      "disposition": "primary_smart",
+      "semantic_delta": "对超过 20 条 recordId 去重分片调用并合并分享链接及逐批结果。",
+      "availability": "available"
     },
     {
-      "suite": "read",
+      "suite": "semantic",
       "service": "aitable",
       "command": "+record-share-url",
       "risk": "read",
-      "status": "real-ok"
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "将最多 20 个 recordId 统一映射为可分享链接列表。",
+      "availability": "available"
     },
     {
-      "suite": "read",
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+record-update",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "primary_smart",
+      "semantic_delta": "自动按 100 条分片更新并逐条读回验证；失败返回可续跑的 nextOffset。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+record-upsert",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "primary_smart",
+      "semantic_delta": "按 recordId 拆分 create/update，自动按 100 条分片；创建必须返回新 recordId，全部结果读回验证。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+record-upsert-by-key",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "primary_smart",
+      "semantic_delta": "先按唯一 fieldId/value 完整查询，0 条创建、1 条更新、2 条以上停止，并读回验证最终 cells。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
       "service": "aitable",
       "command": "+resolve-base",
       "risk": "read",
-      "status": "real-ok"
+      "status": "reviewed_available",
+      "disposition": "primary_smart",
+      "semantic_delta": "按名称搜索并要求唯一匹配，0 或多结果时显式失败而不猜 baseId。",
+      "availability": "available"
     },
     {
-      "suite": "read",
+      "suite": "semantic",
       "service": "aitable",
       "command": "+resolve-table",
       "risk": "read",
-      "status": "real-ok"
+      "status": "reviewed_available",
+      "disposition": "primary_smart",
+      "semantic_delta": "在 Base 内按名称要求唯一匹配，0 或多结果时显式失败而不猜 tableId。",
+      "availability": "available"
     },
     {
-      "suite": "read",
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+role-create",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "统一角色名称、成员与权限配置的创建输入。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+role-delete",
+      "risk": "high-risk-write",
+      "status": "reviewed_available",
+      "disposition": "schema_leaf",
+      "semantic_delta": "不可逆删除自定义角色的一对一入口。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+role-get",
+      "risk": "read",
+      "status": "reviewed_available",
+      "disposition": "schema_leaf",
+      "semantic_delta": "读取单个角色完整权限配置的一对一入口。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
       "service": "aitable",
       "command": "+role-list",
       "risk": "read",
-      "status": "real-ok"
+      "status": "reviewed_available",
+      "disposition": "schema_leaf",
+      "semantic_delta": "列出 Base 全部自定义角色的一对一入口。",
+      "availability": "available"
     },
     {
-      "suite": "read",
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+role-update",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "按 PATCH 语义增量更新角色配置，未提供字段保持不变。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+section-create",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "在 Base 目录树创建用于组织表和仪表盘的文件夹。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+section-delete",
+      "risk": "high-risk-write",
+      "status": "reviewed_available",
+      "disposition": "schema_leaf",
+      "semantic_delta": "不可逆删除目录文件夹的一对一入口。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
       "service": "aitable",
       "command": "+section-list-empty",
       "risk": "read",
-      "status": "real-ok"
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "扫描目录树并只返回没有子节点的空文件夹。",
+      "availability": "available"
     },
     {
-      "suite": "read",
+      "suite": "semantic",
       "service": "aitable",
       "command": "+section-list-nodes",
       "risk": "read",
-      "status": "real-ok"
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "列出当前版本全部 nsheet 节点，提供表、仪表盘和文件夹统一目录。",
+      "availability": "available"
     },
     {
-      "suite": "read",
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+section-move-node",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "统一移动任意 nsheet 节点到文件夹或 Base 根目录，并可指定位置。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+section-rename",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "schema_leaf",
+      "semantic_delta": "重命名目录文件夹的一对一入口。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+section-reorder",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "按当前父目录语义调整文件夹展示顺序。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+table-copy",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "primary_smart",
+      "semantic_delta": "在没有服务端 table-copy task 的情况下，本地重建安全字段、映射 fieldId，并可分片复制及读回验证全部记录。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+table-delete",
+      "risk": "high-risk-write",
+      "status": "reviewed_available",
+      "disposition": "schema_leaf",
+      "semantic_delta": "不可逆删除数据表的一对一入口，显式发布高风险确认事实。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
       "service": "aitable",
       "command": "+table-get",
       "risk": "read",
-      "status": "real-ok"
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "批量读取表级信息、字段和视图目录，为后续 Schema 与记录操作提供 ID。",
+      "availability": "available"
     },
     {
-      "suite": "read",
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+table-update",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "schema_leaf",
+      "semantic_delta": "统一表名、备注和行命名规则的 PATCH 风格更新。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
       "service": "aitable",
       "command": "+template-search",
       "risk": "read",
-      "status": "real-ok"
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "按名称关键词搜索 AI 表格模板并返回可用于创建的 templateId。",
+      "availability": "available"
     },
     {
-      "suite": "read",
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+url-resolve",
+      "risk": "read",
+      "status": "reviewed_available",
+      "disposition": "primary_smart",
+      "semantic_delta": "严格解析 alidocs AI 表格 URL 的 base/table/view/record ID，并可通过最深层只读接口验证目标存在。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+view-delete",
+      "risk": "high-risk-write",
+      "status": "reviewed_available",
+      "disposition": "schema_leaf",
+      "semantic_delta": "不可逆删除视图的一对一入口，显式发布高风险确认事实。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+view-duplicate",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "schema_leaf",
+      "semantic_delta": "复制现有视图配置并生成新视图的一对一入口。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
       "service": "aitable",
       "command": "+view-get",
       "risk": "read",
-      "status": "real-ok"
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "读取视图完整列顺序、筛选、排序和分组配置。",
+      "availability": "available"
     },
     {
-      "suite": "read",
+      "suite": "semantic",
       "service": "aitable",
       "command": "+view-get-frozen-cols",
       "risk": "read",
-      "status": "real-ok"
+      "status": "reviewed_available",
+      "disposition": "schema_leaf",
+      "semantic_delta": "读取 Grid 视图冻结列数的一对一入口。",
+      "availability": "available"
     },
     {
-      "suite": "read",
+      "suite": "semantic",
       "service": "aitable",
       "command": "+view-get-lock",
       "risk": "read",
-      "status": "real-ok"
+      "status": "reviewed_available",
+      "disposition": "schema_leaf",
+      "semantic_delta": "读取视图锁定状态的一对一入口。",
+      "availability": "available"
     },
     {
-      "suite": "read",
+      "suite": "semantic",
       "service": "aitable",
       "command": "+view-get-row-height",
       "risk": "read",
-      "status": "real-ok"
+      "status": "reviewed_available",
+      "disposition": "schema_leaf",
+      "semantic_delta": "读取视图行高像素值的一对一入口。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+view-lock",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "用默认锁定与 --off 解锁统一表达视图锁状态切换。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+view-preset-apply",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "primary_smart",
+      "semantic_delta": "按精确名称幂等创建或更新视图预设，拒绝同名歧义和类型漂移，并读回校验 config。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+view-set-fill-color-rule",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "以全量覆盖语义设置条件填色规则，并用空数组明确清空。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+view-set-frozen-cols",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "设置冻结列数并以 0 明确表达取消冻结。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+view-set-row-height",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "把行高限制为服务支持的 32/56/88/128 像素档位后写入。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+view-update",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "semantic_adapter",
+      "semantic_delta": "统一视图名称、描述、可见列、筛选、排序和分组配置更新。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+workflow-deploy",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "primary_smart",
+      "semantic_delta": "创建或更新完整 workflow-dsl/v1，强制 valid=true/flowId，并可启用后从 list 验证 RUNNING。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+workflow-disable",
+      "risk": "high-risk-write",
+      "status": "reviewed_available",
+      "disposition": "schema_leaf",
+      "semantic_delta": "禁用工作流会影响业务自动化，显式发布高风险确认事实。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+workflow-enable",
+      "risk": "write",
+      "status": "reviewed_available",
+      "disposition": "schema_leaf",
+      "semantic_delta": "启用指定自动化工作流的一对一入口。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+workflow-get",
+      "risk": "read",
+      "status": "reviewed_available",
+      "disposition": "schema_leaf",
+      "semantic_delta": "读取单个自动化工作流详情的一对一入口。",
+      "availability": "available"
+    },
+    {
+      "suite": "semantic",
+      "service": "aitable",
+      "command": "+workflow-list",
+      "risk": "read",
+      "status": "reviewed_available",
+      "disposition": "schema_leaf",
+      "semantic_delta": "分页列出 Base 自动化工作流的一对一入口。",
+      "availability": "available"
     },
     {
       "suite": "read",
@@ -1964,7 +2681,7 @@
       "risk": "write",
       "status": "reviewed_available",
       "disposition": "primary_smart",
-      "semantic_delta": "一体化创建会话、上传、确认转换并轮询导入结果。",
+      "semantic_delta": "一体化创建会话、上传、确认转换并轮询导入结果；白名单外格式自动改走文件上传原样入库（fallback=upload、converted=false）。",
       "availability": "available"
     },
     {

docs/skill-content-framework.md

@@ -101,5 +101,5 @@ skills/mono/
 ## 7. 变更流程
 
 1. 改 / 增内容 → 更新 `skills/content-qa/mono-multi-coverage.yaml`（coverage 或 omit）  
-2. 跑 `make skill-mono-multi-content`（或 `make policy`）  
+2. 跑 `make skill-mono-multi-content`（该独立门禁不包含在默认 `make policy` 中）
 3. 失败则修内容或更新 reviewed omit（disposition + 原因），**禁止**用安装默认值绕过  

docs/skill-mono-multi-qa.md

@@ -2,7 +2,7 @@
 
 > 对照基准：`skills/mono`（单 skill）。被测主体：`skills/multi`。  
 > 机读合同：`skills/content-qa/mono-multi-coverage.yaml`。  
-> 执行：`make skill-mono-multi-content`（已挂入 `make policy`）。
+> 执行：`make skill-mono-multi-content`（独立门禁；默认 `make policy` 按设计不包含该检查）。
 
 ## 1. 质检矩阵
 
@@ -11,7 +11,8 @@
 | **G1 形状** | 结构 | `skills/multi/*` | 仅 `dingtalk-*`（含必选 `dingtalk-shared`）；每目录有 `SKILL.md` |
 | **G2 结构** | 结构 | 各 `SKILL.md` frontmatter | `name`==目录名；非空 `description`；`category`∈{product,shared}；`requires.bins` 含 `dws` |
 | **G3 覆盖** | 覆盖 | mono `references/products/*` 顶层 stem | 每 stem ∈ `coverage` 或 `omit_coverage`；coverage 目标 skill/refs 存在 |
-| **G4 漂移** | 漂移 | scripts、成对文件、全局协议 | orphan 脚本 ∈ allowlist；paired 一致；全局协议存在或 ∈ `omit_global` |
+| **G4 漂移** | 漂移 | scripts、成对文件、全局协议 | orphan 脚本 ∈ allowlist；paired 内容一致（允许合同声明的布局链接替换）；全局协议存在或 ∈ `omit_global` |
+| **G5 链接** | 可达性 | 合同覆盖的 paired Markdown | 内联相对链接目标文件或目录存在且不逃出仓库；外链、纯锚点和锚点内容不在检查范围 |
 
 已有门禁（继续复用，不替代本矩阵）：`check-skill-commands`、`check-skill-context-budget`、`check-multi-im-skill-chain`、`skill_docs_policy`、whiteboard 成对测试。
 
@@ -37,6 +38,16 @@ orphan_scripts_allowlist:
   - path: dingtalk-misc/scripts/report_received_today.py
     disposition: defer
     reason: "pending report.md reference"
+
+paired_files:
+  - mono: references/products/sheet.md
+    multi: dingtalk-misc/references/sheet.md
+    mode: link-normalized
+    link_substitutions:
+      - mono: "../url-patterns.md"
+        multi: "../../dingtalk-shared/references/url-patterns.md"
+      - mono: "../intent-guide.md"
+        multi: "sheet-intent-guide.md"
 ```
 
 **处置原则**：质检失败 → 修**内容**或更新 reviewed omit；**不**改安装/升级默认。

docs/skill-wukong-align-plan.md

@@ -4,7 +4,7 @@
 > 合同短文：[skill-content-framework.md](skill-content-framework.md)  
 > 质检规格：[skill-mono-multi-qa.md](skill-mono-multi-qa.md)  
 > 机读合同：`skills/content-qa/mono-multi-coverage.yaml`  
-> 门禁：`make skill-mono-multi-content`（已入 `make policy`）
+> 门禁：`make skill-mono-multi-content`（独立门禁；默认 `make policy` 按设计不包含该检查）
 >
 > 撰写 / 收窄 / 质检增补 / 执行：2026-08-05  
 > 工作树：`/Users/john/GolandProjects/open-source/dws-multi-skill-align`  

internal/app/param_alias_e2e_test.go

@@ -39,7 +39,7 @@ func (c *paramAliasCaptureCaller) CallTool(_ context.Context, server, tool strin
 		copyArgs[key] = value
 	}
 	c.calls = append(c.calls, paramAliasToolCall{server: server, tool: tool, args: copyArgs})
-	text := paramAliasResponseForTool(tool)
+	text := c.paramAliasResponseForTool(tool)
 	return &edition.ToolResult{Content: []edition.ContentBlock{{Type: "text", Text: text}}}, nil
 }
 
@@ -48,7 +48,7 @@ func (c *paramAliasCaptureCaller) CallTool(_ context.Context, server, tool strin
 // print the transport result and need an empty object; smart shortcuts that
 // inspect a read response receive the smallest shape that lets their full RunE
 // complete without falling back to a validation error.
-func paramAliasResponseForTool(tool string) string {
+func (c *paramAliasCaptureCaller) paramAliasResponseForTool(tool string) string {
 	switch tool {
 	case "list_calendar_events":
 		return `{"result":{"events":[]}}`
@@ -62,9 +62,22 @@ func paramAliasResponseForTool(tool string) string {
 		return `{"result":[{"name":"Fixture User","userId":"fixture-user","openDingTalkId":"D-fixture-user"}]}`
 	case "list_doc_versions":
 		return `{"result":{"items":[{"version":3}]}}`
+	case "revert_doc_version":
+		return `{"version":3}`
 	case "search_doc_templates":
 		return `{"result":[{"templateId":"fixture-template-id"}]}`
+	case "create_document":
+		return `{"nodeId":"fixture-node"}`
 	case "get_document_content":
+		for index := len(c.calls) - 2; index >= 0; index-- {
+			call := c.calls[index]
+			for _, key := range []string{"jsonml", "markdown"} {
+				if content, ok := call.args[key].(string); ok {
+					encoded, _ := json.Marshal(map[string]any{"revision": 1, key: content})
+					return string(encoded)
+				}
+			}
+		}
 		return `{"revision":1}`
 	default:
 		return `{}`
@@ -173,6 +186,24 @@ func executeParamAliasDryRunE2E(t *testing.T, args ...string) (*pipeline.Context
 	return ctx, preview, append([]executor.Invocation(nil), rejectRunner.attempts...), executeErr
 }
 
+func TestCrossPlatformCoverageFlagListDryRunStopsBeforeReadDispatch(t *testing.T) {
+	_, preview, attempts, err := executeParamAliasDryRunE2E(t,
+		"chat", "+flag-list", "--page-size", "20", "--cursor", "0", "--dry-run",
+	)
+	if err != nil {
+		t.Fatalf("flag-list dry-run error = %v", err)
+	}
+	if len(attempts) != 0 {
+		t.Fatalf("flag-list dry-run crossed dispatch boundary: %#v", attempts)
+	}
+	if !preview.DryRun || preview.Executed || preview.Tool != "list_message_favorites" {
+		t.Fatalf("flag-list dry-run preview = %#v", preview)
+	}
+	if preview.Arguments["cursor"] != float64(0) || preview.Arguments["size"] != "20" {
+		t.Fatalf("flag-list dry-run arguments = %#v", preview.Arguments)
+	}
+}
+
 func executeParamAliasE2E(t *testing.T, caller *paramAliasCaptureCaller, args ...string) (*pipeline.Context, error) {
 	t.Helper()
 	originalArgs := os.Args

internal/app/param_alias_payload_equivalence_test.go

@@ -48,7 +48,7 @@ var paramAliasCompleteCommands = map[string][]string{
 	"chat +feed-group-query-item":              {"chat", "+feed-group-query-item", "--category-id", "7", "--conversation-ids", "fixture-conversation"},
 	"chat +flag-cancel":                        {"chat", "+flag-cancel", "--conversation-id", "fixture-conversation", "--message-id", "message-1", "--yes"},
 	"chat +flag-create":                        {"chat", "+flag-create", "--conversation-id", "fixture-conversation", "--message-id", "message-1", "--yes"},
-	"chat +flag-list":                          {"chat", "+flag-list", "--cursor", "0", "--size", "7"},
+	"chat +flag-list":                          {"chat", "+flag-list", "--cursor", "0", "--page-size", "7"},
 	"chat +messages-combine-forward":           {"chat", "+messages-combine-forward", "--src-conversation-id", "fixture-source", "--msg-ids", "message-1,message-2", "--dest-conversation-id", "fixture-destination", "--yes"},
 	"chat +messages-forward":                   {"chat", "+messages-forward", "--src-conversation-id", "fixture-source", "--msg-id", "message-1", "--dest-conversation-id", "fixture-destination", "--yes"},
 	"chat +messages-forward-topic":             {"chat", "+messages-forward-topic", "--src-msg-id", "message-1", "--src-conversation-id", "fixture-source", "--src-thread-id", "convThread-fixture", "--dest-conversation-id", "fixture-destination", "--yes"},
@@ -123,7 +123,7 @@ var paramAliasCompleteCommands = map[string][]string{
 	"doc +version-list":                        {"doc", "+version-list", "--node", "node-1", "--limit", "7", "--cursor", "cursor-1"},
 	"doc +version-revert":                      {"doc", "+version-revert", "--node", "node-1", "--version", "3", "--yes"},
 	"doc +version-save":                        {"doc", "+version-save", "--node", "node-1", "--yes"},
-	"doc +update":                              {"doc", "+update", "--node", "node-1", "--command", "append", "--content", "fixture body", "--expected-revision", "1", "--yes"},
+	"doc +update":                              {"doc", "+update", "--node", "node-1", "--command", "overwrite", "--content", `["root",{}]`, "--doc-format", "jsonml", "--expected-revision", "1", "--yes"},
 	"doc block insert":                         {"doc", "block", "insert", "--node", "node-1", "--text", "fixture paragraph", "--yes"},
 	"doc block update":                         {"doc", "block", "update", "--node", "node-1", "--block-id", "block-1", "--text", "fixture paragraph", "--yes"},
 	"doc comment create":                       {"doc", "comment", "create", "--node", "node-1", "--content", "fixture comment", "--yes"},
@@ -212,7 +212,7 @@ var paramAliasNewIMCases = []struct {
 	{command: "chat +chat-update", emitted: "open-conversation-id", canonical: "group"},
 	{command: "chat +chat-update", emitted: "title", canonical: "name"},
 	{command: "chat +chat-update", emitted: "new-title", canonical: "name"},
-	{command: "chat +flag-list", emitted: "limit", canonical: "size"},
+	{command: "chat +flag-list", emitted: "limit", canonical: "page-size"},
 	{command: "chat +chat-members-list", emitted: "chat-id", canonical: "conversation-id"},
 	{command: "chat +chat-members-list", emitted: "id", canonical: "conversation-id"},
 	{command: "chat +conversation-set-top", emitted: "open-conversation-id", canonical: "conversation-id"},

internal/app/schema_agent_examples_dry_run_test.go

@@ -296,7 +296,14 @@ func newAgentExampleFiles(t testing.TB, root string) agentExampleFiles {
 			t.Fatalf("write dry-run fixture %s: %v", path, err)
 		}
 	}
-	return agentExampleFiles{root: root, markdown: markdown, json: jsonFile, batch: batch, binary: binary, image: image}
+	return agentExampleFiles{
+		root:     root,
+		markdown: "./" + filepath.Base(markdown),
+		json:     "./" + filepath.Base(jsonFile),
+		batch:    "./" + filepath.Base(batch),
+		binary:   "./" + filepath.Base(binary),
+		image:    "./" + filepath.Base(image),
+	}
 }
 
 func materializeAgentExampleArgv(argv []string, files agentExampleFiles) []string {

internal/app/schema_interface_disposition_contract_test.go

@@ -34,7 +34,12 @@ func TestReviewedRoutedInterfacesReachFinalSchema(t *testing.T) {
 		{
 			canonical: "sheet.range_batch_set_style",
 			mode:      "composite",
-			reason:    "The CLI reads a local batch file and performs multiple sheet/update_range calls with local continue-on-error control; the workflow has no single direct MCP interface.",
+			reason:    "The CLI assembles style cell matrices locally from --ranges or a local batch file and submits them as one sheet/batch_update operations array; no single direct MCP interface represents the wrapper input shape.",
+		},
+		{
+			canonical: "sheet.create_with_data",
+			mode:      "composite",
+			reason:    "Reviewed composite workflow: the command calls sheet/create_workspace_sheet, waits for the new document to become writable, resolves the default worksheet, writes the initial data through sheet/set_range_from_csv or sheet/table_put, reads it back with sheet/get_range_as_csv, and optionally applies sheet/set_cell_range, sheet/update_dimension and sheet/merge_cells; no single pinned RPC represents the workflow.",
 		},
 		{
 			canonical: "sheet.range_read",

internal/app/schema_shortcut_contract_test.go

@@ -16,12 +16,12 @@ import (
 )
 
 const (
-	publicShortcutCount = 294
+	publicShortcutCount = 357
 	// schemaPublishedShortcutCount counts every delivered *.shortcut_* tool,
 	// including hidden leaves such as minutes.shortcut_minutes_search.
-	schemaPublishedShortcutCount = 295
+	schemaPublishedShortcutCount = 358
 	// publiclyDeliveredShortcutCount is the public-catalog subset of that surface.
-	publiclyDeliveredShortcutCount = 294
+	publiclyDeliveredShortcutCount = 357
 )
 
 func TestDeliverySchemaCoversOrExactlyExcludesEveryPublicShortcutContract(t *testing.T) {
@@ -114,7 +114,7 @@ func TestDeliveryShortcutProgressiveQueriesReturnCompleteContracts(t *testing.T)
 
 	product := executeShortcutSchemaQuery(t, "chat")
 	productPayload, _ := product["product"].(map[string]any)
-	if got, want := int(product["count"].(float64)), 180; got != want {
+	if got, want := int(product["count"].(float64)), 187; got != want {
 		t.Fatalf("schema chat count = %d, want %d", got, want)
 	}
 	summaries := schemaContractObjectSlice(productPayload["tools"])
@@ -129,6 +129,75 @@ func TestDeliveryShortcutProgressiveQueriesReturnCompleteContracts(t *testing.T)
 	}
 }
 
+func TestDeliveryDocUpdateShortcutPublishesCompleteConditionalContract(t *testing.T) {
+	leaf := executeShortcutSchemaQuery(t, "--cli-path", "doc +update")
+	if got, want := schemaContractString(leaf["confirmation"]), "user_required"; got != want {
+		t.Fatalf("confirmation = %q, want %q", got, want)
+	}
+	parameters := schemaContractMap(leaf["parameters"])
+	if got, want := len(parameters), 11; got != want {
+		t.Fatalf("parameter count = %d, want %d: %#v", got, want, parameters)
+	}
+	if required, _ := parameters["node"]["required"].(bool); !required {
+		t.Errorf("--node required = %#v, want true", parameters["node"]["required"])
+	}
+	if required, _ := parameters["command"]["required"].(bool); required {
+		t.Errorf("--command required = true, want runtime custom validation")
+	}
+	wantProperties := map[string]string{
+		"node": "node", "doc": "node", "command": "command", "content": "content", "text": "content", "doc-format": "docFormat",
+		"block-id": "blockId", "after-block-id": "afterBlockId", "old": "old", "new": "new",
+		"expected-revision": "expectedRevision",
+	}
+	for name, want := range wantProperties {
+		if got := schemaContractString(parameters[name]["property"]); got != want {
+			t.Errorf("--%s property = %q, want %q", name, got, want)
+		}
+	}
+	for _, name := range []string{"content", "block-id", "after-block-id", "old", "new"} {
+		parameter := parameters[name]
+		if required, _ := parameter["required"].(bool); required {
+			t.Errorf("--%s required = true, want runtime custom validation", name)
+		}
+		if got := schemaContractString(parameter["required_when"]); got != "" {
+			t.Errorf("--%s required_when = %q, want compatibility-safe custom validation", name, got)
+		}
+	}
+	if constraints, exists := leaf["constraints"]; exists && constraints != nil {
+		t.Fatalf("enum-discriminated requirements must not be mispublished as relationship constraints: %#v", constraints)
+	}
+}
+
+func TestDeliveryDocCommentExportImportContractsAreCanonical(t *testing.T) {
+	comment := executeShortcutSchemaQuery(t, "--cli-path", "doc +comment-create")
+	commentParameters := schemaContractMap(comment["parameters"])
+	for _, name := range []string{"node", "content", "selection", "block-id", "start", "end", "selected-text", "mention"} {
+		if _, ok := commentParameters[name]; !ok {
+			t.Errorf("comment-create missing --%s: %#v", name, commentParameters)
+		}
+	}
+	for name, want := range map[string]string{"node": "node", "mention": "mention"} {
+		if got := schemaContractString(commentParameters[name]["property"]); got != want {
+			t.Errorf("comment-create --%s property = %q, want %q", name, got, want)
+		}
+	}
+
+	export := executeShortcutSchemaQuery(t, "--cli-path", "doc +export")
+	exportFormat := schemaContractMap(export["parameters"])["export-format"]
+	if required, _ := exportFormat["required"].(bool); required {
+		t.Fatalf("export --export-format required = true, want compatibility default")
+	}
+	if defaultValue := schemaContractString(exportFormat["default"]); defaultValue != "docx" {
+		t.Fatalf("export --export-format default = %q, want docx", defaultValue)
+	}
+
+	importLeaf := executeShortcutSchemaQuery(t, "--cli-path", "doc +import")
+	constraints, _ := importLeaf["constraints"].(map[string]any)
+	if requireOneOf, ok := constraints["require_one_of"]; ok && !schemaContractJSONEqual(requireOneOf, [][]string{}) {
+		t.Fatalf("import unexpectedly requires a target: %#v", constraints)
+	}
+}
+
 func executeShortcutSchemaQuery(t testing.TB, args ...string) map[string]any {
 	t.Helper()
 	root := NewRootCommand()

internal/cli/param_aliases_generated.go

@@ -568,9 +568,9 @@ var generatedParamAliases = []ParamAliasEntry{
 	{
 		CLIPath: "chat +flag-list",
 		Aliases: map[string]string{
-			"limit": "size",
+			"limit": "page-size",
 		},
-		Blocked: []string{"count", "max", "max-results", "max-size", "page", "page-size", "per-page"},
+		Blocked: []string{"count", "max", "max-results", "max-size", "page", "per-page"},
 	},
 	{
 		CLIPath: "chat +group-members",
@@ -2003,7 +2003,6 @@ var generatedParamAliases = []ParamAliasEntry{
 			"next-page-token": "cursor",
 			"next-token":      "cursor",
 			"node-id":         "node",
-			"page-size":       "limit",
 			"page-token":      "cursor",
 			"per-page":        "limit",
 			"size":            "limit",
@@ -2459,25 +2458,15 @@ var generatedParamAliases = []ParamAliasEntry{
 	{
 		CLIPath: "doc +version-list",
 		Aliases: map[string]string{
-			"doc":             "node",
-			"doc-id":          "node",
-			"document-id":     "node",
-			"file-id":         "node",
-			"max-result":      "limit",
-			"max-results":     "limit",
-			"next-cursor":     "cursor",
-			"next-page-token": "cursor",
-			"next-token":      "cursor",
-			"node-id":         "node",
-			"page-size":       "limit",
-			"page-token":      "cursor",
-			"per-page":        "limit",
-			"size":            "limit",
-			"take":            "limit",
-			"top":             "limit",
-			"url":             "node",
+			"doc":         "node",
+			"doc-id":      "node",
+			"document-id": "node",
+			"file-id":     "node",
+			"node-id":     "node",
+			"url":         "node",
 		},
-		Blocked: []string{"block-id", "comment-id", "comment-key", "count", "folder", "folder-id", "id", "job-id", "offset", "page", "parent-id", "revision", "task-id", "template-id", "version", "workspace", "workspace-id"},
+		Blocked:   []string{"block-id", "comment-id", "comment-key", "folder", "folder-id", "id", "job-id", "parent-id", "revision", "task-id", "template-id", "version", "workspace", "workspace-id"},
+		Ambiguous: []string{"max-result", "max-results", "next-cursor", "next-page-token", "next-token", "per-page", "size", "take", "top"},
 	},
 	{
 		CLIPath: "doc +version-revert",

internal/cli/param_concepts.json

@@ -44,6 +44,7 @@
   },
 
   "command_overrides": {
+	"doc +version-list": {"ambiguous": ["size", "max-results", "max-result", "take", "top", "per-page", "next-cursor", "next-token", "next-page-token"], "note": "--limit/--cursor and the shipped visible compatibility flags --page-size/--page-token remain native. Other pagination spellings cannot choose between two visible real flags and must stop before execution."},
     "chat group rename": {"bind": {"id": "open_conversation_id"}, "note": "This command's real --id carries one openConversationId; aliases reduce to --id without changing the value."},
     "chat group members": {"bind": {"id": "open_conversation_id"}},
     "chat group members add": {"bind": {"id": "open_conversation_id"}, "block": ["user-id", "open-dingtalk-id"], "note": "The real --users is a list and may contain mixed userId/openDingTalkId values; singular inputs are not promoted automatically."},
@@ -148,7 +149,7 @@
     "chat +chat-update": {"scoped_aliases": {"conversation-id": "group", "open-conversation-id": "group", "chat-id": "group", "title": "name", "new-title": "name"}, "block": ["id", "group-id", "group-ids", "conversation-ids", "open-conversation-ids"], "scope_strict": true, "note": "--group accepts a name or CID, so only explicit CID spellings are mapped; generic --id is blocked."},
     "chat +conversation-set-top": {"scoped_aliases": {"open-conversation-id": "conversation-id", "chat-id": "conversation-id", "open-conversation-ids": "conversation-ids", "chat-ids": "conversation-ids"}, "block": ["group", "groups", "group-id", "group-ids", "top", "set-top"], "scope_strict": true, "note": "Singular/list cardinality stays explicit; top/set-top cannot be rewritten to the inverse --off switch."},
     "chat +feed-group-query-item": {"scoped_aliases": {"open-conversation-ids": "conversation-ids", "chat-ids": "conversation-ids"}, "block": ["group", "groups", "group-id", "group-ids", "conversation-id", "open-conversation-id"], "scope_strict": true, "note": "The real field is an openConversationId list; group names and singular IDs are not converted."},
-    "chat +flag-list": {"scoped_aliases": {"limit": "size"}, "block": ["max", "max-results", "max-size", "count", "page", "page-size", "per-page"], "scope_strict": true, "note": "Only limit and size are reviewed as the same page bound; total-count and page-number spellings are not equivalent."},
+    "chat +flag-list": {"scoped_aliases": {"limit": "page-size"}, "block": ["max", "max-results", "max-size", "count", "page", "per-page"], "scope_strict": true, "note": "Native --page-size is the canonical page bound, native --size is its command-owned compatibility alias, and --limit is reviewed as value-preservingly equivalent to --page-size; total-count and page-number spellings are not equivalent."},
     "chat +messages-batch-recall-by-bot": {"block": ["msg-id", "message-id", "open-message-id", "msg-ids", "message-ids", "open-message-ids"], "note": "--keys carries processQueryKey values returned by bot sending; it is not an openMessageId field."},
     "chat +messages-combine-forward": {"scoped_aliases": {"src-open-cid": "src-conversation-id", "src-open-conversation-id": "src-conversation-id", "source-conversation-id": "src-conversation-id", "dest-open-cid": "dest-conversation-id", "dest-open-conversation-id": "dest-conversation-id", "target-conversation-id": "dest-conversation-id", "destination-conversation-id": "dest-conversation-id"}, "block": ["group-id", "group-ids"], "ambiguous": ["conversation-id", "open-conversation-id", "group", "chat", "chat-id", "id"], "scope_strict": true, "note": "Source and destination conversation roles remain explicit; role-free CID spellings cannot choose a side."},
     "chat +messages-forward": {"scoped_aliases": {"src-open-cid": "src-conversation-id", "src-open-conversation-id": "src-conversation-id", "source-conversation-id": "src-conversation-id", "dest-open-cid": "dest-conversation-id", "dest-open-conversation-id": "dest-conversation-id", "target-conversation-id": "dest-conversation-id", "destination-conversation-id": "dest-conversation-id", "src-open-message-id": "msg-id", "source-message-id": "msg-id"}, "block": ["group-id", "group-ids"], "ambiguous": ["conversation-id", "open-conversation-id", "group", "chat", "chat-id", "id"], "scope_strict": true, "note": "The message role is uniquely the source message, but source/destination conversation roles cannot be inferred from a generic CID."},
@@ -361,7 +362,7 @@
       {"command": "chat +chat-update", "emitted": "id", "expect": "did-you-mean:blocked", "via": "guard:generic-id-value-domain"},
       {"command": "chat +chat-update", "emitted": "title", "expect": "name", "via": "override:scoped-group-title"},
       {"command": "chat +chat-update", "emitted": "new-title", "expect": "name", "via": "override:scoped-group-title"},
-      {"command": "chat +flag-list", "emitted": "limit", "expect": "size", "via": "override:scoped-page-bound"},
+      {"command": "chat +flag-list", "emitted": "limit", "expect": "page-size", "via": "override:scoped-page-bound"},
       {"command": "chat +flag-list", "emitted": "max", "expect": "did-you-mean:blocked", "via": "guard:page-size-vs-total-count"},
       {"command": "chat +flag-list", "emitted": "max-results", "expect": "did-you-mean:blocked", "via": "guard:page-size-vs-total-count"},
       {"command": "chat +flag-list", "emitted": "max-size", "expect": "did-you-mean:blocked", "via": "guard:page-size-vs-total-count"},

internal/cli/schema_parameter_mapping_ledger.go

@@ -592,10 +592,26 @@ var reviewedSchemaParameterMappingExclusions = map[string]string{
 	"sheet.range_batch_clear --node":                                     "Reviewed unpinned adapter: sheet.range_batch_clear has no singular pinned interface_ref; --node is a CLI wrapper input and does not publish a direct interface property.",
 	"sheet.range_batch_clear --ranges":                                   "Reviewed unpinned adapter: sheet.range_batch_clear has no singular pinned interface_ref; --ranges is a CLI wrapper input and does not publish a direct interface property.",
 	"sheet.range_batch_clear --type":                                     "Reviewed unpinned adapter: sheet.range_batch_clear has no singular pinned interface_ref; --type is a CLI wrapper input and does not publish a direct interface property.",
-	"sheet.range_batch_set_style --batch":                                "Composite wrapper reads this local JSON file and performs zero or more update_range calls; the filesystem path is not a direct RPC property.",
-	"sheet.range_batch_set_style --continue-on-error":                    "Composite wrapper consumes this flag in its local multi-call error loop and never sends it to update_range.",
+	"sheet.range_batch_set_style --batch":                                "Composite wrapper reads this local JSON file and assembles it into a single sheet/batch_update operations array; the filesystem path is not a direct RPC property.",
+	"sheet.range_batch_set_style --continue-on-error":                    "Reviewed unpinned adapter: sheet.range_batch_set_style has no singular pinned interface_ref; --continue-on-error is forwarded to sheet/batch_update as continueOnError and does not publish a direct interface property.",
 	"sheet.range_read --range":                                           "Reviewed unpinned adapter: sheet.range_read has no singular pinned interface_ref; --range is a CLI wrapper input and does not publish a direct interface property.",
 	"sheet.range_read --sheet-id":                                        "Reviewed unpinned adapter: sheet.range_read has no singular pinned interface_ref; --sheet-id is a CLI wrapper input and does not publish a direct interface property.",
+	"sheet.range_set_style --bg-color":                                   "Composite stamp wrapper: set-style expands this flag into cells[][].cellStyles.backgroundColor across the set_cell_range matrix; it has no top-level RPC property.",
+	"sheet.range_set_style --bg-colors-json":                             "Composite stamp wrapper: set-style expands this per-cell matrix into cells[][].cellStyles.backgroundColor across the set_cell_range matrix; it has no top-level RPC property.",
+	"sheet.range_set_style --border-styles-json":                         "Composite stamp wrapper: set-style attaches this object as cells[][].borderStyles on every cell of the set_cell_range matrix; it has no top-level RPC property.",
+	"sheet.range_set_style --font-color":                                 "Composite stamp wrapper: set-style expands this flag into cells[][].cellStyles.fontColor across the set_cell_range matrix; it has no top-level RPC property.",
+	"sheet.range_set_style --font-colors-json":                           "Composite stamp wrapper: set-style expands this per-cell matrix into cells[][].cellStyles.fontColor across the set_cell_range matrix; it has no top-level RPC property.",
+	"sheet.range_set_style --font-family":                                "Composite stamp wrapper: set-style expands this flag into cells[][].cellStyles.fontFamily across the set_cell_range matrix; it has no top-level RPC property.",
+	"sheet.range_set_style --font-line":                                  "Composite stamp wrapper: set-style expands this single-choice flag into the cells[][].cellStyles.textUnderline and textLineThrough booleans of the set_cell_range matrix; it has no top-level RPC property.",
+	"sheet.range_set_style --font-size":                                  "Composite stamp wrapper: set-style expands this flag into cells[][].cellStyles.fontSize across the set_cell_range matrix; it has no top-level RPC property.",
+	"sheet.range_set_style --font-sizes-json":                            "Composite stamp wrapper: set-style expands this per-cell matrix into cells[][].cellStyles.fontSize across the set_cell_range matrix; it has no top-level RPC property.",
+	"sheet.range_set_style --font-style":                                 "Composite stamp wrapper: set-style expands this flag into cells[][].cellStyles.fontStyle across the set_cell_range matrix; it has no top-level RPC property.",
+	"sheet.range_set_style --font-weight":                                "Composite stamp wrapper: set-style expands this flag into cells[][].cellStyles.fontWeight across the set_cell_range matrix; it has no top-level RPC property.",
+	"sheet.range_set_style --font-weights-json":                          "Composite stamp wrapper: set-style expands this per-cell matrix into cells[][].cellStyles.fontWeight across the set_cell_range matrix; it has no top-level RPC property.",
+	"sheet.range_set_style --h-align":                                    "Composite stamp wrapper: set-style expands this flag into cells[][].cellStyles.horizontalAlignment across the set_cell_range matrix; it has no top-level RPC property.",
+	"sheet.range_set_style --h-aligns-json":                              "Composite stamp wrapper: set-style expands this per-cell matrix into cells[][].cellStyles.horizontalAlignment across the set_cell_range matrix; it has no top-level RPC property.",
+	"sheet.range_set_style --v-align":                                    "Composite stamp wrapper: set-style expands this flag into cells[][].cellStyles.verticalAlignment across the set_cell_range matrix; it has no top-level RPC property.",
+	"sheet.range_set_style --v-aligns-json":                              "Composite stamp wrapper: set-style expands this per-cell matrix into cells[][].cellStyles.verticalAlignment across the set_cell_range matrix; it has no top-level RPC property.",
 	"sheet.reply_sheet_comment --comment-key":                            "Reviewed unpinned adapter: sheet.reply_sheet_comment has no singular pinned interface_ref; --comment-key is a CLI wrapper input and does not publish a direct interface property.",
 	"sheet.reply_sheet_comment --content":                                "Reviewed unpinned adapter: sheet.reply_sheet_comment has no singular pinned interface_ref; --content is a CLI wrapper input and does not publish a direct interface property.",
 	"sheet.reply_sheet_comment --emoji":                                  "Reviewed unpinned adapter: sheet.reply_sheet_comment has no singular pinned interface_ref; --emoji is a CLI wrapper input and does not publish a direct interface property.",

internal/corecmd/corecmd.go

@@ -558,6 +558,18 @@ func RegisterFlags(cmd *cobra.Command, flags []FlagSpec) {
 		for _, alias := range flag.Aliases {
 			RegisterFlag(cmd, flag.Kind, alias, "", flag.Usage+" (alias)")
 			_ = cmd.Flags().MarkHidden(alias)
+			if registered := cmd.Flags().Lookup(alias); registered != nil {
+				runtimeannotate.SetFlagAnnotation(
+					registered,
+					runtimeannotate.AnnotationFlagAliasOf,
+					flag.Name,
+				)
+				runtimeannotate.SetFlagAnnotation(
+					registered,
+					runtimeannotate.AnnotationFlagAliasOrigin,
+					runtimeannotate.FlagAliasOriginCorecmdV1,
+				)
+			}
 		}
 		if flag.MarkRequired {
 			_ = cmd.MarkFlagRequired(flag.Name)

internal/corecmd/corecmd_test.go

@@ -90,11 +90,17 @@ func TestCrossPlatformCoverageRegisterFlagsAllKinds(t *testing.T) {
 		}
 	}
 	// Aliases are registered with the main kind and hidden.
-	for _, alias := range []string{"i-alias", "sl-alias"} {
+	for alias, canonical := range map[string]string{"i-alias": "i", "sl-alias": "sl"} {
 		f := cmd.Flags().Lookup(alias)
 		if f == nil || !f.Hidden {
 			t.Fatalf("alias %q = %#v, want registered+hidden", alias, f)
 		}
+		if got := f.Annotations[runtimeannotate.AnnotationFlagAliasOf]; len(got) != 1 || got[0] != canonical {
+			t.Fatalf("alias %q annotation = %#v, want alias_of %q", alias, got, canonical)
+		}
+		if got := f.Annotations[runtimeannotate.AnnotationFlagAliasOrigin]; len(got) != 1 || got[0] != runtimeannotate.FlagAliasOriginCorecmdV1 {
+			t.Fatalf("alias %q origin = %#v, want corecmd FlagSpec marker", alias, got)
+		}
 	}
 	if cmd.Flags().Lookup("i-alias").Value.Type() != "int" {
 		t.Fatal("int alias must be registered as int")

internal/corecmd/runtimeannotate/interface_alias.go

@@ -0,0 +1,25 @@
+// 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 runtimeannotate
+
+// CLI flag alias evidence is kept in this narrow file so the base-owned
+// Interface Snapshot helper can add the protocol constants to an older stable
+// worktree without replacing that revision's complete runtimeannotate package.
+// Only corecmd.FlagSpec.Aliases writes the exact origin; neither field is a
+// Schema synonym or final payload-equivalence proof.
+const (
+	AnnotationFlagAliasOf     = "dws.compat.alias_of"
+	AnnotationFlagAliasOrigin = "dws.compat.alias_origin"
+	FlagAliasOriginCorecmdV1  = "corecmd.flag_spec_aliases.v1"
+)

internal/helpers/aitable.go

@@ -5,12 +5,15 @@ import (
 	"encoding/json"
 	"fmt"
 	"io"
+	"math"
 	"os"
 	"strconv"
 	"strings"
 	"time"
 
 	"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/pkg/paging"
 	"github.com/spf13/cobra"
 
 	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/corecmd/contract"
@@ -123,113 +126,195 @@ func resolveWorkflowDSL(cmd *cobra.Command) (map[string]any, error) {
 }
 
 // recordQueryFetchAll implements --all auto-pagination for record query.
-//   - pageLimit controls max pages (default 50, 0 = unlimited)
-//   - Mid-loop errors break gracefully, outputting already-fetched data
-//   - Merged result preserves last page's cursor so caller can resume
+// It prints only a complete result. A page limit, empty/invalid response,
+// transport failure, or cursor cycle returns a non-zero structured error whose
+// details retain the incomplete records and retry cursor.
 func recordQueryFetchAll(toolArgs map[string]any, pageLimit int) error {
-	const pageDelayMs = 200
 	const serverID = "aitable"
 
 	ctx := context.Background()
-	var allRecords []any
-	page := 0
-	lastCursor := ""
-
-	for {
-		page++
-
-		if page > 1 {
-			helperSleep(time.Duration(pageDelayMs) * time.Millisecond)
-		}
-
-		fmt.Fprintf(os.Stderr, "[page %d] fetching...\n", page)
-
-		result, err := deps.Caller.CallTool(ctx, serverID, "query_records", toolArgs)
-		if err != nil {
-			if page == 1 {
-				return WrapError(err)
-			}
-			fmt.Fprintf(os.Stderr, "[page %d] error, stopping pagination: %v\n", page, err)
-			break
-		}
-
-		// Extract text content from tool result
-		var text string
-		for _, c := range result.Content {
-			if c.Type == "text" && c.Text != "" {
-				text = c.Text
-				break
-			}
-		}
-		if text == "" {
-			break
-		}
-
-		var resp map[string]any
-		if err := json.Unmarshal([]byte(text), &resp); err != nil {
-			if page == 1 {
-				deps.Out.PrintRaw(text)
-				return nil
-			}
-			break
-		}
-
-		// Extract records from response
-		var pageRecords []any
-		if data, ok := resp["data"].(map[string]any); ok {
-			if recs, ok := data["records"].([]any); ok {
-				pageRecords = recs
-			}
-		} else if recs, ok := resp["records"].([]any); ok {
-			pageRecords = recs
-		}
-		allRecords = append(allRecords, pageRecords...)
-
-		// Extract cursor for next page (MCP returns "nextCursor" or "cursor")
-		cursor := ""
-		if data, ok := resp["data"].(map[string]any); ok {
-			if c, ok := data["nextCursor"].(string); ok && c != "" {
-				cursor = c
-			} else if c, ok := data["cursor"].(string); ok && c != "" {
-				cursor = c
-			}
-		} else if c, ok := resp["nextCursor"].(string); ok && c != "" {
-			cursor = c
-		} else if c, ok := resp["cursor"].(string); ok && c != "" {
-			cursor = c
-		}
+	requestArgs := make(map[string]any, len(toolArgs))
+	for key, value := range toolArgs {
+		requestArgs[key] = value
+	}
+	initialCursor, _ := requestArgs["cursor"].(string)
 
+	effectivePageLimit := pageLimit
+	if pageLimit == 0 {
+		effectivePageLimit = paging.UnlimitedPageLimit
+	}
+	result := paging.FetchAll(ctx, func(ctx context.Context, cursor string) (paging.Page, error) {
 		if cursor == "" {
-			lastCursor = ""
-			break
+			delete(requestArgs, "cursor")
+		} else {
+			requestArgs["cursor"] = cursor
 		}
-
-		// Check page limit (0 = unlimited)
-		if pageLimit > 0 && page >= pageLimit {
-			lastCursor = cursor
-			fmt.Fprintf(os.Stderr, "[pagination] reached page limit (%d), stopping. Use --page-limit 0 to fetch all.\n", pageLimit)
-			break
+		toolResult, err := deps.Caller.CallTool(ctx, serverID, "query_records", requestArgs)
+		if err != nil {
+			return paging.Page{}, err
 		}
+		if toolResult == nil {
+			return paging.Page{}, fmt.Errorf("query_records returned a nil result")
+		}
+		for _, content := range toolResult.Content {
+			if content.Type == "text" && strings.TrimSpace(content.Text) != "" {
+				return parseRecordQueryPage(content.Text)
+			}
+		}
+		return paging.Page{}, fmt.Errorf("query_records returned no non-empty text content")
+	}, paging.Options{
+		PageLimit:      effectivePageLimit,
+		InterPageDelay: paging.DefaultInterPageDelay,
+		InitialCursor:  initialCursor,
+	})
 
-		lastCursor = ""
-		toolArgs["cursor"] = cursor
+	if !result.Complete {
+		return recordQueryIncompleteError(result, pageLimit)
 	}
 
-	fmt.Fprintf(os.Stderr, "[pagination] done: %d pages, %d total records\n", page, len(allRecords))
-
-	// Build merged output — preserve cursor & has_more for resume capability
+	fmt.Fprintf(os.Stderr, "[pagination] complete: %d pages, %d fetched records\n", result.Pages, len(result.Records))
 	mergedData := map[string]any{
-		"records":    allRecords,
-		"totalCount": len(allRecords),
+		"records":      result.Records,
+		"fetchedCount": len(result.Records),
+		"hasMore":      false,
+		"complete":     true,
+		"pages":        result.Pages,
 	}
-	if lastCursor != "" {
-		mergedData["cursor"] = lastCursor
-		mergedData["hasMore"] = true
-	} else {
-		mergedData["hasMore"] = false
+	if result.TotalCount != nil {
+		mergedData["totalCount"] = *result.TotalCount
+	}
+	return deps.Out.PrintJSON(map[string]any{"data": mergedData})
+}
+
+func parseRecordQueryPage(text string) (paging.Page, error) {
+	decoder := json.NewDecoder(strings.NewReader(text))
+	decoder.UseNumber()
+	var response map[string]any
+	if err := decoder.Decode(&response); err != nil {
+		return paging.Page{}, fmt.Errorf("query_records returned invalid JSON: %w", err)
+	}
+	var trailing any
+	if err := decoder.Decode(&trailing); err != io.EOF {
+		if err == nil {
+			return paging.Page{}, fmt.Errorf("query_records returned more than one JSON value")
+		}
+		return paging.Page{}, fmt.Errorf("query_records returned trailing invalid JSON: %w", err)
+	}
+	if response == nil {
+		return paging.Page{}, fmt.Errorf("query_records returned null instead of an object")
 	}
 
-	return deps.Out.PrintJSON(map[string]any{"data": mergedData})
+	payload := response
+	if rawData, exists := response["data"]; exists {
+		data, ok := rawData.(map[string]any)
+		if !ok {
+			return paging.Page{}, fmt.Errorf("query_records data must be an object, got %T", rawData)
+		}
+		payload = data
+	}
+
+	rawRecords, exists := payload["records"]
+	if !exists {
+		return paging.Page{}, fmt.Errorf("query_records response is missing records")
+	}
+	records, ok := rawRecords.([]any)
+	if !ok {
+		return paging.Page{}, fmt.Errorf("query_records records must be an array, got %T", rawRecords)
+	}
+	for index, record := range records {
+		if _, ok := record.(map[string]any); !ok {
+			return paging.Page{}, fmt.Errorf("query_records records[%d] must be an object, got %T", index, record)
+		}
+	}
+
+	nextCursor := firstNonEmptyString(payload, "nextCursor", "cursor")
+	if hasMore, ok := payload["hasMore"].(bool); ok && hasMore && nextCursor == "" {
+		return paging.Page{}, fmt.Errorf("query_records reported hasMore=true without a next cursor")
+	}
+	totalCount, err := parseOptionalNonNegativeInt(payload["totalCount"])
+	if err != nil {
+		return paging.Page{}, fmt.Errorf("query_records totalCount: %w", err)
+	}
+	return paging.Page{Records: records, NextCursor: nextCursor, TotalCount: totalCount}, nil
+}
+
+func firstNonEmptyString(values map[string]any, keys ...string) string {
+	for _, key := range keys {
+		if value, ok := values[key].(string); ok && strings.TrimSpace(value) != "" {
+			return value
+		}
+	}
+	return ""
+}
+
+func parseOptionalNonNegativeInt(value any) (*int, error) {
+	if value == nil {
+		return nil, nil
+	}
+	var parsed int64
+	switch typed := value.(type) {
+	case json.Number:
+		integer, err := typed.Int64()
+		if err != nil {
+			return nil, fmt.Errorf("must be an integer, got %q", typed)
+		}
+		parsed = integer
+	case float64:
+		if math.Trunc(typed) != typed {
+			return nil, fmt.Errorf("must be an integer, got %v", typed)
+		}
+		parsed = int64(typed)
+	case int:
+		parsed = int64(typed)
+	case int64:
+		parsed = typed
+	default:
+		return nil, fmt.Errorf("must be a non-negative integer, got %T", value)
+	}
+	if parsed < 0 || int64(int(parsed)) != parsed {
+		return nil, fmt.Errorf("must fit a non-negative int, got %d", parsed)
+	}
+	result := int(parsed)
+	return &result, nil
+}
+
+func recordQueryIncompleteError(result paging.Result, pageLimit int) error {
+	incomplete := map[string]any{
+		"records":      result.Records,
+		"fetchedCount": len(result.Records),
+		"pages":        result.Pages,
+		"attempts":     result.Attempts,
+		"complete":     false,
+		"hasMore":      result.HasMore,
+		"stopReason":   result.StopReason,
+	}
+	if result.LastCursor != "" {
+		incomplete["cursor"] = result.LastCursor
+	}
+	if result.TotalCount != nil {
+		incomplete["totalCount"] = *result.TotalCount
+	}
+
+	hint := "retry the command; the structured error details preserve fetched records and the retry cursor"
+	if result.StopReason == paging.StopPageLimit {
+		hint = "rerun with --page-limit 0 to require a complete result, or resume from details.cursor"
+	}
+	message := fmt.Sprintf("record pagination incomplete after %d successful page(s) and %d fetched record(s)", result.Pages, len(result.Records))
+	return apperrors.NewAPI(message,
+		apperrors.WithOperation("aitable.query_records.all"),
+		apperrors.WithServerKey("aitable"),
+		apperrors.WithOrigin("mcp"),
+		apperrors.WithFailureStage("pagination"),
+		apperrors.WithExecutionStarted(true),
+		apperrors.WithRetryable(true),
+		apperrors.WithReason("pagination_"+string(result.StopReason)),
+		apperrors.WithHint(hint),
+		apperrors.WithDetails(map[string]any{
+			"page_limit":        pageLimit,
+			"incomplete_result": incomplete,
+		}),
+		apperrors.WithCause(result.Err),
+	)
 }
 
 // ─── filters 格式校验 ──────────────────────────────────────────────────────
@@ -6966,8 +7051,8 @@ parentSectionId 为空串表示该节点在 Base 根目录下。
 	recordQueryCmd.Flags().Int("page-size", 0, "--limit 的别名（兼容 LLM 常见误用）")
 	_ = recordQueryCmd.Flags().MarkHidden("page-size")
 	recordQueryCmd.Flags().String("cursor", "", "分页游标，首次查询不传；cursor 为空表示已取完全部记录")
-	recordQueryCmd.Flags().Bool("all", false, "自动翻页获取全部记录。传入时自动循环直到无更多数据或达到 --page-limit 上限")
-	recordQueryCmd.Flags().Int("page-limit", 50, "自动翻页最大页数（仅 --all 时生效）。默认 50 页（5000 条），设为 0 表示无限制")
+	recordQueryCmd.Flags().Bool("all", false, "自动翻页获取完整记录集；达到 --page-limit 且仍有更多页时返回非零结构化错误，不把不完整结果作为成功输出")
+	recordQueryCmd.Flags().Int("page-limit", 50, "自动翻页最大页数（仅 --all 时生效）。默认 50 页（约 5000 条）；设为 0 表示显式不限页数；超限时错误详情保留已取记录和续传 cursor")
 	recordQueryCmd.Flags().String("view-id", "", "视图 ID（record query 不支持按视图过滤，此参数会被忽略并给出提示）")
 	_ = recordQueryCmd.Flags().MarkHidden("view-id")
 	recordCreateCmd.Flags().String("base-id", "", "Base ID，可通过 base list 或 base search 获取 (必填)")

internal/helpers/aitable_core_test.go

@@ -292,15 +292,15 @@ func TestCrossPlatformCoverageAitableToolResponseAndPaginationHelpers(t *testing
 		t.Fatalf("update view top-level: %v", err)
 	}
 
-	caller = &aitableTestCaller{responses: []string{`{"data":{"records":[{"id":1}],"nextCursor":"next"}}`}}
+	caller = &aitableTestCaller{responses: []string{`{"data":{"records":[{"id":1}],"totalCount":17}}`}}
 	out = installAitableDeps(t, caller)
-	if err := recordQueryFetchAll(map[string]any{}, 1); err != nil || !strings.Contains(out.String(), "totalCount") {
+	if err := recordQueryFetchAll(map[string]any{}, 1); err != nil || !strings.Contains(out.String(), `"totalCount": 17`) || !strings.Contains(out.String(), `"fetchedCount": 1`) {
 		t.Fatalf("paginated records = %q, %v", out.String(), err)
 	}
 	caller = &aitableTestCaller{responses: []string{"not-json"}}
 	out = installAitableDeps(t, caller)
-	if err := recordQueryFetchAll(map[string]any{}, 1); err != nil || !strings.Contains(out.String(), "not-json") {
-		t.Fatalf("raw first page = %q, %v", out.String(), err)
+	if err := recordQueryFetchAll(map[string]any{}, 1); err == nil || out.Len() != 0 {
+		t.Fatalf("invalid first page must fail without success output = %q, %v", out.String(), err)
 	}
 	caller = &aitableTestCaller{responses: []string{`{"records":[{"id":1}]}`}}
 	installAitableDeps(t, caller)

internal/helpers/aitable_coverage_gap_test.go

@@ -0,0 +1,93 @@
+// Copyright 2026 Alibaba Group
+// SPDX-License-Identifier: Apache-2.0
+
+package helpers
+
+import (
+	"encoding/json"
+	"fmt"
+	"strings"
+	"testing"
+
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/pkg/paging"
+)
+
+func TestCrossPlatformCoverageAITablePageParsingRemainingEdges(t *testing.T) {
+	for name, raw := range map[string]string{
+		"trailing invalid JSON": `{"records":[]} {`,
+		"data not object":       `{"data":[]}`,
+	} {
+		t.Run(name, func(t *testing.T) {
+			if _, err := parseRecordQueryPage(raw); err == nil {
+				t.Fatalf("parseRecordQueryPage(%q) succeeded", raw)
+			}
+		})
+	}
+}
+
+func TestCrossPlatformCoverageAITableOptionalCountNumericTypes(t *testing.T) {
+	cases := []struct {
+		name    string
+		value   any
+		want    int
+		wantErr bool
+	}{
+		{name: "bad number", value: json.Number("1.5"), wantErr: true},
+		{name: "fractional float", value: float64(1.5), wantErr: true},
+		{name: "whole float", value: float64(2), want: 2},
+		{name: "int", value: int(3), want: 3},
+		{name: "int64", value: int64(4), want: 4},
+		{name: "wrong type", value: "5", wantErr: true},
+		{name: "negative", value: int64(-1), wantErr: true},
+	}
+	for _, tc := range cases {
+		t.Run(tc.name, func(t *testing.T) {
+			got, err := parseOptionalNonNegativeInt(tc.value)
+			if tc.wantErr {
+				if err == nil {
+					t.Fatalf("value %#v succeeded: %#v", tc.value, got)
+				}
+				return
+			}
+			if err != nil || got == nil || *got != tc.want {
+				t.Fatalf("value %#v = %#v, %v", tc.value, got, err)
+			}
+		})
+	}
+}
+
+func TestCrossPlatformCoverageAITableIncompleteResultMetadata(t *testing.T) {
+	total := 9
+	err := recordQueryIncompleteError(paging.Result{
+		Records: []any{map[string]any{"id": "r"}}, Pages: 1, Attempts: 1,
+		HasMore: true, LastCursor: "next", StopReason: paging.StopPageLimit,
+		TotalCount: &total,
+	}, 1)
+	if err == nil || !strings.Contains(err.Error(), "pagination incomplete") {
+		t.Fatalf("incomplete result error = %v", err)
+	}
+}
+
+func TestCrossPlatformCoverageAITableExplicitUnlimitedPaginationE2E(t *testing.T) {
+	responses := make([]string, paging.DefaultPageLimit+1)
+	for index := range responses {
+		next := ""
+		if index < len(responses)-1 {
+			next = fmt.Sprintf(`,"nextCursor":"cursor-%d"`, index+1)
+		}
+		responses[index] = fmt.Sprintf(`{"records":[{"recordId":"record-%d"}]%s}`, index, next)
+	}
+	caller := &aitableTestCaller{responses: responses}
+	out := installAitableDeps(t, caller)
+	if err := recordQueryFetchAll(map[string]any{}, 0); err != nil {
+		t.Fatalf("explicit unlimited pagination failed: %v", err)
+	}
+	if len(caller.calls) != len(responses) {
+		t.Fatalf("explicit unlimited pagination calls = %d, want %d", len(caller.calls), len(responses))
+	}
+	for _, want := range []string{`"pages": 51`, `"fetchedCount": 51`, `"complete": true`} {
+		if !strings.Contains(out.String(), want) {
+			t.Fatalf("explicit unlimited output missing %s: %s", want, out.String())
+		}
+	}
+}

internal/helpers/aitable_record_query_e2e_test.go

@@ -0,0 +1,192 @@
+// Copyright 2026 Alibaba Group
+// SPDX-License-Identifier: Apache-2.0
+
+package helpers
+
+import (
+	"bytes"
+	"context"
+	"errors"
+	"fmt"
+	"os"
+	"reflect"
+	"strings"
+	"testing"
+
+	apperrors "github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/errors"
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/testseam"
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/pkg/edition"
+)
+
+type recordQueryE2EStep struct {
+	result *edition.ToolResult
+	err    error
+}
+
+type recordQueryE2ECaller struct {
+	steps []recordQueryE2EStep
+	calls []aitableTestCall
+}
+
+func (c *recordQueryE2ECaller) CallTool(_ context.Context, server, tool string, args map[string]any) (*edition.ToolResult, error) {
+	cloned := make(map[string]any, len(args))
+	for key, value := range args {
+		cloned[key] = value
+	}
+	c.calls = append(c.calls, aitableTestCall{server: server, tool: tool, args: cloned})
+	index := len(c.calls) - 1
+	if index >= len(c.steps) {
+		return nil, fmt.Errorf("unexpected tool call %d", index+1)
+	}
+	return c.steps[index].result, c.steps[index].err
+}
+
+func (*recordQueryE2ECaller) Format() string { return "json" }
+func (*recordQueryE2ECaller) DryRun() bool   { return false }
+func (*recordQueryE2ECaller) Fields() string { return "" }
+func (*recordQueryE2ECaller) JQ() string     { return "" }
+
+func runRecordQueryCLI(t *testing.T, caller *recordQueryE2ECaller, extraArgs ...string) (string, error) {
+	t.Helper()
+	testseam.Protect(t, &deps)
+	oldArgs := os.Args
+	t.Cleanup(func() { os.Args = oldArgs })
+
+	InitDeps(caller)
+	out := &bytes.Buffer{}
+	deps.Out.w = out
+	deps.Out.errW = out
+	args := []string{"record", "query", "--base-id", "base-e2e", "--table-id", "table-e2e", "--all"}
+	args = append(args, extraArgs...)
+	os.Args = append([]string{"dws", "aitable"}, args...)
+
+	command := newAitableCommand()
+	command.SilenceErrors = true
+	command.SilenceUsage = true
+	command.SetArgs(args)
+	err := command.Execute()
+	return out.String(), err
+}
+
+func recordQueryTextStep(text string) recordQueryE2EStep {
+	return recordQueryE2EStep{result: textToolResult(text)}
+}
+
+func TestCrossPlatformCoverageRecordQueryCLICompleteE2E(t *testing.T) {
+	caller := &recordQueryE2ECaller{steps: []recordQueryE2EStep{
+		recordQueryTextStep(`{"data":{"records":[{"id":"r1"},{"id":"r2"}],"nextCursor":"cursor-2","totalCount":3}}`),
+		recordQueryTextStep(`{"data":{"records":[{"id":"r3"}],"nextCursor":"","totalCount":3}}`),
+	}}
+	out, err := runRecordQueryCLI(t, caller, "--page-limit", "0")
+	if err != nil {
+		t.Fatalf("record query CLI returned error: %v", err)
+	}
+	for _, want := range []string{`"complete": true`, `"fetchedCount": 3`, `"totalCount": 3`, `"id": "r3"`} {
+		if !bytes.Contains([]byte(out), []byte(want)) {
+			t.Fatalf("record query CLI output missing %s:\n%s", want, out)
+		}
+	}
+	if len(caller.calls) != 2 {
+		t.Fatalf("tool calls = %d, want 2", len(caller.calls))
+	}
+	if got := caller.calls[0].args["cursor"]; got != nil {
+		t.Fatalf("first-page cursor = %#v, want absent", got)
+	}
+	if got := caller.calls[1].args["cursor"]; got != "cursor-2" {
+		t.Fatalf("second-page cursor = %#v, want cursor-2", got)
+	}
+}
+
+func TestCrossPlatformCoverageRecordQueryCLIFailsClosedE2E(t *testing.T) {
+	tests := []struct {
+		name  string
+		steps []recordQueryE2EStep
+		args  []string
+	}{
+		{name: "nil result", steps: []recordQueryE2EStep{{result: nil}}},
+		{name: "empty content", steps: []recordQueryE2EStep{{result: &edition.ToolResult{}}}},
+		{name: "empty text", steps: []recordQueryE2EStep{recordQueryTextStep("")}},
+		{name: "invalid json", steps: []recordQueryE2EStep{recordQueryTextStep("{")}},
+		{name: "null payload", steps: []recordQueryE2EStep{recordQueryTextStep("null")}},
+		{name: "missing records", steps: []recordQueryE2EStep{recordQueryTextStep(`{"data":{"nextCursor":"c"}}`)}},
+		{name: "records wrong type", steps: []recordQueryE2EStep{recordQueryTextStep(`{"records":{}}`)}},
+		{name: "records null", steps: []recordQueryE2EStep{recordQueryTextStep(`{"records":null}`)}},
+		{name: "record item wrong type", steps: []recordQueryE2EStep{recordQueryTextStep(`{"records":["bad"]}`)}},
+		{name: "trailing json", steps: []recordQueryE2EStep{recordQueryTextStep(`{"records":[]} {}`)}},
+		{name: "has more without cursor", steps: []recordQueryE2EStep{recordQueryTextStep(`{"records":[],"hasMore":true}`)}},
+		{name: "invalid total", steps: []recordQueryE2EStep{recordQueryTextStep(`{"records":[],"totalCount":"many"}`)}},
+		{name: "first transport error", steps: []recordQueryE2EStep{{err: errors.New("offline")}}},
+		{
+			name: "second page error",
+			steps: []recordQueryE2EStep{
+				recordQueryTextStep(`{"records":[{"id":"kept"}],"nextCursor":"retry-me"}`),
+				{err: errors.New("upstream reset")},
+			},
+			args: []string{"--page-limit", "0"},
+		},
+		{
+			name:  "page limit",
+			steps: []recordQueryE2EStep{recordQueryTextStep(`{"records":[{"id":"kept"}],"nextCursor":"resume-me"}`)},
+			args:  []string{"--page-limit", "1"},
+		},
+		{
+			name: "cursor cycle",
+			steps: []recordQueryE2EStep{
+				recordQueryTextStep(`{"records":[{"id":"r1"}],"nextCursor":"same"}`),
+				recordQueryTextStep(`{"records":[{"id":"r2"}],"nextCursor":"same"}`),
+			},
+			args: []string{"--page-limit", "0"},
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			caller := &recordQueryE2ECaller{steps: test.steps}
+			out, err := runRecordQueryCLI(t, caller, test.args...)
+			if err == nil {
+				t.Fatalf("CLI treated %s as success; output=%s", test.name, out)
+			}
+			if out != "" {
+				t.Fatalf("CLI emitted success output for %s: %q", test.name, out)
+			}
+			var typed *apperrors.Error
+			if !errors.As(err, &typed) {
+				t.Fatalf("error type = %T, want *errors.Error", err)
+			}
+			if typed.Reason == "" || typed.Details["incomplete_result"] == nil {
+				t.Fatalf("error lacks recovery metadata: %#v", typed)
+			}
+		})
+	}
+}
+
+func TestCrossPlatformCoverageRecordQueryCLIInitialCursorE2E(t *testing.T) {
+	caller := &recordQueryE2ECaller{steps: []recordQueryE2EStep{
+		recordQueryTextStep(`{"records":[]}`),
+	}}
+	if out, err := runRecordQueryCLI(t, caller, "--cursor", "resume-from-here", "--page-limit", "0"); err != nil || out == "" {
+		t.Fatalf("resume query = %q, %v", out, err)
+	}
+	want := map[string]any{"baseId": "base-e2e", "tableId": "table-e2e", "cursor": "resume-from-here"}
+	if !reflect.DeepEqual(caller.calls[0].args, want) {
+		t.Fatalf("resume args = %#v, want %#v", caller.calls[0].args, want)
+	}
+}
+
+func TestCrossPlatformCoverageRecordQueryCLIPageLimitHelpE2E(t *testing.T) {
+	command := newAitableCommand()
+	command.SilenceErrors = true
+	command.SilenceUsage = true
+	out := &bytes.Buffer{}
+	command.SetOut(out)
+	command.SetErr(out)
+	command.SetArgs([]string{"record", "query", "--help"})
+	if err := command.Execute(); err != nil {
+		t.Fatalf("record query help failed: %v", err)
+	}
+	for _, want := range []string{"返回非零结构化错误", "不完整结果", "错误详情保留已取记录和续传 cursor"} {
+		if !strings.Contains(out.String(), want) {
+			t.Fatalf("record query help missing %q:\n%s", want, out.String())
+		}
+	}
+}

internal/helpers/chat/deps.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 chat
+
+import (
+	"fmt"
+	"strconv"
+	"strings"
+
+	"github.com/spf13/cobra"
+
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/corecmd"
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/corecmd/contract"
+	apperrors "github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/errors"
+)
+
+type LeafContract = corecmd.ContractDecl
+
+type LeafSpec struct {
+	Safety   contract.SafetySpec
+	Contract LeafContract
+}
+
+type Deps struct {
+	GroupRunE            func(*cobra.Command, []string) error
+	CallMCPToolOnServer  func(string, string, map[string]any) error
+	DeclareLeafMetadata  func(*cobra.Command, LeafSpec) *cobra.Command
+	ValidateRequiredFlag func(*cobra.Command, ...string) error
+}
+
+var deps = Deps{
+	GroupRunE: func(cmd *cobra.Command, args []string) error {
+		return cmd.Help()
+	},
+	CallMCPToolOnServer: func(serverID, toolName string, args map[string]any) error {
+		return fmt.Errorf("im helpers dependency CallMCPToolOnServer is not initialized")
+	},
+	DeclareLeafMetadata: func(cmd *cobra.Command, spec LeafSpec) *cobra.Command {
+		return cmd
+	},
+	ValidateRequiredFlag: validateRequiredFlagsLocal,
+}
+
+func SetDeps(next Deps) {
+	if next.GroupRunE != nil {
+		deps.GroupRunE = next.GroupRunE
+	}
+	if next.CallMCPToolOnServer != nil {
+		deps.CallMCPToolOnServer = next.CallMCPToolOnServer
+	}
+	if next.DeclareLeafMetadata != nil {
+		deps.DeclareLeafMetadata = next.DeclareLeafMetadata
+	}
+	if next.ValidateRequiredFlag != nil {
+		deps.ValidateRequiredFlag = next.ValidateRequiredFlag
+	}
+}
+
+func NewChatToolbarCommand() *cobra.Command {
+	return newChatToolbarCommand()
+}
+
+func groupRunE(cmd *cobra.Command, args []string) error {
+	return deps.GroupRunE(cmd, args)
+}
+
+func callMCPToolOnServer(serverID, toolName string, args map[string]any) error {
+	return deps.CallMCPToolOnServer(serverID, toolName, args)
+}
+
+func DeclareLeafMetadata(cmd *cobra.Command, spec LeafSpec) *cobra.Command {
+	return deps.DeclareLeafMetadata(cmd, spec)
+}
+
+func validateRequiredFlags(cmd *cobra.Command, names ...string) error {
+	return deps.ValidateRequiredFlag(cmd, names...)
+}
+
+func mustGetFlag(cmd *cobra.Command, name string) string {
+	val, _ := cmd.Flags().GetString(name)
+	if val == "" {
+		val, _ = cmd.InheritedFlags().GetString(name)
+	}
+	return strings.TrimSpace(val)
+}
+
+func parseCSVInt64(raw string) ([]int64, error) {
+	parts := strings.Split(raw, ",")
+	result := make([]int64, 0, len(parts))
+	for _, p := range parts {
+		v := strings.TrimSpace(p)
+		if v == "" {
+			continue
+		}
+		n, err := strconv.ParseInt(v, 10, 64)
+		if err != nil {
+			return nil, fmt.Errorf("invalid integer %q", v)
+		}
+		result = append(result, n)
+	}
+	if len(result) == 0 {
+		return nil, fmt.Errorf("at least one ID is required")
+	}
+	return result, nil
+}
+
+func validateRequiredFlagsLocal(cmd *cobra.Command, names ...string) error {
+	for _, name := range names {
+		if strings.TrimSpace(mustGetFlag(cmd, name)) == "" {
+			return apperrors.NewValidation(
+				fmt.Sprintf("flag --%s is required", name),
+				apperrors.WithReason("missing_required_flag"),
+				apperrors.WithHint(fmt.Sprintf("使用 --%s 指定必填参数", name)),
+			)
+		}
+	}
+	return nil
+}
+
+func commandBoolFlag(cmd *cobra.Command, name string) bool {
+	if cmd == nil {
+		return false
+	}
+	value, _ := cmd.Flags().GetBool(name)
+	if !value {
+		value, _ = cmd.InheritedFlags().GetBool(name)
+	}
+	return value
+}
+
+func boolPtr(v bool) *bool {
+	return &v
+}

internal/helpers/chat/toolbar.go

@@ -0,0 +1,38 @@
+// 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 chat
+
+import "github.com/spf13/cobra"
+
+func newChatToolbarCommand() *cobra.Command {
+	toolbarCmd := &cobra.Command{
+		Use:   "toolbar",
+		Short: "快捷栏管理",
+		Long:  "管理会话快捷栏入口：查询、添加、隐藏、排序及自定义入口 CRUD。",
+		RunE:  groupRunE,
+	}
+	toolbarCmd.DisableAutoGenTag = true
+
+	toolbarCmd.AddCommand(
+		newToolbarListCommand(),
+		newToolbarAddCommand(),
+		newToolbarHideCommand(),
+		newToolbarSortCommand(),
+		newToolbarCreateCustomCommand(),
+		newToolbarRemoveCustomCommand(),
+		newToolbarUpdateCustomCommand(),
+	)
+
+	return toolbarCmd
+}

internal/helpers/chat/toolbar_add.go

@@ -0,0 +1,92 @@
+// 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 chat
+
+import (
+	"fmt"
+
+	"github.com/spf13/cobra"
+
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/corecmd/contract"
+)
+
+func newToolbarAddCommand() *cobra.Command {
+	cmd := &cobra.Command{
+		Use:   "add",
+		Short: "将入口添加到快捷栏可见区",
+		Long:  "将一个或多个快捷入口从隐藏区移至可见区。入口 ID 使用逗号分隔。",
+		Example: `  dws chat toolbar add --conversation-id <cid> --shortcut-ids 101,102
+  # 查询入口 ID: dws chat toolbar list --conversation-id <cid>`,
+		RunE: func(cmd *cobra.Command, args []string) error {
+			cid, err := toolbarConversationID(cmd)
+			if err != nil {
+				return err
+			}
+			if err := validateRequiredFlags(cmd, "shortcut-ids"); err != nil {
+				return err
+			}
+			ids, err := parseCSVInt64(mustGetFlag(cmd, "shortcut-ids"))
+			if err != nil {
+				return fmt.Errorf("--shortcut-ids: %w", err)
+			}
+			err = callMCPToolOnServer("im", "add_shortcut_to_bar", map[string]any{
+				"openConversationId": cid,
+				"shortcutIds":        ids,
+			})
+			if isSystemBusy(err) {
+				return toolbarNewSystemBusyError()
+			}
+			return err
+		},
+	}
+	cmd.Flags().String("conversation-id", "", "会话 openConversationId")
+	cmd.Flags().String("shortcut-ids", "", "入口 ID 列表（逗号分隔）")
+	_ = cmd.MarkFlagRequired("conversation-id")
+	_ = cmd.MarkFlagRequired("shortcut-ids")
+	cmd.DisableAutoGenTag = true
+
+	DeclareLeafMetadata(cmd, LeafSpec{
+		Safety: contract.SafetySpec{
+			Effect: "write", Risk: "low",
+			Confirmation: "not_required", Idempotency: "idempotent",
+		},
+		Contract: LeafContract{
+			Identity: contract.ToolIdentitySpec{
+				ProductID:      "chat",
+				Name:           "add_shortcut_to_bar",
+				CanonicalPath:  "chat.add_shortcut_to_bar",
+				CLIPath:        "chat toolbar add",
+				PrimaryCLIPath: "chat toolbar add",
+			},
+			Description: "将快捷入口添加到会话快捷栏可见区",
+			Interface: &contract.InterfaceSpec{
+				Mode:         "mcp",
+				Availability: "available",
+				Ref:          &contract.InterfaceRefSpec{ProductID: "im", RPCName: "add_shortcut_to_bar"},
+			},
+			Selection: contract.SelectionSpec{
+				AgentSummary: "将快捷入口添加到会话快捷栏可见区",
+				UseWhen:      []string{"需要将隐藏的快捷入口重新显示在快捷栏中"},
+				AvoidWhen:    []string{"创建新的自定义入口应使用 chat toolbar create-custom"},
+				Examples:     []string{"dws chat toolbar add --conversation-id <cid> --shortcut-ids 101,102"},
+			},
+			Parameters: []contract.ParamDecl{
+				{Name: "conversation-id", Property: "openConversationId", Required: boolPtr(true)},
+				{Name: "shortcut-ids", Property: "shortcutIds", Required: boolPtr(true)},
+			},
+		},
+	})
+
+	return cmd
+}

internal/helpers/chat/toolbar_create_custom.go

@@ -0,0 +1,143 @@
+// 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 chat
+
+import (
+	"github.com/spf13/cobra"
+
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/corecmd/contract"
+)
+
+func newToolbarCreateCustomCommand() *cobra.Command {
+	cmd := &cobra.Command{
+		Use:   "create-custom",
+		Short: "创建自定义快捷栏入口",
+		Long: `创建一个新的自定义快捷栏入口。
+
+必填参数：
+  --conversation-id  会话 openConversationId
+  --title            入口标题
+  --url              入口跳转链接
+  --icon-url         入口图标 URL
+  --pc-url           PC 端跳转链接（可与 --url 相同）
+
+可选参数：
+  --extension        扩展信息，格式 key=value，可重复使用
+  --desc             入口描述（为空时使用 --title）
+  --tag              入口标签
+  --sort-index       排序权重`,
+		Example: `  dws chat toolbar create-custom --conversation-id <cid> --title "周报" --url "https://example.com" --icon-url "https://example.com/icon.png" --pc-url "https://example.com"
+  dws chat toolbar create-custom --conversation-id <cid> --title "工具" --url "https://example.com" --icon-url "https://example.com/icon.png" --pc-url "https://example.com" --extension color=blue --desc "常用工具"`,
+		RunE: func(cmd *cobra.Command, args []string) error {
+			cid, err := toolbarConversationID(cmd)
+			if err != nil {
+				return err
+			}
+			if err := validateRequiredFlags(cmd, "title", "url", "icon-url", "pc-url"); err != nil {
+				return err
+			}
+
+			toolArgs := map[string]any{
+				"openConversationId": cid,
+				"name":               toolbarI18nJSONString(mustGetFlag(cmd, "title")),
+				"url":                mustGetFlag(cmd, "url"),
+				"icons":              toolbarIconsJSONString(mustGetFlag(cmd, "icon-url")),
+				"pcUrl":              mustGetFlag(cmd, "pc-url"),
+			}
+
+			descValue := mustGetFlag(cmd, "desc")
+			if descValue == "" {
+				descValue = mustGetFlag(cmd, "title")
+			}
+			toolArgs["desc"] = toolbarI18nJSONString(descValue)
+			if tag, _ := cmd.Flags().GetString("tag"); tag != "" {
+				toolArgs["tag"] = tag
+			}
+			if cmd.Flags().Changed("sort-index") {
+				sortIndex, _ := cmd.Flags().GetInt("sort-index")
+				toolArgs["sortIndex"] = sortIndex
+			}
+
+			ext, err := parseExtension(cmd)
+			if err != nil {
+				return err
+			}
+			if ext != nil {
+				toolArgs["extension"] = ext
+			}
+
+			err = callMCPToolOnServer("im", "create_custom_shortcut", toolArgs)
+			if isSystemBusy(err) {
+				return toolbarNewSystemBusyError()
+			}
+			return err
+		},
+	}
+	cmd.Flags().String("conversation-id", "", "会话 openConversationId")
+	cmd.Flags().String("title", "", "入口标题")
+	cmd.Flags().String("url", "", "入口跳转链接")
+	cmd.Flags().String("icon-url", "", "入口图标 URL")
+	cmd.Flags().String("pc-url", "", "PC 端跳转链接")
+	cmd.Flags().StringArray("extension", nil, "扩展信息，格式 key=value，可重复使用")
+	cmd.Flags().String("desc", "", "入口描述（为空时使用 --title）")
+	cmd.Flags().String("tag", "", "入口标签")
+	cmd.Flags().Int("sort-index", 0, "排序权重")
+	_ = cmd.MarkFlagRequired("conversation-id")
+	_ = cmd.MarkFlagRequired("title")
+	_ = cmd.MarkFlagRequired("url")
+	_ = cmd.MarkFlagRequired("icon-url")
+	_ = cmd.MarkFlagRequired("pc-url")
+	cmd.DisableAutoGenTag = true
+
+	DeclareLeafMetadata(cmd, LeafSpec{
+		Safety: contract.SafetySpec{
+			Effect: "write", Risk: "medium",
+			Confirmation: "not_required", Idempotency: "non_idempotent",
+		},
+		Contract: LeafContract{
+			Identity: contract.ToolIdentitySpec{
+				ProductID:      "chat",
+				Name:           "create_custom_shortcut",
+				CanonicalPath:  "chat.create_custom_shortcut",
+				CLIPath:        "chat toolbar create-custom",
+				PrimaryCLIPath: "chat toolbar create-custom",
+			},
+			Description: "创建自定义快捷栏入口",
+			Interface: &contract.InterfaceSpec{
+				Mode:         "mcp",
+				Availability: "available",
+				Ref:          &contract.InterfaceRefSpec{ProductID: "im", RPCName: "create_custom_shortcut"},
+			},
+			Selection: contract.SelectionSpec{
+				AgentSummary: "创建自定义快捷栏入口",
+				UseWhen:      []string{"需要在快捷栏中新增一个自定义入口（含标题、链接、图标等）"},
+				AvoidWhen:    []string{"移动已有入口使用 chat toolbar add / hide；更新已有自定义入口使用 chat toolbar update-custom"},
+				Examples:     []string{"dws chat toolbar create-custom --conversation-id <cid> --title \"周报\" --url \"https://example.com\" --icon-url \"https://example.com/icon.png\" --pc-url \"https://example.com\""},
+			},
+			Parameters: []contract.ParamDecl{
+				{Name: "conversation-id", Property: "openConversationId", Required: boolPtr(true)},
+				{Name: "title", Property: "name", Required: boolPtr(true)},
+				{Name: "url", Property: "url", Required: boolPtr(true)},
+				{Name: "icon-url", Property: "icons", Required: boolPtr(true)},
+				{Name: "pc-url", Property: "pcUrl", Required: boolPtr(true)},
+				{Name: "extension", Property: "extension", Required: boolPtr(false)},
+				{Name: "desc", Property: "desc", Required: boolPtr(false)},
+				{Name: "tag", Property: "tag", Required: boolPtr(false)},
+				{Name: "sort-index", Property: "sortIndex", Required: boolPtr(false)},
+			},
+		},
+	})
+
+	return cmd
+}

internal/helpers/chat/toolbar_custom_payload.go

@@ -0,0 +1,26 @@
+// 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 chat
+
+import "encoding/json"
+
+func toolbarI18nJSONString(value string) string {
+	payload, _ := json.Marshal(map[string]string{"zh_CN": value})
+	return string(payload)
+}
+
+func toolbarIconsJSONString(iconMobile string) string {
+	payload, _ := json.Marshal(map[string]string{"iconMobile": iconMobile})
+	return string(payload)
+}

internal/helpers/chat/toolbar_helpers.go

@@ -0,0 +1,93 @@
+// 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 chat
+
+import (
+	"fmt"
+	"strings"
+
+	"github.com/spf13/cobra"
+
+	apperrors "github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/errors"
+)
+
+func hasIntersection(a, b []int64) bool {
+	if len(a) == 0 || len(b) == 0 {
+		return false
+	}
+	seen := make(map[int64]struct{}, len(a))
+	for _, v := range a {
+		seen[v] = struct{}{}
+	}
+	for _, v := range b {
+		if _, ok := seen[v]; ok {
+			return true
+		}
+	}
+	return false
+}
+
+func isSystemBusy(err error) bool {
+	if err == nil {
+		return false
+	}
+	return strings.Contains(err.Error(), "SYSTEM_BUSY")
+}
+
+func parseExtension(cmd *cobra.Command) (map[string]string, error) {
+	raw, _ := cmd.Flags().GetStringArray("extension")
+	if len(raw) == 0 {
+		return nil, nil
+	}
+	result := make(map[string]string, len(raw))
+	for _, pair := range raw {
+		idx := strings.Index(pair, "=")
+		if idx <= 0 {
+			return nil, fmt.Errorf("--extension 格式错误，期望 key=value，实际 %q", pair)
+		}
+		key := strings.TrimSpace(pair[:idx])
+		value := strings.TrimSpace(pair[idx+1:])
+		if key == "" {
+			return nil, fmt.Errorf("--extension key 不能为空: %q", pair)
+		}
+		if _, exists := result[key]; exists {
+			return nil, fmt.Errorf("--extension 重复 key %q，每个 key 只能指定一次", key)
+		}
+		result[key] = value
+	}
+	return result, nil
+}
+
+func toolbarConversationID(cmd *cobra.Command) (string, error) {
+	cid, _ := cmd.Flags().GetString("conversation-id")
+	cid = strings.TrimSpace(cid)
+	if cid == "" {
+		return "", apperrors.NewValidation(
+			"flag --conversation-id is required",
+			apperrors.WithReason("missing_required_flag"),
+			apperrors.WithHint("使用 --conversation-id 指定会话 openConversationId"),
+			apperrors.WithActions("通过 dws chat search 或 dws chat group list 获取会话 ID"),
+		)
+	}
+	return cid, nil
+}
+
+func toolbarNewSystemBusyError() error {
+	return apperrors.NewValidation(
+		"服务端繁忙（SYSTEM_BUSY），请稍后重试",
+		apperrors.WithReason("system_busy"),
+		apperrors.WithHint("服务端当前负载较高，建议等待几秒后重试"),
+		apperrors.WithActions("等待后使用相同参数重试"),
+	)
+}

internal/helpers/chat/toolbar_helpers_test.go

@@ -0,0 +1,733 @@
+// 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 chat
+
+import (
+	"context"
+	"errors"
+	"io"
+	"reflect"
+	"strings"
+	"testing"
+
+	"github.com/spf13/cobra"
+
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/testseam"
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/pkg/edition"
+)
+
+type toolbarCall struct {
+	product string
+	tool    string
+	args    map[string]any
+}
+
+type toolbarTestCaller struct {
+	product string
+	tool    string
+	args    map[string]any
+	err     error
+	calls   []toolbarCall
+}
+
+func (c *toolbarTestCaller) CallTool(_ context.Context, productID, toolName string, args map[string]any) (*edition.ToolResult, error) {
+	c.calls = append(c.calls, toolbarCall{product: productID, tool: toolName, args: args})
+	c.product = productID
+	c.tool = toolName
+	c.args = args
+	if c.err != nil {
+		return nil, c.err
+	}
+	return &edition.ToolResult{Content: []edition.ContentBlock{{Type: "text", Text: "{}"}}}, nil
+}
+
+func (c *toolbarTestCaller) Format() string { return "json" }
+func (c *toolbarTestCaller) DryRun() bool   { return false }
+func (c *toolbarTestCaller) Fields() string { return "" }
+func (c *toolbarTestCaller) JQ() string     { return "" }
+
+func executeToolbarCommand(t *testing.T, cmd *cobra.Command, caller *toolbarTestCaller, args ...string) error {
+	t.Helper()
+	initToolbarDepsForTest(t, caller)
+	cmd.SetOut(io.Discard)
+	cmd.SetErr(io.Discard)
+	cmd.SetArgs(args)
+	return cmd.Execute()
+}
+
+func initToolbarDepsForTest(t *testing.T, caller *toolbarTestCaller) {
+	t.Helper()
+	previous := deps
+	SetDeps(Deps{
+		GroupRunE: func(cmd *cobra.Command, args []string) error {
+			return cmd.Help()
+		},
+		CallMCPToolOnServer: func(productID, toolName string, args map[string]any) error {
+			_, err := caller.CallTool(context.Background(), productID, toolName, args)
+			return err
+		},
+		DeclareLeafMetadata: func(cmd *cobra.Command, spec LeafSpec) *cobra.Command {
+			return cmd
+		},
+		ValidateRequiredFlag: validateRequiredFlagsLocal,
+	})
+	t.Cleanup(func() {
+		deps = previous
+	})
+}
+
+func requireTypedConfirmationError(t *testing.T, err error) {
+	t.Helper()
+	if err == nil {
+		t.Fatal("expected confirmation error, got nil")
+	}
+	if !strings.Contains(err.Error(), "获得用户确认后加 --yes 执行") {
+		t.Fatalf("expected confirmation error, got %v", err)
+	}
+}
+
+func requireToolbarCall(
+	t *testing.T,
+	caller *toolbarTestCaller,
+	wantProduct, wantTool string,
+	wantArgs map[string]any,
+) {
+	t.Helper()
+	if caller.product != wantProduct || caller.tool != wantTool {
+		t.Fatalf("tool call = %s/%s, want %s/%s", caller.product, caller.tool, wantProduct, wantTool)
+	}
+	if !reflect.DeepEqual(caller.args, wantArgs) {
+		t.Fatalf("tool args = %#v, want %#v", caller.args, wantArgs)
+	}
+}
+
+func toolbarCustomArgs(extra ...string) []string {
+	args := []string{"--conversation-id", "cid", "--title", "Title", "--url", "https://example.com/mobile", "--icon-url", "https://example.com/icon.png", "--pc-url", "https://example.com/pc"}
+	return append(args, extra...)
+}
+
+func TestHasIntersection(t *testing.T) {
+	tests := []struct {
+		name string
+		a, b []int64
+		want bool
+	}{
+		{"both empty", nil, nil, false},
+		{"a empty", nil, []int64{1}, false},
+		{"b empty", []int64{1}, nil, false},
+		{"no overlap", []int64{1, 2, 3}, []int64{4, 5, 6}, false},
+		{"has overlap", []int64{1, 2, 3}, []int64{3, 4, 5}, true},
+		{"single match", []int64{1}, []int64{1}, true},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			got := hasIntersection(tt.a, tt.b)
+			if got != tt.want {
+				t.Errorf("hasIntersection(%v, %v) = %v, want %v", tt.a, tt.b, got, tt.want)
+			}
+		})
+	}
+}
+
+func TestIsSystemBusy(t *testing.T) {
+	if isSystemBusy(nil) {
+		t.Error("isSystemBusy(nil) should be false")
+	}
+	if !isSystemBusy(errors.New("error code: SYSTEM_BUSY")) {
+		t.Error("isSystemBusy should detect SYSTEM_BUSY in error message")
+	}
+	if isSystemBusy(errors.New("some other error")) {
+		t.Error("isSystemBusy should return false for non-SYSTEM_BUSY errors")
+	}
+}
+
+func TestParseExtension(t *testing.T) {
+	cmd := &cobra.Command{Use: "test"}
+	cmd.Flags().StringArray("extension", nil, "")
+
+	ext, err := parseExtension(cmd)
+	if err != nil {
+		t.Fatalf("empty extension should not error: %v", err)
+	}
+	if ext != nil {
+		t.Fatalf("empty extension should return nil, got %v", ext)
+	}
+
+	_ = cmd.Flags().Set("extension", "key1=val1")
+	_ = cmd.Flags().Set("extension", "key2=val2")
+	ext, err = parseExtension(cmd)
+	if err != nil {
+		t.Fatalf("valid extension should not error: %v", err)
+	}
+	if ext["key1"] != "val1" || ext["key2"] != "val2" {
+		t.Fatalf("unexpected extension map: %v", ext)
+	}
+
+	cmd2 := &cobra.Command{Use: "test2"}
+	cmd2.Flags().StringArray("extension", nil, "")
+	_ = cmd2.Flags().Set("extension", "badformat")
+	_, err = parseExtension(cmd2)
+	if err == nil {
+		t.Fatal("invalid extension format should error")
+	}
+
+	cmd3 := &cobra.Command{Use: "test3"}
+	cmd3.Flags().StringArray("extension", nil, "")
+	_ = cmd3.Flags().Set("extension", "color=blue")
+	_ = cmd3.Flags().Set("extension", "color=red")
+	_, err = parseExtension(cmd3)
+	if err == nil {
+		t.Fatal("duplicate extension keys should error")
+	}
+
+	cmd4 := &cobra.Command{Use: "test4"}
+	cmd4.Flags().StringArray("extension", nil, "")
+	_ = cmd4.Flags().Set("extension", " =value")
+	_, err = parseExtension(cmd4)
+	if err == nil {
+		t.Fatal("blank extension key should error")
+	}
+}
+
+func TestToolbarConversationID(t *testing.T) {
+	cmd := &cobra.Command{Use: "test"}
+	cmd.Flags().String("conversation-id", "", "")
+
+	_, err := toolbarConversationID(cmd)
+	if err == nil {
+		t.Fatal("missing conversation-id should error")
+	}
+
+	_ = cmd.Flags().Set("conversation-id", "cid123")
+	cid, err := toolbarConversationID(cmd)
+	if err != nil {
+		t.Fatalf("valid conversation-id should not error: %v", err)
+	}
+	if cid != "cid123" {
+		t.Fatalf("expected cid123, got %s", cid)
+	}
+}
+
+func TestToolbarDepsDefaultWrappersAndValidationEdges(t *testing.T) {
+	cmd := &cobra.Command{Use: "toolbar"}
+	cmd.SetOut(io.Discard)
+	cmd.SetErr(io.Discard)
+	if err := groupRunE(cmd, nil); err != nil {
+		t.Fatalf("default groupRunE returned error: %v", err)
+	}
+	if err := callMCPToolOnServer("im", "tool", nil); err == nil ||
+		!strings.Contains(err.Error(), "CallMCPToolOnServer is not initialized") {
+		t.Fatalf("default callMCPToolOnServer error = %v", err)
+	}
+	if got := DeclareLeafMetadata(cmd, LeafSpec{}); got != cmd {
+		t.Fatal("default DeclareLeafMetadata should return the original command")
+	}
+
+	toolbar := NewChatToolbarCommand()
+	if toolbar.Name() != "toolbar" || len(toolbar.Commands()) != 7 {
+		t.Fatalf("toolbar command = %q with %d children, want toolbar with 7 children", toolbar.Name(), len(toolbar.Commands()))
+	}
+
+	parent := &cobra.Command{Use: "parent"}
+	parent.PersistentFlags().String("conversation-id", "", "")
+	child := &cobra.Command{Use: "child"}
+	parent.AddCommand(child)
+	_ = parent.PersistentFlags().Set("conversation-id", " inherited ")
+	if got := mustGetFlag(child, "conversation-id"); got != "inherited" {
+		t.Fatalf("mustGetFlag inherited value = %q, want inherited", got)
+	}
+
+	if _, err := parseCSVInt64(" , "); err == nil || !strings.Contains(err.Error(), "at least one ID") {
+		t.Fatalf("empty csv error = %v", err)
+	}
+	if _, err := parseCSVInt64("1,bad"); err == nil || !strings.Contains(err.Error(), `invalid integer "bad"`) {
+		t.Fatalf("invalid csv error = %v", err)
+	}
+	required := &cobra.Command{Use: "required"}
+	required.Flags().String("title", "", "")
+	if err := validateRequiredFlagsLocal(required, "title"); err == nil ||
+		!strings.Contains(err.Error(), "flag --title is required") {
+		t.Fatalf("validateRequiredFlagsLocal error = %v", err)
+	}
+	if commandBoolFlag(nil, "yes") {
+		t.Fatal("commandBoolFlag(nil) should be false")
+	}
+	boolParent := &cobra.Command{Use: "parent"}
+	boolParent.PersistentFlags().Bool("yes", false, "")
+	boolChild := &cobra.Command{Use: "child"}
+	boolParent.AddCommand(boolChild)
+	_ = boolParent.PersistentFlags().Set("yes", "true")
+	if !commandBoolFlag(boolChild, "yes") {
+		t.Fatal("commandBoolFlag should read inherited true value")
+	}
+}
+
+func TestToolbarCommandsValidationEdges(t *testing.T) {
+	tests := []struct {
+		name      string
+		cmd       func() *cobra.Command
+		args      []string
+		wantError string
+	}{
+		{
+			name:      "add blank conversation id",
+			cmd:       newToolbarAddCommand,
+			args:      []string{"--conversation-id", " ", "--shortcut-ids", "1"},
+			wantError: "flag --conversation-id is required",
+		},
+		{
+			name:      "add blank shortcut ids",
+			cmd:       newToolbarAddCommand,
+			args:      []string{"--conversation-id", "cid", "--shortcut-ids", " "},
+			wantError: "flag --shortcut-ids is required",
+		},
+		{
+			name:      "add invalid shortcut ids",
+			cmd:       newToolbarAddCommand,
+			args:      []string{"--conversation-id", "cid", "--shortcut-ids", "1,bad"},
+			wantError: "--shortcut-ids",
+		},
+		{
+			name:      "hide blank conversation id",
+			cmd:       newToolbarHideCommand,
+			args:      []string{"--conversation-id", " ", "--shortcut-ids", "1"},
+			wantError: "flag --conversation-id is required",
+		},
+		{
+			name:      "hide blank shortcut ids",
+			cmd:       newToolbarHideCommand,
+			args:      []string{"--conversation-id", "cid", "--shortcut-ids", " "},
+			wantError: "flag --shortcut-ids is required",
+		},
+		{
+			name:      "hide invalid shortcut ids",
+			cmd:       newToolbarHideCommand,
+			args:      []string{"--conversation-id", "cid", "--shortcut-ids", "1,bad"},
+			wantError: "--shortcut-ids",
+		},
+		{
+			name:      "list blank conversation id",
+			cmd:       newToolbarListCommand,
+			args:      []string{"--conversation-id", " "},
+			wantError: "flag --conversation-id is required",
+		},
+		{
+			name:      "sort blank conversation id",
+			cmd:       newToolbarSortCommand,
+			args:      []string{"--conversation-id", " ", "--sorted-ids", "1"},
+			wantError: "flag --conversation-id is required",
+		},
+		{
+			name:      "sort blank sorted ids",
+			cmd:       newToolbarSortCommand,
+			args:      []string{"--conversation-id", "cid", "--sorted-ids", " "},
+			wantError: "flag --sorted-ids is required",
+		},
+		{
+			name:      "sort invalid sorted ids",
+			cmd:       newToolbarSortCommand,
+			args:      []string{"--conversation-id", "cid", "--sorted-ids", "1,bad"},
+			wantError: "--sorted-ids",
+		},
+		{
+			name:      "sort invalid unsorted ids",
+			cmd:       newToolbarSortCommand,
+			args:      []string{"--conversation-id", "cid", "--sorted-ids", "1", "--unsorted-ids", "bad"},
+			wantError: "--unsorted-ids",
+		},
+		{
+			name:      "create blank conversation id",
+			cmd:       newToolbarCreateCustomCommand,
+			args:      toolbarCustomArgs("--conversation-id", " "),
+			wantError: "flag --conversation-id is required",
+		},
+		{
+			name:      "create blank title",
+			cmd:       newToolbarCreateCustomCommand,
+			args:      toolbarCustomArgs("--title", " "),
+			wantError: "flag --title is required",
+		},
+		{
+			name:      "create invalid extension",
+			cmd:       newToolbarCreateCustomCommand,
+			args:      toolbarCustomArgs("--extension", "bad"),
+			wantError: "--extension",
+		},
+		{
+			name:      "remove blank conversation id",
+			cmd:       newToolbarRemoveCustomCommand,
+			args:      []string{"--conversation-id", " ", "--shortcut-id", "42", "--yes"},
+			wantError: "flag --conversation-id is required",
+		},
+		{
+			name:      "update blank conversation id",
+			cmd:       newToolbarUpdateCustomCommand,
+			args:      toolbarCustomArgs("--shortcut-id", "42", "--conversation-id", " "),
+			wantError: "flag --conversation-id is required",
+		},
+		{
+			name:      "update blank title",
+			cmd:       newToolbarUpdateCustomCommand,
+			args:      toolbarCustomArgs("--shortcut-id", "42", "--title", " "),
+			wantError: "flag --title is required",
+		},
+		{
+			name:      "update invalid extension",
+			cmd:       newToolbarUpdateCustomCommand,
+			args:      toolbarCustomArgs("--shortcut-id", "42", "--extension", "bad"),
+			wantError: "--extension",
+		},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			err := executeToolbarCommand(t, tt.cmd(), &toolbarTestCaller{}, tt.args...)
+			if err == nil || !strings.Contains(err.Error(), tt.wantError) {
+				t.Fatalf("error = %v, want containing %q", err, tt.wantError)
+			}
+		})
+	}
+}
+
+func TestToolbarCommandsConvertSystemBusy(t *testing.T) {
+	tests := []struct {
+		name string
+		cmd  func() *cobra.Command
+		args []string
+	}{
+		{
+			name: "add",
+			cmd:  newToolbarAddCommand,
+			args: []string{"--conversation-id", "cid", "--shortcut-ids", "1,2"},
+		},
+		{
+			name: "hide",
+			cmd:  newToolbarHideCommand,
+			args: []string{"--conversation-id", "cid", "--shortcut-ids", "1,2"},
+		},
+		{
+			name: "sort",
+			cmd:  newToolbarSortCommand,
+			args: []string{"--conversation-id", "cid", "--sorted-ids", "1,2"},
+		},
+		{
+			name: "remove-custom",
+			cmd:  newToolbarRemoveCustomCommand,
+			args: []string{"--conversation-id", "cid", "--shortcut-id", "1", "--yes"},
+		},
+		{
+			name: "create-custom",
+			cmd:  newToolbarCreateCustomCommand,
+			args: toolbarCustomArgs(),
+		},
+		{
+			name: "update-custom",
+			cmd:  newToolbarUpdateCustomCommand,
+			args: toolbarCustomArgs("--shortcut-id", "99"),
+		},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			err := executeToolbarCommand(t, tt.cmd(), &toolbarTestCaller{
+				err: errors.New("remote SYSTEM_BUSY"),
+			}, tt.args...)
+			if err == nil || !strings.Contains(err.Error(), "SYSTEM_BUSY") {
+				t.Fatalf("expected system_busy validation error, got %v", err)
+			}
+		})
+	}
+}
+
+func TestToolbarCreateAndUpdateCustomIncludeOptionalFields(t *testing.T) {
+	createCaller := &toolbarTestCaller{}
+	err := executeToolbarCommand(t, newToolbarCreateCustomCommand(), createCaller, toolbarCustomArgs(
+		"--desc", "Description",
+		"--tag", "Tag",
+		"--sort-index", "7",
+		"--extension", "color=blue",
+	)...)
+	if err != nil {
+		t.Fatalf("create-custom returned error: %v", err)
+	}
+	requireToolbarCall(t, createCaller, "im", "create_custom_shortcut", map[string]any{
+		"openConversationId": "cid",
+		"name":               `{"zh_CN":"Title"}`,
+		"url":                "https://example.com/mobile",
+		"icons":              `{"iconMobile":"https://example.com/icon.png"}`,
+		"pcUrl":              "https://example.com/pc",
+		"desc":               `{"zh_CN":"Description"}`,
+		"tag":                "Tag",
+		"sortIndex":          7,
+		"extension":          map[string]string{"color": "blue"},
+	})
+
+	updateCaller := &toolbarTestCaller{}
+	err = executeToolbarCommand(t, newToolbarUpdateCustomCommand(), updateCaller, toolbarCustomArgs(
+		"--shortcut-id", "99",
+		"--desc", "Description",
+		"--tag", "Tag",
+		"--sort-index", "7",
+		"--extension", "color=blue",
+	)...)
+	if err != nil {
+		t.Fatalf("update-custom returned error: %v", err)
+	}
+	requireToolbarCall(t, updateCaller, "im", "update_custom_shortcut", map[string]any{
+		"openConversationId": "cid",
+		"shortcutId":         int64(99),
+		"name":               `{"zh_CN":"Title"}`,
+		"url":                "https://example.com/mobile",
+		"icons":              `{"iconMobile":"https://example.com/icon.png"}`,
+		"pcUrl":              "https://example.com/pc",
+		"desc":               `{"zh_CN":"Description"}`,
+		"tag":                "Tag",
+		"sortIndex":          7,
+		"extension":          map[string]string{"color": "blue"},
+	})
+}
+
+func TestToolbarCreateCustomDefaultsDescToTitle(t *testing.T) {
+	tests := []struct {
+		name string
+		args []string
+	}{
+		{name: "missing desc", args: toolbarCustomArgs()},
+		{name: "blank desc", args: toolbarCustomArgs("--desc", "   ")},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			caller := &toolbarTestCaller{}
+			if err := executeToolbarCommand(t, newToolbarCreateCustomCommand(), caller, tt.args...); err != nil {
+				t.Fatalf("create-custom returned error: %v", err)
+			}
+			requireToolbarCall(t, caller, "im", "create_custom_shortcut", map[string]any{
+				"openConversationId": "cid",
+				"name":               `{"zh_CN":"Title"}`,
+				"url":                "https://example.com/mobile",
+				"icons":              `{"iconMobile":"https://example.com/icon.png"}`,
+				"pcUrl":              "https://example.com/pc",
+				"desc":               `{"zh_CN":"Title"}`,
+			})
+		})
+	}
+}
+
+func TestToolbarUpdateCustomDefaultsDescToTitle(t *testing.T) {
+	tests := []struct {
+		name string
+		args []string
+	}{
+		{name: "missing desc", args: toolbarCustomArgs("--shortcut-id", "99")},
+		{name: "blank desc", args: toolbarCustomArgs("--shortcut-id", "99", "--desc", "   ")},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			caller := &toolbarTestCaller{}
+			if err := executeToolbarCommand(t, newToolbarUpdateCustomCommand(), caller, tt.args...); err != nil {
+				t.Fatalf("update-custom returned error: %v", err)
+			}
+			requireToolbarCall(t, caller, "im", "update_custom_shortcut", map[string]any{
+				"openConversationId": "cid",
+				"shortcutId":         int64(99),
+				"name":               `{"zh_CN":"Title"}`,
+				"url":                "https://example.com/mobile",
+				"icons":              `{"iconMobile":"https://example.com/icon.png"}`,
+				"pcUrl":              "https://example.com/pc",
+				"desc":               `{"zh_CN":"Title"}`,
+			})
+		})
+	}
+}
+
+func TestToolbarSortValidatesOptionalUnsortedIDs(t *testing.T) {
+	intersectionErr := executeToolbarCommand(t, newToolbarSortCommand(), &toolbarTestCaller{},
+		"--conversation-id", "cid",
+		"--sorted-ids", "1,2",
+		"--unsorted-ids", "2,3",
+	)
+	if intersectionErr == nil || !strings.Contains(intersectionErr.Error(), "不能有交集") {
+		t.Fatalf("expected id_intersection error, got %v", intersectionErr)
+	}
+
+	caller := &toolbarTestCaller{}
+	err := executeToolbarCommand(t, newToolbarSortCommand(), caller,
+		"--conversation-id", "cid",
+		"--sorted-ids", "1,2",
+		"--unsorted-ids", "3,4",
+	)
+	if err != nil {
+		t.Fatalf("sort returned error: %v", err)
+	}
+	requireToolbarCall(t, caller, "im", "sort_shortcut_bar", map[string]any{
+		"openConversationId":  "cid",
+		"sortedShortcutIds":   []int64{1, 2},
+		"unsortedShortcutIds": []int64{3, 4},
+	})
+}
+
+func TestCrossPlatformCoverageToolbarAddHideAndListCallMCPWithExactArgs(t *testing.T) {
+	tests := []struct {
+		name     string
+		cmd      func() *cobra.Command
+		args     []string
+		wantTool string
+		wantArgs map[string]any
+	}{
+		{
+			name:     "add",
+			cmd:      newToolbarAddCommand,
+			args:     []string{"--conversation-id", "cidX", "--shortcut-ids", "101,102"},
+			wantTool: "add_shortcut_to_bar",
+			wantArgs: map[string]any{
+				"openConversationId": "cidX",
+				"shortcutIds":        []int64{101, 102},
+			},
+		},
+		{
+			name:     "hide",
+			cmd:      newToolbarHideCommand,
+			args:     []string{"--conversation-id", "cidX", "--shortcut-ids", "201,202"},
+			wantTool: "hide_shortcut_from_bar",
+			wantArgs: map[string]any{
+				"openConversationId": "cidX",
+				"shortcutIds":        []int64{201, 202},
+			},
+		},
+		{
+			name:     "list",
+			cmd:      newToolbarListCommand,
+			args:     []string{"--conversation-id", "cidX"},
+			wantTool: "list_chat_toolbar_shortcuts",
+			wantArgs: map[string]any{
+				"openConversationId": "cidX",
+			},
+		},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			caller := &toolbarTestCaller{}
+			if err := executeToolbarCommand(t, tt.cmd(), caller, tt.args...); err != nil {
+				t.Fatalf("%s returned error: %v", tt.name, err)
+			}
+			requireToolbarCall(t, caller, "im", tt.wantTool, tt.wantArgs)
+		})
+	}
+}
+
+func TestCrossPlatformCoverageToolbarRemoveCustomRejectsWithoutYes(t *testing.T) {
+	caller := &toolbarTestCaller{}
+	err := executeToolbarCommand(t, newToolbarRemoveCustomCommand(), caller,
+		"--conversation-id", "cidX", "--shortcut-id", "42")
+	requireTypedConfirmationError(t, err)
+	if len(caller.calls) != 0 {
+		t.Fatalf("expected 0 MCP calls before --yes, got %d: %+v", len(caller.calls), caller.calls)
+	}
+	if caller.product != "" || caller.tool != "" || caller.args != nil {
+		t.Fatalf("expected no MCP call recorded, got product=%q tool=%q args=%#v", caller.product, caller.tool, caller.args)
+	}
+}
+
+func TestCrossPlatformCoverageToolbarRemoveCustomCallsMCPWithExactArgsWhenYes(t *testing.T) {
+	caller := &toolbarTestCaller{}
+	err := executeToolbarCommand(t, newToolbarRemoveCustomCommand(), caller,
+		"--conversation-id", "cidX", "--shortcut-id", "42", "--yes")
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	if len(caller.calls) != 1 {
+		t.Fatalf("expected exactly 1 MCP call, got %d: %+v", len(caller.calls), caller.calls)
+	}
+	call := caller.calls[0]
+	if call.product != "im" {
+		t.Fatalf("product = %q, want im", call.product)
+	}
+	if call.tool != "remove_custom_shortcut" {
+		t.Fatalf("tool = %q, want remove_custom_shortcut", call.tool)
+	}
+	wantArgs := map[string]any{"openConversationId": "cidX", "shortcutId": int64(42)}
+	if !reflect.DeepEqual(call.args, wantArgs) {
+		t.Fatalf("tool args = %#v, want %#v", call.args, wantArgs)
+	}
+}
+
+type removeCustomSeamCall struct {
+	openConversationId string
+	shortcutId         int64
+}
+
+type removeCustomSeamStub struct {
+	calls []removeCustomSeamCall
+	err   error
+}
+
+// invoke records the seam call and forwards to deps.Caller.CallTool so the
+// command path still records the same remote invocation shape as production.
+func (s *removeCustomSeamStub) invoke(openConversationId string, shortcutId int64) error {
+	s.calls = append(s.calls, removeCustomSeamCall{openConversationId: openConversationId, shortcutId: shortcutId})
+	if err := callMCPToolOnServer("im", "remove_custom_shortcut", map[string]any{
+		"openConversationId": openConversationId,
+		"shortcutId":         shortcutId,
+	}); err != nil {
+		return err
+	}
+	return s.err
+}
+
+func TestCrossPlatformCoverageToolbarRemoveCustomSeamRejectsWithoutYes(t *testing.T) {
+	stub := &removeCustomSeamStub{}
+	testseam.Swap(t, &removeChatToolbarCustomShortcutFn, stub.invoke)
+
+	err := executeToolbarCommand(t, newToolbarRemoveCustomCommand(), &toolbarTestCaller{},
+		"--conversation-id", "cidX", "--shortcut-id", "42")
+	requireTypedConfirmationError(t, err)
+	if len(stub.calls) != 0 {
+		t.Fatalf("expected 0 seam calls before --yes, got %d: %+v", len(stub.calls), stub.calls)
+	}
+}
+
+func TestCrossPlatformCoverageToolbarRemoveCustomSeamCallsWithExactArgsWhenYes(t *testing.T) {
+	stub := &removeCustomSeamStub{}
+	testseam.Swap(t, &removeChatToolbarCustomShortcutFn, stub.invoke)
+
+	err := executeToolbarCommand(t, newToolbarRemoveCustomCommand(), &toolbarTestCaller{},
+		"--conversation-id", "cidX", "--shortcut-id", "42", "--yes")
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	if len(stub.calls) != 1 {
+		t.Fatalf("expected exactly 1 seam call, got %d: %+v", len(stub.calls), stub.calls)
+	}
+	call := stub.calls[0]
+	if call.openConversationId != "cidX" {
+		t.Fatalf("seam openConversationId = %q, want cidX", call.openConversationId)
+	}
+	if call.shortcutId != int64(42) {
+		t.Fatalf("seam shortcutId = %d, want 42", call.shortcutId)
+	}
+}
+
+func TestToolbarRemoveCustomDeclaresDestructiveHighRiskConfirmation(t *testing.T) {
+	var got LeafSpec
+	next := deps
+	next.DeclareLeafMetadata = func(cmd *cobra.Command, spec LeafSpec) *cobra.Command {
+		got = spec
+		return cmd
+	}
+	testseam.Swap(t, &deps, next)
+
+	_ = newToolbarRemoveCustomCommand()
+	if got.Safety.Effect != "destructive" || got.Safety.Risk != "high" ||
+		got.Safety.Confirmation != "user_required" {
+		t.Fatalf("remove-custom safety = %#v, want destructive/high/user_required", got.Safety)
+	}
+}

internal/helpers/chat/toolbar_hide.go

@@ -0,0 +1,92 @@
+// 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 chat
+
+import (
+	"fmt"
+
+	"github.com/spf13/cobra"
+
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/corecmd/contract"
+)
+
+func newToolbarHideCommand() *cobra.Command {
+	cmd := &cobra.Command{
+		Use:   "hide",
+		Short: "将入口从快捷栏可见区隐藏",
+		Long:  "将一个或多个快捷入口从可见区移至隐藏区。入口 ID 使用逗号分隔。",
+		Example: `  dws chat toolbar hide --conversation-id <cid> --shortcut-ids 101,102
+  # 查询入口 ID: dws chat toolbar list --conversation-id <cid>`,
+		RunE: func(cmd *cobra.Command, args []string) error {
+			cid, err := toolbarConversationID(cmd)
+			if err != nil {
+				return err
+			}
+			if err := validateRequiredFlags(cmd, "shortcut-ids"); err != nil {
+				return err
+			}
+			ids, err := parseCSVInt64(mustGetFlag(cmd, "shortcut-ids"))
+			if err != nil {
+				return fmt.Errorf("--shortcut-ids: %w", err)
+			}
+			err = callMCPToolOnServer("im", "hide_shortcut_from_bar", map[string]any{
+				"openConversationId": cid,
+				"shortcutIds":        ids,
+			})
+			if isSystemBusy(err) {
+				return toolbarNewSystemBusyError()
+			}
+			return err
+		},
+	}
+	cmd.Flags().String("conversation-id", "", "会话 openConversationId")
+	cmd.Flags().String("shortcut-ids", "", "入口 ID 列表（逗号分隔）")
+	_ = cmd.MarkFlagRequired("conversation-id")
+	_ = cmd.MarkFlagRequired("shortcut-ids")
+	cmd.DisableAutoGenTag = true
+
+	DeclareLeafMetadata(cmd, LeafSpec{
+		Safety: contract.SafetySpec{
+			Effect: "write", Risk: "low",
+			Confirmation: "not_required", Idempotency: "idempotent",
+		},
+		Contract: LeafContract{
+			Identity: contract.ToolIdentitySpec{
+				ProductID:      "chat",
+				Name:           "hide_shortcut_from_bar",
+				CanonicalPath:  "chat.hide_shortcut_from_bar",
+				CLIPath:        "chat toolbar hide",
+				PrimaryCLIPath: "chat toolbar hide",
+			},
+			Description: "将快捷入口从会话快捷栏可见区移至隐藏区",
+			Interface: &contract.InterfaceSpec{
+				Mode:         "mcp",
+				Availability: "available",
+				Ref:          &contract.InterfaceRefSpec{ProductID: "im", RPCName: "hide_shortcut_from_bar"},
+			},
+			Selection: contract.SelectionSpec{
+				AgentSummary: "将快捷入口从会话快捷栏可见区隐藏",
+				UseWhen:      []string{"需要将快捷栏中的入口隐藏到隐藏区"},
+				AvoidWhen:    []string{"永久删除自定义入口应使用 chat toolbar remove-custom"},
+				Examples:     []string{"dws chat toolbar hide --conversation-id <cid> --shortcut-ids 101,102"},
+			},
+			Parameters: []contract.ParamDecl{
+				{Name: "conversation-id", Property: "openConversationId", Required: boolPtr(true)},
+				{Name: "shortcut-ids", Property: "shortcutIds", Required: boolPtr(true)},
+			},
+		},
+	})
+
+	return cmd
+}

internal/helpers/chat/toolbar_list.go

@@ -0,0 +1,75 @@
+// 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 chat
+
+import (
+	"github.com/spf13/cobra"
+
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/corecmd/contract"
+)
+
+func newToolbarListCommand() *cobra.Command {
+	cmd := &cobra.Command{
+		Use:   "list",
+		Short: "查询会话快捷栏入口列表",
+		Long:  "查询指定会话的快捷栏入口列表，包括可见区和隐藏区的所有入口。",
+		Example: `  dws chat toolbar list --conversation-id <cid>
+  # 查询群 ID: dws chat search --query "群名"`,
+		RunE: func(cmd *cobra.Command, args []string) error {
+			cid, err := toolbarConversationID(cmd)
+			if err != nil {
+				return err
+			}
+			return callMCPToolOnServer("im", "list_chat_toolbar_shortcuts", map[string]any{
+				"openConversationId": cid,
+			})
+		},
+	}
+	cmd.Flags().String("conversation-id", "", "会话 openConversationId")
+	_ = cmd.MarkFlagRequired("conversation-id")
+	cmd.DisableAutoGenTag = true
+
+	DeclareLeafMetadata(cmd, LeafSpec{
+		Safety: contract.SafetySpec{
+			Effect: "read", Risk: "low",
+			Confirmation: "not_required", Idempotency: "idempotent",
+		},
+		Contract: LeafContract{
+			Identity: contract.ToolIdentitySpec{
+				ProductID:      "chat",
+				Name:           "list_chat_toolbar_shortcuts",
+				CanonicalPath:  "chat.list_chat_toolbar_shortcuts",
+				CLIPath:        "chat toolbar list",
+				PrimaryCLIPath: "chat toolbar list",
+			},
+			Description: "查询会话快捷栏入口列表",
+			Interface: &contract.InterfaceSpec{
+				Mode:         "mcp",
+				Availability: "available",
+				Ref:          &contract.InterfaceRefSpec{ProductID: "im", RPCName: "list_chat_toolbar_shortcuts"},
+			},
+			Selection: contract.SelectionSpec{
+				AgentSummary: "查询会话快捷栏入口列表",
+				UseWhen:      []string{"需要查看会话快捷栏中有哪些入口、获取入口 ID 时"},
+				AvoidWhen:    []string{"修改快捷栏排序使用 chat toolbar sort；创建自定义入口使用 chat toolbar create-custom"},
+				Examples:     []string{"dws chat toolbar list --conversation-id <cid>"},
+			},
+			Parameters: []contract.ParamDecl{
+				{Name: "conversation-id", Property: "openConversationId", Required: boolPtr(true)},
+			},
+		},
+	})
+
+	return cmd
+}

internal/helpers/chat/toolbar_remove_custom.go

@@ -0,0 +1,103 @@
+// 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 chat
+
+import (
+	"github.com/spf13/cobra"
+
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/corecmd/contract"
+	apperrors "github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/errors"
+)
+
+// removeChatToolbarCustomShortcutFn is the package-level injection seam for
+// the remove-custom command's remote call. The default implementation routes
+// through callMCPToolOnServer against the im server, so production behavior is
+// unchanged. Tests swap it with testseam.Swap(t, &removeChatToolbarCustomShortcutFn, stub)
+// to capture the call without hitting a real remote. Like every package-var
+// seam in this repo it is not safe for t.Parallel tests.
+var removeChatToolbarCustomShortcutFn = func(openConversationId string, shortcutId int64) error {
+	return callMCPToolOnServer("im", "remove_custom_shortcut", map[string]any{
+		"openConversationId": openConversationId,
+		"shortcutId":         shortcutId,
+	})
+}
+
+func newToolbarRemoveCustomCommand() *cobra.Command {
+	cmd := &cobra.Command{
+		Use:     "remove-custom",
+		Short:   "删除自定义快捷栏入口",
+		Long:    "删除一个自定义快捷栏入口。该操作不可逆；必须先获得用户确认，再追加 --yes 执行。",
+		Example: `  dws chat toolbar remove-custom --conversation-id <cid> --shortcut-id 123`,
+		RunE: func(cmd *cobra.Command, args []string) error {
+			cid, err := toolbarConversationID(cmd)
+			if err != nil {
+				return err
+			}
+			shortcutId, _ := cmd.Flags().GetInt64("shortcut-id")
+			if !commandBoolFlag(cmd, "yes") {
+				return apperrors.NewValidation(
+					"删除自定义快捷入口不可逆；获得用户确认后加 --yes 执行",
+					apperrors.WithReason("confirmation_required"),
+					apperrors.WithHint("先确认目标入口及影响范围；用户明确同意后以相同参数追加 --yes"),
+					apperrors.WithActions("确认目标自定义入口", "获得用户确认后使用 --yes 执行"),
+				)
+			}
+			err = removeChatToolbarCustomShortcutFn(cid, shortcutId)
+			if isSystemBusy(err) {
+				return toolbarNewSystemBusyError()
+			}
+			return err
+		},
+	}
+	cmd.Flags().String("conversation-id", "", "会话 openConversationId")
+	cmd.Flags().Int64("shortcut-id", 0, "自定义入口 ID")
+	cmd.Flags().Bool("yes", false, "确认执行删除操作")
+	_ = cmd.MarkFlagRequired("conversation-id")
+	_ = cmd.MarkFlagRequired("shortcut-id")
+	cmd.DisableAutoGenTag = true
+
+	DeclareLeafMetadata(cmd, LeafSpec{
+		Safety: contract.SafetySpec{
+			Effect: "destructive", Risk: "high",
+			Confirmation: "user_required", Idempotency: "idempotent",
+		},
+		Contract: LeafContract{
+			Identity: contract.ToolIdentitySpec{
+				ProductID:      "chat",
+				Name:           "remove_custom_shortcut",
+				CanonicalPath:  "chat.remove_custom_shortcut",
+				CLIPath:        "chat toolbar remove-custom",
+				PrimaryCLIPath: "chat toolbar remove-custom",
+			},
+			Description: "删除自定义快捷栏入口",
+			Interface: &contract.InterfaceSpec{
+				Mode:         "mcp",
+				Availability: "available",
+				Ref:          &contract.InterfaceRefSpec{ProductID: "im", RPCName: "remove_custom_shortcut"},
+			},
+			Selection: contract.SelectionSpec{
+				AgentSummary: "删除自定义快捷栏入口",
+				UseWhen:      []string{"需要永久删除某个自定义快捷栏入口"},
+				AvoidWhen:    []string{"仅隐藏入口使用 chat toolbar hide；更新入口使用 chat toolbar update-custom"},
+				Examples:     []string{"dws chat toolbar remove-custom --conversation-id <cid> --shortcut-id 123"},
+			},
+			Parameters: []contract.ParamDecl{
+				{Name: "conversation-id", Property: "openConversationId", Required: boolPtr(true)},
+				{Name: "shortcut-id", Property: "shortcutId", Required: boolPtr(true)},
+			},
+		},
+	})
+
+	return cmd
+}

internal/helpers/chat/toolbar_sort.go

@@ -0,0 +1,115 @@
+// 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 chat
+
+import (
+	"fmt"
+
+	"github.com/spf13/cobra"
+
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/corecmd/contract"
+	apperrors "github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/errors"
+)
+
+func newToolbarSortCommand() *cobra.Command {
+	cmd := &cobra.Command{
+		Use:   "sort",
+		Short: "排序快捷栏入口",
+		Long:  "对会话快捷栏入口进行排序。sorted-ids 指定排序后的可见区入口 ID 列表，unsorted-ids 指定不参与排序放在末尾的入口 ID 列表。两个列表不能有交集。",
+		Example: `  dws chat toolbar sort --conversation-id <cid> --sorted-ids 101,102,103
+  dws chat toolbar sort --conversation-id <cid> --sorted-ids 101,102 --unsorted-ids 103,104
+  # 查询入口 ID: dws chat toolbar list --conversation-id <cid>`,
+		RunE: func(cmd *cobra.Command, args []string) error {
+			cid, err := toolbarConversationID(cmd)
+			if err != nil {
+				return err
+			}
+			if err := validateRequiredFlags(cmd, "sorted-ids"); err != nil {
+				return err
+			}
+			sortedIds, err := parseCSVInt64(mustGetFlag(cmd, "sorted-ids"))
+			if err != nil {
+				return fmt.Errorf("--sorted-ids: %w", err)
+			}
+
+			toolArgs := map[string]any{
+				"openConversationId": cid,
+				"sortedShortcutIds":  sortedIds,
+			}
+
+			if unsortedRaw, _ := cmd.Flags().GetString("unsorted-ids"); unsortedRaw != "" {
+				unsortedIds, err := parseCSVInt64(unsortedRaw)
+				if err != nil {
+					return fmt.Errorf("--unsorted-ids: %w", err)
+				}
+				if hasIntersection(sortedIds, unsortedIds) {
+					return apperrors.NewValidation(
+						"--sorted-ids 与 --unsorted-ids 不能有交集",
+						apperrors.WithReason("id_intersection"),
+						apperrors.WithHint("检查两个 ID 列表，确保同一个 ID 不同时出现在两个列表中"),
+						apperrors.WithActions("修正 ID 列表后重试"),
+					)
+				}
+				toolArgs["unsortedShortcutIds"] = unsortedIds
+			}
+
+			err = callMCPToolOnServer("im", "sort_shortcut_bar", toolArgs)
+			if isSystemBusy(err) {
+				return toolbarNewSystemBusyError()
+			}
+			return err
+		},
+	}
+	cmd.Flags().String("conversation-id", "", "会话 openConversationId")
+	cmd.Flags().String("sorted-ids", "", "排序后的入口 ID 列表（逗号分隔）")
+	cmd.Flags().String("unsorted-ids", "", "不参与排序放在末尾的入口 ID 列表（逗号分隔）")
+	_ = cmd.MarkFlagRequired("conversation-id")
+	_ = cmd.MarkFlagRequired("sorted-ids")
+	cmd.DisableAutoGenTag = true
+
+	DeclareLeafMetadata(cmd, LeafSpec{
+		Safety: contract.SafetySpec{
+			Effect: "write", Risk: "low",
+			Confirmation: "not_required", Idempotency: "idempotent",
+		},
+		Contract: LeafContract{
+			Identity: contract.ToolIdentitySpec{
+				ProductID:      "chat",
+				Name:           "sort_shortcut_bar",
+				CanonicalPath:  "chat.sort_shortcut_bar",
+				CLIPath:        "chat toolbar sort",
+				PrimaryCLIPath: "chat toolbar sort",
+			},
+			Description: "对会话快捷栏入口进行排序",
+			Interface: &contract.InterfaceSpec{
+				Mode:         "mcp",
+				Availability: "available",
+				Ref:          &contract.InterfaceRefSpec{ProductID: "im", RPCName: "sort_shortcut_bar"},
+			},
+			Selection: contract.SelectionSpec{
+				AgentSummary: "对会话快捷栏入口进行排序",
+				UseWhen:      []string{"需要调整快捷栏入口的显示顺序"},
+				AvoidWhen:    []string{"仅添加或隐藏入口使用 chat toolbar add / hide"},
+				Examples:     []string{"dws chat toolbar sort --conversation-id <cid> --sorted-ids 101,102,103"},
+			},
+			Parameters: []contract.ParamDecl{
+				{Name: "conversation-id", Property: "openConversationId", Required: boolPtr(true)},
+				{Name: "sorted-ids", Property: "sortedShortcutIds", Required: boolPtr(true)},
+				{Name: "unsorted-ids", Property: "unsortedShortcutIds", Required: boolPtr(false)},
+			},
+		},
+	})
+
+	return cmd
+}

internal/helpers/chat/toolbar_update_custom.go

@@ -0,0 +1,149 @@
+// 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 chat
+
+import (
+	"github.com/spf13/cobra"
+
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/corecmd/contract"
+)
+
+func newToolbarUpdateCustomCommand() *cobra.Command {
+	cmd := &cobra.Command{
+		Use:   "update-custom",
+		Short: "更新自定义快捷栏入口",
+		Long: `更新一个已有的自定义快捷栏入口。
+
+必填参数：
+  --conversation-id  会话 openConversationId
+  --shortcut-id      自定义入口 ID
+  --title            入口标题
+  --url              入口跳转链接
+  --icon-url         入口图标 URL
+  --pc-url           PC 端跳转链接
+
+可选参数：
+  --extension        扩展信息，格式 key=value，可重复使用
+  --desc             入口描述
+  --tag              入口标签
+  --sort-index       排序权重`,
+		Example: `  dws chat toolbar update-custom --conversation-id <cid> --shortcut-id 123 --title "周报" --url "https://example.com" --icon-url "https://example.com/icon.png" --pc-url "https://example.com"
+  # 查询入口 ID: dws chat toolbar list --conversation-id <cid>`,
+		RunE: func(cmd *cobra.Command, args []string) error {
+			cid, err := toolbarConversationID(cmd)
+			if err != nil {
+				return err
+			}
+			shortcutId, _ := cmd.Flags().GetInt64("shortcut-id")
+			if err := validateRequiredFlags(cmd, "title", "url", "icon-url", "pc-url"); err != nil {
+				return err
+			}
+
+			toolArgs := map[string]any{
+				"openConversationId": cid,
+				"shortcutId":         shortcutId,
+				"name":               toolbarI18nJSONString(mustGetFlag(cmd, "title")),
+				"url":                mustGetFlag(cmd, "url"),
+				"icons":              toolbarIconsJSONString(mustGetFlag(cmd, "icon-url")),
+				"pcUrl":              mustGetFlag(cmd, "pc-url"),
+			}
+
+			descValue := mustGetFlag(cmd, "desc")
+			if descValue == "" {
+				descValue = mustGetFlag(cmd, "title")
+			}
+			toolArgs["desc"] = toolbarI18nJSONString(descValue)
+			if tag, _ := cmd.Flags().GetString("tag"); tag != "" {
+				toolArgs["tag"] = tag
+			}
+			if cmd.Flags().Changed("sort-index") {
+				sortIndex, _ := cmd.Flags().GetInt("sort-index")
+				toolArgs["sortIndex"] = sortIndex
+			}
+
+			ext, err := parseExtension(cmd)
+			if err != nil {
+				return err
+			}
+			if ext != nil {
+				toolArgs["extension"] = ext
+			}
+
+			err = callMCPToolOnServer("im", "update_custom_shortcut", toolArgs)
+			if isSystemBusy(err) {
+				return toolbarNewSystemBusyError()
+			}
+			return err
+		},
+	}
+	cmd.Flags().String("conversation-id", "", "会话 openConversationId")
+	cmd.Flags().Int64("shortcut-id", 0, "自定义入口 ID")
+	cmd.Flags().String("title", "", "入口标题")
+	cmd.Flags().String("url", "", "入口跳转链接")
+	cmd.Flags().String("icon-url", "", "入口图标 URL")
+	cmd.Flags().String("pc-url", "", "PC 端跳转链接")
+	cmd.Flags().StringArray("extension", nil, "扩展信息，格式 key=value，可重复使用")
+	cmd.Flags().String("desc", "", "入口描述")
+	cmd.Flags().String("tag", "", "入口标签")
+	cmd.Flags().Int("sort-index", 0, "排序权重")
+	_ = cmd.MarkFlagRequired("conversation-id")
+	_ = cmd.MarkFlagRequired("shortcut-id")
+	_ = cmd.MarkFlagRequired("title")
+	_ = cmd.MarkFlagRequired("url")
+	_ = cmd.MarkFlagRequired("icon-url")
+	_ = cmd.MarkFlagRequired("pc-url")
+	cmd.DisableAutoGenTag = true
+
+	DeclareLeafMetadata(cmd, LeafSpec{
+		Safety: contract.SafetySpec{
+			Effect: "write", Risk: "medium",
+			Confirmation: "not_required", Idempotency: "idempotent",
+		},
+		Contract: LeafContract{
+			Identity: contract.ToolIdentitySpec{
+				ProductID:      "chat",
+				Name:           "update_custom_shortcut",
+				CanonicalPath:  "chat.update_custom_shortcut",
+				CLIPath:        "chat toolbar update-custom",
+				PrimaryCLIPath: "chat toolbar update-custom",
+			},
+			Description: "更新自定义快捷栏入口",
+			Interface: &contract.InterfaceSpec{
+				Mode:         "mcp",
+				Availability: "available",
+				Ref:          &contract.InterfaceRefSpec{ProductID: "im", RPCName: "update_custom_shortcut"},
+			},
+			Selection: contract.SelectionSpec{
+				AgentSummary: "更新自定义快捷栏入口",
+				UseWhen:      []string{"需要修改已有自定义快捷栏入口的标题、链接、图标等属性"},
+				AvoidWhen:    []string{"创建新入口使用 chat toolbar create-custom；删除入口使用 chat toolbar remove-custom"},
+				Examples:     []string{"dws chat toolbar update-custom --conversation-id <cid> --shortcut-id 123 --title \"周报\" --url \"https://example.com\" --icon-url \"https://example.com/icon.png\" --pc-url \"https://example.com\""},
+			},
+			Parameters: []contract.ParamDecl{
+				{Name: "conversation-id", Property: "openConversationId", Required: boolPtr(true)},
+				{Name: "shortcut-id", Property: "shortcutId", Required: boolPtr(true)},
+				{Name: "title", Property: "name", Required: boolPtr(true)},
+				{Name: "url", Property: "url", Required: boolPtr(true)},
+				{Name: "icon-url", Property: "icons", Required: boolPtr(true)},
+				{Name: "pc-url", Property: "pcUrl", Required: boolPtr(true)},
+				{Name: "extension", Property: "extension", Required: boolPtr(false)},
+				{Name: "desc", Property: "desc", Required: boolPtr(false)},
+				{Name: "tag", Property: "tag", Required: boolPtr(false)},
+				{Name: "sort-index", Property: "sortIndex", Required: boolPtr(false)},
+			},
+		},
+	})
+
+	return cmd
+}

internal/helpers/chat_favorites_test.go

@@ -190,23 +190,23 @@ func TestChatListFavoritesSuppliesOpenDefaults(t *testing.T) {
 func TestChatListFavoritesMapsExplicitPagination(t *testing.T) {
 	caller := &chatFavoritesCaller{}
 	err := executeChatFavoritesCommand(t, caller,
-		"message", "list-favorites", "--cursor", "42", "--size", "50")
+		"message", "list-favorites", "--cursor", "42", "--size", "30")
 	if err != nil {
 		t.Fatalf("list-favorites returned error: %v", err)
 	}
 
-	want := map[string]any{"cursor": int64(42), "size": "50"}
+	want := map[string]any{"cursor": int64(42), "size": "30"}
 	if len(caller.calls) != 1 || caller.calls[0].productID != "im" || caller.calls[0].toolName != "list_message_favorites" || !reflect.DeepEqual(caller.calls[0].args, want) {
 		t.Fatalf("calls = %#v, want im/list_message_favorites %#v", caller.calls, want)
 	}
 }
 
 func TestChatListFavoritesRejectsInvalidSize(t *testing.T) {
-	for _, size := range []string{"-1", "0", "101"} {
+	for _, size := range []string{"-1", "0", "31"} {
 		t.Run(size, func(t *testing.T) {
 			caller := &chatFavoritesCaller{}
 			err := executeChatFavoritesCommand(t, caller, "message", "list-favorites", "--size", size)
-			if err == nil || !strings.Contains(err.Error(), "--size must be between 1 and 100") {
+			if err == nil || !strings.Contains(err.Error(), "--size must be between 1 and 30") {
 				t.Fatalf("error = %v, want size validation error", err)
 			}
 			if got := apperrors.ExitCode(err); got != 3 {

internal/helpers/chat_message_help_test.go

@@ -0,0 +1,92 @@
+// 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 helpers
+
+import (
+	"bytes"
+	"strings"
+	"testing"
+)
+
+func TestCrossPlatformCoverageChatMessageHelpDocumentsPostSendIDChain(t *testing.T) {
+	tests := []struct {
+		name       string
+		command    string
+		contains   []string
+		notContain string
+	}{
+		{
+			name:    "send returns task ID",
+			command: "send",
+			contains: []string{
+				"openTaskId",
+				"query-send-status --open-task-id <openTaskId>",
+			},
+		},
+		{
+			name:    "query returns message and conversation IDs",
+			command: "query-send-status",
+			contains: []string{
+				"openTaskId",
+				"openMessageId",
+				"openConversationId",
+				"chat message edit",
+				"chat message recall",
+			},
+		},
+		{
+			name:    "edit includes post-send workflow",
+			command: "edit",
+			contains: []string{
+				"send -> query-send-status -> edit",
+				"query-send-status --open-task-id <上一步返回的openTaskId>",
+				"edit --conversation-id <上一步返回的openConversationId> --msg-id <上一步返回的openMessageId>",
+			},
+			notContain: "chat message list",
+		},
+		{
+			name:    "recall includes post-send workflow",
+			command: "recall",
+			contains: []string{
+				"send -> query-send-status -> recall",
+				"query-send-status --open-task-id <上一步返回的openTaskId>",
+				"recall --conversation-id <上一步返回的openConversationId> --msg-id <上一步返回的openMessageId>",
+			},
+			notContain: "chat message list",
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			cmd := newChatCommand()
+			var output bytes.Buffer
+			cmd.SetOut(&output)
+			cmd.SetErr(&output)
+			cmd.SetArgs([]string{"message", test.command, "--help"})
+			if err := cmd.Execute(); err != nil {
+				t.Fatalf("chat message %s --help: %v\n%s", test.command, err, output.String())
+			}
+
+			help := output.String()
+			for _, want := range test.contains {
+				if !strings.Contains(help, want) {
+					t.Errorf("chat message %s help missing %q:\n%s", test.command, want, help)
+				}
+			}
+			if test.notContain != "" && strings.Contains(help, test.notContain) {
+				t.Errorf("chat message %s help still contains %q:\n%s", test.command, test.notContain, help)
+			}
+		})
+	}
+}

internal/helpers/chat_safety_catalog_test.go

@@ -0,0 +1,62 @@
+// Copyright 2026 Alibaba Group
+// Licensed under the Apache License, Version 2.0 (the "License");
+
+package helpers
+
+import (
+	"strings"
+	"testing"
+
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/corecmd/contractfinal"
+	"github.com/spf13/cobra"
+)
+
+func TestCrossPlatformCoverageChatCatalogSafetyMetadataIsExplicit(t *testing.T) {
+	root := &cobra.Command{Use: "dws"}
+	root.AddCommand(newChatCommand())
+
+	checked := 0
+	walkChatCatalogLeaves(root, func(cmd *cobra.Command) {
+		final, ok := contractfinal.RuntimeContractFinal(cmd)
+		if !ok || final.Identity == nil || final.Safety == nil {
+			return
+		}
+		if strings.TrimSpace(final.Identity.ProductID) != "chat" {
+			return
+		}
+		checked++
+		canonical := strings.TrimSpace(final.Identity.CanonicalPath)
+		safety := final.Safety
+		if safety.Effect == "" || safety.Risk == "" ||
+			safety.Confirmation == "" || safety.Idempotency == "" {
+			t.Fatalf("chat Catalog leaf %s has incomplete Safety: %+v", canonical, *safety)
+		}
+		if safety.Effect == "read" {
+			if safety.Risk != "low" ||
+				safety.Confirmation != "not_required" ||
+				safety.Idempotency != "idempotent" {
+				t.Fatalf("chat read Catalog leaf %s Safety = %+v, want read/low/not_required/idempotent",
+					canonical, *safety)
+			}
+		}
+	})
+	if checked == 0 {
+		t.Fatal("no chat Catalog leaves checked")
+	}
+}
+
+func walkChatCatalogLeaves(cmd *cobra.Command, fn func(*cobra.Command)) {
+	if cmd == nil {
+		return
+	}
+	if cmd.Runnable() && !cmd.HasSubCommands() {
+		fn(cmd)
+		return
+	}
+	for _, child := range cmd.Commands() {
+		if child.Name() == "help" {
+			continue
+		}
+		walkChatCatalogLeaves(child, fn)
+	}
+}

internal/helpers/chat_toolbar.go

@@ -0,0 +1,34 @@
+// 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 helpers
+
+import (
+	"github.com/spf13/cobra"
+
+	chathelpers "github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/helpers/chat"
+)
+
+func newChatToolbarCommand() *cobra.Command {
+	chathelpers.SetDeps(chathelpers.Deps{
+		GroupRunE:           GroupRunE,
+		CallMCPToolOnServer: CallMCPToolOnServer,
+		DeclareLeafMetadata: func(cmd *cobra.Command, spec chathelpers.LeafSpec) *cobra.Command {
+			return DeclareLeafMetadata(cmd, LeafSpec{
+				Safety:   spec.Safety,
+				Contract: spec.Contract,
+			})
+		},
+	})
+	return chathelpers.NewChatToolbarCommand()
+}

internal/helpers/chat_toolbar_test.go

@@ -0,0 +1,52 @@
+package helpers
+
+import (
+	"context"
+	"io"
+	"reflect"
+	"testing"
+
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/pkg/edition"
+)
+
+type chatToolbarParentCaller struct {
+	product string
+	tool    string
+	args    map[string]any
+}
+
+func (c *chatToolbarParentCaller) CallTool(_ context.Context, productID, toolName string, args map[string]any) (*edition.ToolResult, error) {
+	c.product = productID
+	c.tool = toolName
+	c.args = args
+	return textToolResult(`{"ok":true}`), nil
+}
+
+func (*chatToolbarParentCaller) Format() string { return "json" }
+func (*chatToolbarParentCaller) DryRun() bool   { return false }
+func (*chatToolbarParentCaller) Fields() string { return "" }
+func (*chatToolbarParentCaller) JQ() string     { return "" }
+
+func TestCrossPlatformCoverageChatToolbarMountedThroughParentDeps(t *testing.T) {
+	caller := &chatToolbarParentCaller{}
+	installHelpersCoreDeps(t, caller)
+
+	root := newChatCommand()
+	root.SetOut(io.Discard)
+	root.SetErr(io.Discard)
+	if command, _, err := root.Find([]string{"toolbar", "list"}); err != nil || command == nil || command.Name() != "list" {
+		t.Fatalf("find chat toolbar list = command %v, err %v", command, err)
+	}
+
+	root.SetArgs([]string{"toolbar", "list", "--conversation-id", "cidX"})
+	if err := root.Execute(); err != nil {
+		t.Fatalf("chat toolbar list returned error: %v", err)
+	}
+	if caller.product != "im" || caller.tool != "list_chat_toolbar_shortcuts" {
+		t.Fatalf("tool call = %s/%s, want im/list_chat_toolbar_shortcuts", caller.product, caller.tool)
+	}
+	wantArgs := map[string]any{"openConversationId": "cidX"}
+	if !reflect.DeepEqual(caller.args, wantArgs) {
+		t.Fatalf("tool args = %#v, want %#v", caller.args, wantArgs)
+	}
+}

internal/helpers/coverage_pure_more_test.go

@@ -170,7 +170,6 @@ func TestCrossPlatformCoverageChartMailAndSheetPureHelpersCoverage(t *testing.T)
 		_ = isEmptyDataValidation(value)
 		_ = isEmptyHyperlink(value)
 	}
-	_ = fillIntMatrix(2, 3, 4)
 	_ = maxColLenStr([][]string{{"a"}, {"b", "c"}})
 	_ = maxColLen2D([][]int{{1}, {2, 3}})
 	_ = checkMatrixShape(1, 2, 1, 2, "x")
@@ -213,7 +212,7 @@ func TestCrossPlatformCoverageChartMailAndSheetPureHelpersCoverage(t *testing.T)
 		{styleSpec{}, 1, 1},
 	}
 	for _, style := range styles {
-		_ = applyStyleSpec(&style.spec, style.rows, style.cols, map[string]any{})
+		_, _ = buildStyleCells(&style.spec, style.rows, style.cols)
 	}
 	for _, tc := range []struct {
 		scalar, raw string
@@ -227,7 +226,9 @@ func TestCrossPlatformCoverageChartMailAndSheetPureHelpersCoverage(t *testing.T)
 		{"", `[[""]]`, hAlignEnum},
 		{"", `[["left"]]`, hAlignEnum},
 	} {
-		_ = apply2DString(tc.scalar, tc.raw, 1, 1, "align", "alignments", tc.enum, map[string]any{})
+		if get, err := strGrid(tc.scalar, tc.raw, "align", tc.enum, 1, 1); err == nil && get != nil {
+			_, _ = get(0, 0)
+		}
 	}
 	views := []map[string]any{{"id": "one"}, {"filterViewId": "two"}}
 	_, _ = findFilterViewByID(views, "one")

internal/helpers/destructive_confirmation_test.go

@@ -81,6 +81,53 @@ func requireTypedConfirmationError(t *testing.T, err error) {
 	}
 }
 
+func TestChatChmodRequiresConfirmationBeforeRemoteCall(t *testing.T) {
+	caller := &guardedMutationCaller{}
+	err := executeGuardedMutationCommand(t, caller, newChatCommand,
+		"chmod", "chat.message:send",
+		"--agentCode", "wukong",
+		"--grant-type", "timed",
+		"--ttl", "24h",
+		"--conversation-id", "cidX",
+	)
+	requireTypedConfirmationError(t, err)
+	if len(caller.calls) != 0 {
+		t.Fatalf("expected 0 MCP calls before --yes, got %d: %+v", len(caller.calls), caller.calls)
+	}
+}
+
+func TestChatChmodCallsRemoteWithExactArgsWhenConfirmed(t *testing.T) {
+	caller := &guardedMutationCaller{}
+	err := executeGuardedMutationCommand(t, caller, newChatCommand,
+		"chmod", "chat.message:send",
+		"--agentCode", "wukong",
+		"--grant-type", "timed",
+		"--ttl", "24h",
+		"--conversation-id", "cidX",
+		"--yes",
+	)
+	if err != nil {
+		t.Fatalf("chat chmod with --yes returned error: %v", err)
+	}
+	if len(caller.calls) != 1 {
+		t.Fatalf("expected exactly 1 MCP call, got %d: %+v", len(caller.calls), caller.calls)
+	}
+	call := caller.calls[0]
+	if call.productID != "im" || call.toolName != "chat_permission_grant" {
+		t.Fatalf("tool call = %s/%s, want im/chat_permission_grant", call.productID, call.toolName)
+	}
+	wantArgs := map[string]any{
+		"agentCode":   "wukong",
+		"scope":       "chat.message:send",
+		"grantType":   "timed",
+		"ttl":         "24h",
+		"grantParams": `{"conversationId":"cidX","openCid":"cidX","openConversationId":"cidX"}`,
+	}
+	if !reflect.DeepEqual(call.args, wantArgs) {
+		t.Fatalf("tool args = %#v, want %#v", call.args, wantArgs)
+	}
+}
+
 type guardedStepCaller struct {
 	calls  []guardedMutationCall
 	steps  []string

internal/helpers/doc.go

@@ -221,6 +221,43 @@ func runDocUpload(cmd *cobra.Command, _ []string) error {
 	return callMCPTool("commit_uploaded_file", commitArgs)
 }
 
+// docSpaceUploadCommitText 执行文档空间三步上传（凭证 → PUT → 入库）并
+// 返回 commit 响应原文，供 doc import 的白名单外回退链路组装结构化结果。
+// 与 runDocUpload 的区别：不打印输出、不携带 doc upload 的 --workspace
+// 兼容告警，调用方负责结果投影。
+func docSpaceUploadCommitText(ctx context.Context, filePath, fileName string, fileSize int64, folder, workspace string) (string, error) {
+	step1Args := map[string]any{}
+	if folder != "" {
+		step1Args["folderId"] = folder
+	}
+	if workspace != "" {
+		step1Args["workspaceId"] = workspace
+	}
+	text, err := callMCPToolReturnText(ctx, "get_file_upload_info", step1Args)
+	if err != nil {
+		return "", err
+	}
+	resourceURL, uploadKey, ossHeaders, err := parseUploadInfo(text)
+	if err != nil {
+		return "", err
+	}
+	if err := httpPutFile(ctx, resourceURL, ossHeaders, filePath, fileSize); err != nil {
+		return "", err
+	}
+	commitArgs := map[string]any{
+		"uploadKey": uploadKey,
+		"name":      fileName,
+		"fileSize":  float64(fileSize),
+	}
+	if folder != "" {
+		commitArgs["folderId"] = folder
+	}
+	if workspace != "" {
+		commitArgs["workspaceId"] = workspace
+	}
+	return callMCPToolReturnText(ctx, "commit_uploaded_file", commitArgs)
+}
+
 // parseUploadInfo extracts resourceUrl, uploadKey and headers from the MCP tool response.
 func parseUploadInfo(text string) (resourceURL, uploadKey string, headers map[string]string, err error) {
 	var data map[string]any
@@ -495,16 +532,23 @@ func runMediaInsert(cmd *cobra.Command, _ []string) error {
 	fileSize := fileInfo.Size()
 
 	if deps.Caller.DryRun() {
-		deps.Out.PrintKeyValue("操作", "上传附件并插入文档")
-		deps.Out.PrintKeyValue("文档", nodeID)
-		deps.Out.PrintKeyValue("文件", filePath)
-		deps.Out.PrintKeyValue("名称", fileName)
-		deps.Out.PrintKeyValue("类型", mimeType)
-		deps.Out.PrintKeyValue("大小", fmt.Sprintf("%d bytes", fileSize))
-		return nil
+		return deps.Out.PrintJSON(map[string]any{
+			"contractVersion": "doc.operation.v1",
+			"dry_run":         true,
+			"preview_kind":    "plan",
+			"ok":              true,
+			"status":          "success",
+			"complete":        true,
+			"operation":       "doc.media_insert",
+			"data": map[string]any{
+				"executed": false, "nodeId": nodeID, "file": filePath,
+				"fileName": fileName, "mimeType": mimeType, "sizeBytes": fileSize,
+			},
+			"steps": []map[string]any{{"name": "validate_local_file", "status": "success"}},
+		})
 	}
 
-	ctx := context.Background()
+	ctx := cmd.Context()
 
 	// Step 1: get upload credentials (uploadUrl + resourceId)
 	deps.Out.PrintInfo(fmt.Sprintf("[1/3] 获取附件上传凭证 (%s, %d bytes)...", fileName, fileSize))
@@ -531,7 +575,20 @@ func runMediaInsert(cmd *cobra.Command, _ []string) error {
 		"Content-Type": mimeType,
 	}
 	if err := httpPutFile(ctx, uploadURL, ossHeaders, filePath, fileSize); err != nil {
-		return err
+		return apperrors.NewAPI(
+			"附件上传结果未知；尚未确认正文 block 已插入，请先检查文档媒体列表，禁止改用手写 HTTP",
+			apperrors.WithOperation("doc.media_insert"),
+			apperrors.WithReason("doc_media_upload_unknown"),
+			apperrors.WithFailureStage("upload_oss"),
+			apperrors.WithExecutionStarted(true),
+			apperrors.WithRetryable(false),
+			apperrors.WithActions("运行 dws doc +media-list 检查当前文档", "确认没有对应媒体后才重新执行 +media-insert", "不要 curl 上传地址或安装本地依赖"),
+			apperrors.WithDetails(map[string]any{
+				"contractVersion": "doc.operation.v1", "status": "unknown", "nodeId": nodeID,
+				"resourceId": resourceID, "fileName": fileName, "stage": "upload_oss",
+			}),
+			apperrors.WithCause(err),
+		)
 	}
 
 	// Step 3: insert block into document
@@ -588,15 +645,43 @@ func runMediaInsert(cmd *cobra.Command, _ []string) error {
 	}
 
 	if err := callMCPTool("insert_document_block", insertArgs); err != nil {
-		return err
+		return apperrors.NewAPI(
+			"附件已上传，但正文 block 插入结果未知；请先检查媒体列表，不要重复上传或插入",
+			apperrors.WithOperation("doc.media_insert"),
+			apperrors.WithReason("doc_media_insert_partial"),
+			apperrors.WithFailureStage("insert_block"),
+			apperrors.WithExecutionStarted(true),
+			apperrors.WithRetryable(false),
+			apperrors.WithActions("运行 dws doc +media-list 检查 resourceId/blockId", "不要直接重试 +media-insert，不要使用 resourceUrl 手写请求"),
+			apperrors.WithDetails(map[string]any{
+				"contractVersion": "doc.operation.v1", "status": "partial_success", "nodeId": nodeID,
+				"resourceId": resourceID, "resourceUrl": resourceURL, "fileName": fileName,
+				"steps": []map[string]any{
+					{"name": "resolve_upload", "status": "success"},
+					{"name": "upload_oss", "status": "success"},
+					{"name": "insert_block", "status": "unknown"},
+				},
+			}),
+			apperrors.WithCause(err),
+		)
 	}
 
-	if strings.HasPrefix(mimeType, "image/") {
-		deps.Out.PrintInfo(fmt.Sprintf("图片已插入文档: %s (resourceUrl=%s)", fileName, resourceURL))
-	} else {
-		deps.Out.PrintInfo(fmt.Sprintf("附件已插入文档: %s (resourceId=%s)", fileName, resourceID))
-	}
-	return nil
+	return deps.Out.PrintJSON(map[string]any{
+		"contractVersion": "doc.operation.v1",
+		"ok":              true,
+		"status":          "success",
+		"complete":        true,
+		"operation":       "doc.media_insert",
+		"data": map[string]any{
+			"nodeId": nodeID, "resourceId": resourceID, "resourceUrl": resourceURL,
+			"fileName": fileName, "mimeType": mimeType, "sizeBytes": fileSize, "inserted": true,
+		},
+		"steps": []map[string]any{
+			{"name": "resolve_upload", "status": "success"},
+			{"name": "upload_oss", "status": "success"},
+			{"name": "insert_block", "status": "success"},
+		},
+	})
 }
 
 // parseAttachmentUploadInfo extracts uploadUrl, resourceId and resourceUrl from the MCP tool response.
@@ -3764,6 +3849,9 @@ CLI 内部自动完成全部流程：
   xlsx, xls   → 电子表格
   md, txt     → 文字文档
   xmind, mark → 脑图
+  其他格式（html/pdf/zip 等）→ 不做在线文档转换，自动改走文件上传链路，
+  以原文件形式存入 --folder/--workspace 指定位置；如需在线文档请先转换
+  为 md；上传到钉盘请用 dws drive upload
 
 文件大小限制: 20MB
 
@@ -3791,8 +3879,8 @@ CLI 内部自动完成全部流程:
 		},
 	}
 	importCmd.Flags().String("file", "", "本地文件路径 (必填)")
-	importCmd.Flags().String("folder", "", "目标文件夹 ID 或 URL (可选，与 --workspace 至少传一个)")
-	importCmd.Flags().String("workspace", "", "目标知识库 ID 或 URL (可选，与 --folder 至少传一个)")
+	importCmd.Flags().String("folder", "", "目标文件夹 ID 或 URL (可选；folder/workspace 都不传时导入到默认根目录)")
+	importCmd.Flags().String("workspace", "", "目标知识库 ID 或 URL (可选；folder/workspace 都不传时导入到默认根目录)")
 	importCmd.Flags().StringP("name", "n", "", "导入后文档名称 (可选，默认取文件名)")
 	importCmd.Flags().String("folder-id", "", "")
 	_ = importCmd.Flags().MarkHidden("folder-id")

internal/helpers/doc_write_pipeline.go

@@ -9,6 +9,7 @@ import (
 	"syscall"
 	"unicode/utf8"
 
+	apperrors "github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/errors"
 	"github.com/spf13/cobra"
 )
 
@@ -17,9 +18,6 @@ const (
 	// Server-side OSS delta resolution is now fixed, so large chunks are safe.
 	initialChunkSize = 10000
 
-	// minChunkSize is the floor; below this we report an error instead of retrying.
-	minChunkSize = 5000
-
 	// longContentWarningThreshold triggers a hint to use --content-file.
 	longContentWarningThreshold = 2048
 )
@@ -95,10 +93,11 @@ func docWritePipeline(cmd *cobra.Command, toolName string, toolArgs map[string]a
 		// Single write path
 		nodeID, lastResponse, writeErr = singleWrite(ctx, toolName, toolArgs)
 		chunksWritten = 1
-		if writeErr != nil && isTimeoutError(writeErr.Error()) && runeCount > minChunkSize {
-			// Timeout on single write — fallback to chunked with halved size
-			deps.Out.PrintInfo("[INFO] 单次写入超时，自动切换为分片写入...")
-			nodeID, chunksWritten, lastResponse, writeErr = chunkedWrite(ctx, toolName, toolArgs, markdown, operation, initialChunkSize/2)
+		if writeErr != nil && isTimeoutError(writeErr.Error()) {
+			// The server may have committed the write before the client observed the
+			// timeout. Replaying create/append here can duplicate a document or
+			// content, so fail closed and require inspection before any retry.
+			writeErr = docWriteUnknownStateError(operation, nodeID, "single_write", 0, 1, writeErr)
 		}
 	} else {
 		// Chunked write path
@@ -156,6 +155,9 @@ func chunkedWrite(ctx context.Context, toolName string, toolArgs map[string]any,
 			len(chunks), utf8.RuneCountInString(chunks[0])))
 		resultText, err := callMCPToolReturnText(ctx, "create_document", createArgs)
 		if err != nil {
+			if isTimeoutError(err.Error()) {
+				return "", 0, resultText, docWriteUnknownStateError(operation, "", "chunk_1", 0, len(chunks), err)
+			}
 			return "", 0, resultText, fmt.Errorf("创建文档失败: %w", err)
 		}
 		nodeID = extractNodeIDFromResult(resultText)
@@ -181,6 +183,9 @@ func chunkedWrite(ctx context.Context, toolName string, toolArgs map[string]any,
 			len(chunks), utf8.RuneCountInString(chunks[0]), firstMode))
 		resultText, err := callMCPToolReturnText(ctx, "update_document", updateArgs)
 		if err != nil {
+			if isTimeoutError(err.Error()) {
+				return nodeID, 0, resultText, docWriteUnknownStateError(operation, nodeID, "chunk_1", 0, len(chunks), err)
+			}
 			return nodeID, 0, resultText, fmt.Errorf("第 1 片写入失败: %w", err)
 		}
 		lastResponse = resultText
@@ -209,29 +214,9 @@ func chunkedWrite(ctx context.Context, toolName string, toolArgs map[string]any,
 		resultText, err := callMCPToolReturnText(ctx, "update_document", updateArgs)
 		if err != nil {
 			if isTimeoutError(err.Error()) {
-				newSize := chunkSize / 2
-				if newSize < minChunkSize {
-					return nodeID, writtenCount, resultText, &CLIError{
-						Code: CodeContentTruncated,
-						Message: fmt.Sprintf("分片写入持续超时，已写入 %d 片。当前分片大小 %d 字符已低于最小阈值 %d",
-							writtenCount, chunkSize, minChunkSize),
-						Suggestion: fmt.Sprintf("后端写入超时无法恢复。已成功写入部分内容到 nodeId=%s，请使用 dws doc read --node %s 查看已写入部分",
-							nodeID, nodeID),
-						Operation: operation,
-					}
-				}
-				deps.Out.PrintInfo(fmt.Sprintf("[INFO] 写入超时，分片大小减半为 %d 字符后重试...", newSize))
-				chunkSize = newSize
-
-				var remaining strings.Builder
-				remaining.WriteString(chunk)
-				for j := i + 1; j < len(chunks); j++ {
-					remaining.WriteString(chunks[j])
-				}
-				newChunks := splitMarkdownSafe(remaining.String(), chunkSize)
-				chunks = append(chunks[:i], newChunks...)
-				i-- // retry current index
-				continue
+				return nodeID, writtenCount, resultText, docWriteUnknownStateError(
+					operation, nodeID, fmt.Sprintf("chunk_%d", i+1), writtenCount, len(chunks), err,
+				)
 			}
 			return nodeID, writtenCount, resultText, fmt.Errorf("分片 %d 写入失败: %w", writtenCount+1, err)
 		}
@@ -243,6 +228,27 @@ func chunkedWrite(ctx context.Context, toolName string, toolArgs map[string]any,
 	return nodeID, writtenCount, lastResponse, nil
 }
 
+func docWriteUnknownStateError(operation, nodeID, stage string, written, total int, cause error) error {
+	details := map[string]any{
+		"status":        "unknown",
+		"nodeId":        nodeID,
+		"chunksWritten": written,
+		"chunksTotal":   total,
+		"failedStage":   stage,
+	}
+	return apperrors.NewAPI(
+		"文档写入响应超时，服务端提交状态未知；为避免重复创建或重复追加，已停止自动重试",
+		apperrors.WithOperation(operation),
+		apperrors.WithReason("doc_write_commit_unknown"),
+		apperrors.WithFailureStage(stage),
+		apperrors.WithExecutionStarted(true),
+		apperrors.WithRetryable(false),
+		apperrors.WithActions("先读取目标文档确认实际写入状态", "仅在确认服务端未提交后重新执行"),
+		apperrors.WithDetails(details),
+		apperrors.WithCause(cause),
+	)
+}
+
 // isTimeoutError checks if an error message indicates a server-side timeout.
 func isTimeoutError(msg string) bool {
 	lower := strings.ToLower(msg)

internal/helpers/doc_write_pipeline_coverage_more_test.go

@@ -7,6 +7,7 @@ import (
 	"strings"
 	"testing"
 
+	apperrors "github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/errors"
 	"github.com/spf13/cobra"
 )
 
@@ -40,10 +41,12 @@ func TestCrossPlatformCoverageDocWritePipelineStrategyRemainingCoverage(t *testi
 		t.Fatalf("long literal single write: %v", err)
 	}
 
-	fallback := strings.Repeat("x", minChunkSize+100)
-	installScriptedCaller(t, &scriptedToolCaller{steps: []scriptedToolStep{{err: errors.New("timeout")}, {text: `{}`}, {text: `{}`}}})
-	if err := docWritePipeline(docWriteCoverageCommand(), "update_document", map[string]any{"nodeId": "node", "mode": "overwrite", "markdown": fallback}, fallback, "update"); err != nil {
-		t.Fatalf("single timeout fallback: %v", err)
+	fallback := strings.Repeat("x", 5100)
+	timeoutCaller := &scriptedToolCaller{steps: []scriptedToolStep{{err: errors.New("timeout")}, {text: `{}`}}}
+	installScriptedCaller(t, timeoutCaller)
+	err := docWritePipeline(docWriteCoverageCommand(), "update_document", map[string]any{"nodeId": "node", "mode": "overwrite", "markdown": fallback}, fallback, "update")
+	if err == nil || !strings.Contains(err.Error(), "提交状态未知") || timeoutCaller.calls != 1 {
+		t.Fatalf("single timeout must stop without replay: err=%v calls=%d", err, timeoutCaller.calls)
 	}
 
 	chunked := strings.Repeat("x", initialChunkSize+100)
@@ -58,9 +61,43 @@ func TestCrossPlatformCoverageChunkedWriteAdaptiveRetryRemainingCoverage(t *test
 	os.Args = []string{"dws", "doc"}
 	t.Cleanup(func() { os.Args = oldArgs })
 	markdown := strings.Repeat("x", 24000)
-	installScriptedCaller(t, &scriptedToolCaller{steps: []scriptedToolStep{{text: `{}`}, {err: errors.New("HSFTimeoutException")}, {text: `{}`}}})
+	caller := &scriptedToolCaller{steps: []scriptedToolStep{{text: `{}`}, {err: errors.New("HSFTimeoutException")}, {text: `{}`}}}
+	installScriptedCaller(t, caller)
 	_, written, _, err := chunkedWrite(context.Background(), "update_document", map[string]any{"nodeId": "node"}, markdown, "update", 10000)
-	if err != nil || written < 3 {
-		t.Fatalf("adaptive retry written=%d err=%v", written, err)
+	if err == nil || written != 1 || caller.calls != 2 || !strings.Contains(err.Error(), "提交状态未知") {
+		t.Fatalf("timeout must stop without replay: written=%d calls=%d err=%v", written, caller.calls, err)
+	}
+}
+
+func TestCrossPlatformCoverageDocWriteFirstChunkTimeoutIsUnknown(t *testing.T) {
+	oldArgs := os.Args
+	os.Args = []string{"dws", "doc"}
+	t.Cleanup(func() { os.Args = oldArgs })
+	markdown := strings.Repeat("x", 24000)
+	tests := []struct {
+		name     string
+		tool     string
+		args     map[string]any
+		wantNode string
+	}{
+		{name: "create", tool: "create_document", args: map[string]any{"name": "doc"}},
+		{name: "update", tool: "update_document", args: map[string]any{"nodeId": "node", "mode": "overwrite"}, wantNode: "node"},
+	}
+	for _, tc := range tests {
+		t.Run(tc.name, func(t *testing.T) {
+			caller := &scriptedToolCaller{steps: []scriptedToolStep{{err: errors.New("HSFTimeoutException")}, {text: `{}`}}}
+			installScriptedCaller(t, caller)
+			nodeID, written, _, err := chunkedWrite(context.Background(), tc.tool, tc.args, markdown, tc.name, 10000)
+			var typed *apperrors.Error
+			if err == nil || !errors.As(err, &typed) {
+				t.Fatalf("error = %#v", err)
+			}
+			if nodeID != tc.wantNode || written != 0 || caller.calls != 1 {
+				t.Fatalf("node=%q written=%d calls=%d", nodeID, written, caller.calls)
+			}
+			if typed.Reason != "doc_write_commit_unknown" || typed.FailureStage != "chunk_1" || typed.ExecutionStarted == nil || !*typed.ExecutionStarted || !typed.RetryableSet || typed.Retryable {
+				t.Fatalf("unknown commit metadata = %#v", typed)
+			}
+		})
 	}
 }

internal/helpers/helpers.go

@@ -210,8 +210,16 @@ func parseMCPToolTextResult(serverID, toolName string, result *edition.ToolResul
 		}
 		return "", WrapError(err)
 	}
+	if result == nil {
+		return "", &CLIError{
+			Code:       CodeMCPToolError,
+			Message:    "MCP 工具返回 nil result，无法判断操作结果",
+			Suggestion: "不要把空回复当作成功；请重试并携带 operation/trace 信息排查服务端",
+			Operation:  serverID + "/" + toolName,
+		}
+	}
 	for _, c := range result.Content {
-		if c.Type == "text" && c.Text != "" {
+		if c.Type == "text" && strings.TrimSpace(c.Text) != "" {
 			dumpRawToolResponse(serverID, toolName, c.Text)
 			var errBody map[string]any
 			if json.Unmarshal([]byte(c.Text), &errBody) == nil {
@@ -243,6 +251,12 @@ func parseMCPToolTextResult(serverID, toolName string, result *edition.ToolResul
 			return c.Text, nil
 		}
 	}
+	// Some legacy tools intentionally use an empty text response as an
+	// acknowledgement. This low-level parser cannot know the business contract,
+	// so preserve that representation. Data-returning callers must validate the
+	// operation-specific shape (for example records/tables/valid) before they
+	// report success. Orchestrators whose contract requires a business result use
+	// CallMCPWriteDataStrict and may prove an unknown effect by independent read-back.
 	return "", nil
 }
 

internal/helpers/helpers_core_test.go

@@ -196,7 +196,11 @@ func TestCrossPlatformCoverageMCPReturnTextClassification(t *testing.T) {
 	}
 	caller.result = &edition.ToolResult{Content: []edition.ContentBlock{{Type: "image", Text: "ignored"}, {Type: "text"}}}
 	if got, err := callMCPToolReturnTextOnServer(context.Background(), "server", "tool", nil); err != nil || got != "" {
-		t.Fatalf("empty text result = %q, %v", got, err)
+		t.Fatalf("low-level empty text representation = %q, %v", got, err)
+	}
+	caller.result = nil
+	if _, err := callMCPToolReturnTextOnServer(context.Background(), "server", "tool", nil); err == nil {
+		t.Fatal("nil result must fail")
 	}
 	caller.result = textToolResult("shortcut result")
 	if got, err := CallMCPToolTextOnServer("server", "tool", nil); err != nil || got != "shortcut result" {

internal/helpers/import_flow.go

@@ -58,6 +58,11 @@ type importFlowConfig struct {
 	timeoutAsResult      bool
 	nextCommand          string
 	poll                 importPollPolicy
+	// uploadFallback 开启后，所有不在 supportedFormats 白名单内的文件
+	// （html/pdf/zip/无扩展名等）不再报错断链，统一移交文档空间文件上传
+	// 链路原样入库；白名单即后端转换能力的封闭集合，无需第二份格式枚举。
+	// 回退共享 prepareImportFile 的存在性 / 20MB / 空文件校验。
+	uploadFallback bool
 }
 
 type preparedImportFile struct {
@@ -98,6 +103,10 @@ func docImportFlowConfig() importFlowConfig {
 		workspaceFlags:       []string{"workspace", "workspace-id"},
 		nextCommand:          "dws doc import get --task-id %s",
 		poll:                 defaultImportPollPolicy(),
+		// 白名单外的格式改走文档空间的文件上传链路
+		// （与 drive upload --workspace 同一条 doc-space 上传原语），
+		// 目标 flags（--folder/--workspace）与 import 同构，链路不中断。
+		uploadFallback: true,
 	}
 }
 
@@ -165,7 +174,11 @@ func prepareImportFile(cmd *cobra.Command, args []string, cfg importFlowConfig)
 	}
 
 	extension := strings.TrimPrefix(strings.ToLower(filepath.Ext(filePath)), ".")
-	if !cfg.supportedFormats[extension] {
+	// 非回退配置保持基线校验顺序：扩展名门禁先于导入目标校验
+	// （sheet import 对无目标的非 Excel 文件必须先报 unsupported）。
+	// uploadFallback 配置的白名单外文件继续走完共享校验，由
+	// runImportCommand 分派到上传回退。
+	if !cfg.supportedFormats[extension] && !cfg.uploadFallback {
 		return preparedImportFile{}, fmt.Errorf("unsupported file format %q, supported: %s", extension, cfg.supportedFormatsText)
 	}
 
@@ -197,11 +210,122 @@ func (cfg importFlowConfig) callTool(ctx context.Context, toolName string, args
 	return callMCPToolReturnText(ctx, toolName, args)
 }
 
+// runImportUploadFallback 承接白名单外格式：不再报错断链，改走文档空间
+// 文件上传链路原样入库。回退在 prepareImportFile 之后执行，共享存在性 /
+// 20MB / 空文件校验；不复用 runDocUpload，避免携带 doc upload 的
+// --workspace 兼容告警。移交事实通过 stderr 显式告知，机器可读结果统一
+// 携带 fallback=upload / converted=false 标记，防止 Agent 误判已完成
+// 在线文档转换。
+func runImportUploadFallback(cmd *cobra.Command, cfg importFlowConfig, file preparedImportFile) error {
+	label := file.extension
+	if label == "" {
+		label = "无扩展名"
+	}
+	deps.Out.PrintWarning(fmt.Sprintf(
+		"%s 文件不支持转换为在线文档（支持: %s），已自动改走文件上传链路，以原文件形式存入 --folder/--workspace 指定的目标位置；如需在线文档，请先将内容转换为 md 后重新执行 doc import；上传到钉盘请用 dws drive upload",
+		label, cfg.supportedFormatsText))
+
+	// prepareImportFile 的 name 去掉了扩展名；上传保留原始文件名形态
+	uploadName := file.name
+	if filepath.Ext(uploadName) == "" && file.extension != "" {
+		uploadName += "." + file.extension
+	}
+	jsonMode := deps.Caller.Format() == "json"
+
+	if deps.Caller.DryRun() {
+		if jsonMode {
+			return deps.Out.PrintJSON(map[string]any{
+				"dry_run":             true,
+				"executed":            false,
+				"preview_kind":        "plan",
+				"operation":           "上传文件到钉钉文档",
+				"requested_operation": cfg.operation,
+				"fallback":            "upload",
+				"converted":           false,
+				"file":                file.path,
+				"name":                uploadName,
+				"format":              file.extension,
+				"size":                file.size,
+			})
+		}
+		deps.Out.PrintKeyValue("操作", "上传文件到钉钉文档（doc import 回退）")
+		deps.Out.PrintKeyValue("文件", file.path)
+		deps.Out.PrintKeyValue("名称", uploadName)
+		deps.Out.PrintKeyValue("格式", file.extension)
+		deps.Out.PrintKeyValue("大小", fmt.Sprintf("%d bytes", file.size))
+		return nil
+	}
+
+	ctx := cmd.Context()
+	if ctx == nil {
+		ctx = context.Background()
+	}
+	ctx, cancel := context.WithTimeout(ctx, 10*time.Minute)
+	defer cancel()
+
+	if !jsonMode {
+		deps.Out.PrintInfo("按原文件上传中（未转换为在线文档）...")
+	}
+	text, err := docSpaceUploadCommitText(ctx, file.path, uploadName, file.size, file.folder, file.workspace)
+	if err != nil {
+		return err
+	}
+	// fail-closed：commit 响应必须可解析且带文件标识才算成功；
+	// 空响应（legacy ack）、非 JSON 或缺少标识都不得包装为 success
+	commit, dentryID, err := parseUploadCommitResult(text)
+	if err != nil {
+		return err
+	}
+	return deps.Out.PrintJSON(map[string]any{
+		"success":             true,
+		"operation":           "上传文件到钉钉文档",
+		"requested_operation": cfg.operation,
+		"fallback":            "upload",
+		"converted":           false,
+		"name":                uploadName,
+		"format":              file.extension,
+		"dentry_id":           dentryID,
+		"result":              commit,
+	})
+}
+
+// uploadCommitIDKeys 是 commit_uploaded_file 响应中可作为文件标识的字段，
+// 按优先级排列；服务端可能返回平铺对象或包一层 result envelope。
+var uploadCommitIDKeys = []string{"dentryUuid", "dentryId", "nodeId", "fileId", "id"}
+
+// parseUploadCommitResult 校验入库响应：拒绝空响应，要求 JSON 对象且
+// 含文件标识，返回解析后的对象与标识值。任何不满足都返回错误，
+// 由调用方向用户提示核对入库结果，而不是伪装成功。
+func parseUploadCommitResult(text string) (map[string]any, string, error) {
+	trimmed := strings.TrimSpace(text)
+	if trimmed == "" {
+		return nil, "", fmt.Errorf("上传入库未返回结果（commit_uploaded_file 响应为空），无法确认文件已入库；请用 dws doc list 核对目标位置")
+	}
+	var parsed map[string]any
+	if err := json.Unmarshal([]byte(trimmed), &parsed); err != nil {
+		return nil, "", fmt.Errorf("上传入库响应无法解析为 JSON，无法确认文件已入库；原始响应: %s", trimmed)
+	}
+	payload := parsed
+	if inner, ok := parsed["result"].(map[string]any); ok {
+		payload = inner
+	}
+	for _, key := range uploadCommitIDKeys {
+		if v, ok := payload[key].(string); ok && strings.TrimSpace(v) != "" {
+			return parsed, v, nil
+		}
+	}
+	return nil, "", fmt.Errorf("上传入库响应缺少文件标识（%s 均为空），无法确认文件已入库；原始响应: %s", strings.Join(uploadCommitIDKeys, "/"), trimmed)
+}
+
 func runImportCommand(cmd *cobra.Command, args []string, cfg importFlowConfig) error {
 	file, err := prepareImportFile(cmd, args, cfg)
 	if err != nil {
 		return err
 	}
+	// 非回退配置的白名单外文件已在 prepareImportFile 中按基线顺序拒绝
+	if cfg.uploadFallback && !cfg.supportedFormats[file.extension] {
+		return runImportUploadFallback(cmd, cfg, file)
+	}
 	jsonMode := deps.Caller.Format() == "json"
 
 	if deps.Caller.DryRun() {

internal/helpers/import_html_redirect_test.go

@@ -0,0 +1,379 @@
+// 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 helpers
+
+import (
+	"bytes"
+	"context"
+	"encoding/json"
+	"errors"
+	"os"
+	"path/filepath"
+	"strings"
+	"testing"
+
+	"github.com/spf13/cobra"
+)
+
+var (
+	errUploadInfo = errors.New("upload info failed")
+	errUploadPut  = errors.New("put failed")
+)
+
+func htmlFallbackCommand(t *testing.T, filePath string) *cobra.Command {
+	t.Helper()
+	// callMCPToolReturnText 从 os.Args 解析产品名（doc）
+	oldArgs := os.Args
+	os.Args = []string{"dws", "doc"}
+	t.Cleanup(func() { os.Args = oldArgs })
+	cmd := &cobra.Command{Use: "import"}
+	cmd.Flags().String("file", "", "")
+	cmd.Flags().String("name", "", "")
+	cmd.Flags().String("folder", "", "")
+	cmd.Flags().String("workspace", "", "")
+	cmd.Flags().String("folder-id", "", "")
+	cmd.Flags().String("workspace-id", "", "")
+	if filePath != "" {
+		if err := cmd.Flags().Set("file", filePath); err != nil {
+			t.Fatalf("set import file: %v", err)
+		}
+	}
+	return cmd
+}
+
+func TestCrossPlatformCoverageDocImportHTMLUploadRedirect(t *testing.T) {
+	uploadSteps := []scriptedToolStep{
+		{text: `{"resourceUrl":"https://upload.example.test/object","uploadKey":"key-1"}`},
+		{text: `{"dentryUuid":"node-1","name":"sales.html"}`},
+	}
+
+	t.Run("html upload fallback emits marked json without legacy warnings", func(t *testing.T) {
+		caller := &scriptedToolCaller{format: "json", steps: uploadSteps}
+		installScriptedCaller(t, caller)
+		var stdout, warnings bytes.Buffer
+		deps.Out.w = &stdout
+		deps.Out.errW = &warnings
+		SetHTTPPutFile(func(context.Context, string, map[string]string, string, int64) error { return nil })
+		t.Cleanup(func() { SetHTTPPutFile(nil) })
+
+		cmd := htmlFallbackCommand(t, writeImportFixture(t, "html"))
+		if err := cmd.Flags().Set("workspace", "ws-1"); err != nil {
+			t.Fatal(err)
+		}
+		if err := runImportCommand(cmd, nil, docImportFlowConfig()); err != nil {
+			t.Fatalf("runImportCommand() error = %v, want upload fallback success", err)
+		}
+		if caller.calls != 2 || caller.tool != "commit_uploaded_file" {
+			t.Fatalf("fallback calls = %d last tool = %q, want 2 calls ending in commit_uploaded_file", caller.calls, caller.tool)
+		}
+		if got := caller.args["workspaceId"]; got != "ws-1" {
+			t.Fatalf("commit workspaceId = %v, want ws-1", got)
+		}
+		if got := caller.args["name"]; got != "sales.html" {
+			t.Fatalf("commit name = %v, want original file name with extension", got)
+		}
+		var payload map[string]any
+		if err := json.Unmarshal(stdout.Bytes(), &payload); err != nil {
+			t.Fatalf("fallback result must be one JSON document: %v\n%s", err, stdout.String())
+		}
+		if payload["success"] != true || payload["fallback"] != "upload" || payload["converted"] != false {
+			t.Fatalf("fallback markers missing: %#v", payload)
+		}
+		if payload["dentry_id"] != "node-1" {
+			t.Fatalf("dentry_id = %v, want node-1", payload["dentry_id"])
+		}
+		if payload["requested_operation"] != "导入本地文件为在线文档" {
+			t.Fatalf("requested_operation = %v", payload["requested_operation"])
+		}
+		if !strings.Contains(warnings.String(), "文件上传链路") {
+			t.Fatalf("fallback must announce the upload on stderr, got %q", warnings.String())
+		}
+		if strings.Contains(warnings.String(), "deprecated") {
+			t.Fatalf("fallback must not emit the doc upload deprecation warning, got %q", warnings.String())
+		}
+	})
+
+	t.Run("json dry run stays a single json document", func(t *testing.T) {
+		caller := &scriptedToolCaller{format: "json", dry: true}
+		installScriptedCaller(t, caller)
+		var stdout bytes.Buffer
+		deps.Out.w = &stdout
+
+		cmd := htmlFallbackCommand(t, writeImportFixture(t, "html"))
+		if err := runImportCommand(cmd, nil, docImportFlowConfig()); err != nil {
+			t.Fatalf("runImportCommand() error = %v", err)
+		}
+		if caller.calls != 0 {
+			t.Fatalf("dry run must not call MCP, calls = %d", caller.calls)
+		}
+		var payload map[string]any
+		if err := json.Unmarshal(stdout.Bytes(), &payload); err != nil {
+			t.Fatalf("json dry-run output is not JSON: %v\n%s", err, stdout.String())
+		}
+		if payload["dry_run"] != true || payload["executed"] != false || payload["fallback"] != "upload" || payload["converted"] != false {
+			t.Fatalf("dry-run fallback payload = %#v", payload)
+		}
+	})
+
+	t.Run("any non-importable format is redirected, not enumerated", func(t *testing.T) {
+		for _, ext := range []string{"pdf", "zip", "png", "mp4"} {
+			caller := &scriptedToolCaller{steps: uploadSteps}
+			installScriptedCaller(t, caller)
+			SetHTTPPutFile(func(context.Context, string, map[string]string, string, int64) error { return nil })
+			t.Cleanup(func() { SetHTTPPutFile(nil) })
+
+			cmd := htmlFallbackCommand(t, writeImportFixture(t, ext))
+			if err := runImportCommand(cmd, nil, docImportFlowConfig()); err != nil {
+				t.Fatalf("runImportCommand(%s) error = %v, want upload fallback success", ext, err)
+			}
+			if caller.tool != "commit_uploaded_file" {
+				t.Fatalf("%s last tool = %q, want commit_uploaded_file", ext, caller.tool)
+			}
+		}
+	})
+
+	t.Run("uppercase htm extension via positional argument", func(t *testing.T) {
+		caller := &scriptedToolCaller{steps: uploadSteps}
+		installScriptedCaller(t, caller)
+		SetHTTPPutFile(func(context.Context, string, map[string]string, string, int64) error { return nil })
+		t.Cleanup(func() { SetHTTPPutFile(nil) })
+
+		path := writeImportFixture(t, "HTM")
+		cmd := htmlFallbackCommand(t, "")
+		if err := runImportCommand(cmd, []string{path}, docImportFlowConfig()); err != nil {
+			t.Fatalf("runImportCommand() error = %v, want upload fallback success", err)
+		}
+		if caller.tool != "commit_uploaded_file" {
+			t.Fatalf("last tool = %q, want commit_uploaded_file", caller.tool)
+		}
+	})
+
+	t.Run("hidden folder-id alias reaches the upload chain", func(t *testing.T) {
+		caller := &scriptedToolCaller{steps: uploadSteps}
+		installScriptedCaller(t, caller)
+		SetHTTPPutFile(func(context.Context, string, map[string]string, string, int64) error { return nil })
+		t.Cleanup(func() { SetHTTPPutFile(nil) })
+
+		cmd := htmlFallbackCommand(t, writeImportFixture(t, "html"))
+		if err := cmd.Flags().Set("folder-id", "folder-abc"); err != nil {
+			t.Fatal(err)
+		}
+		if err := runImportCommand(cmd, nil, docImportFlowConfig()); err != nil {
+			t.Fatalf("runImportCommand() error = %v, want upload fallback success", err)
+		}
+		if got := caller.args["folderId"]; got != "folder-abc" {
+			t.Fatalf("commit folderId = %v, want folder-abc from --folder-id alias", got)
+		}
+	})
+
+	t.Run("extensionless file is redirected with a readable label", func(t *testing.T) {
+		caller := &scriptedToolCaller{steps: uploadSteps}
+		installScriptedCaller(t, caller)
+		var warnings bytes.Buffer
+		deps.Out.errW = &warnings
+		SetHTTPPutFile(func(context.Context, string, map[string]string, string, int64) error { return nil })
+		t.Cleanup(func() { SetHTTPPutFile(nil) })
+
+		noExt := filepath.Join(t.TempDir(), "README")
+		if err := os.WriteFile(noExt, []byte("plain"), 0o600); err != nil {
+			t.Fatal(err)
+		}
+		cmd := htmlFallbackCommand(t, noExt)
+		if err := runImportCommand(cmd, nil, docImportFlowConfig()); err != nil {
+			t.Fatalf("runImportCommand() error = %v, want upload fallback success", err)
+		}
+		if !strings.Contains(warnings.String(), "无扩展名") {
+			t.Fatalf("warning must label the extensionless file, got %q", warnings.String())
+		}
+	})
+
+	t.Run("fallback enforces the shared 20MB limit", func(t *testing.T) {
+		installScriptedCaller(t, &scriptedToolCaller{})
+		big := filepath.Join(t.TempDir(), "big.html")
+		f, err := os.Create(big)
+		if err != nil {
+			t.Fatal(err)
+		}
+		if err := f.Truncate(importMaxFileSize + 1); err != nil {
+			t.Fatal(err)
+		}
+		if err := f.Close(); err != nil {
+			t.Fatal(err)
+		}
+		cmd := htmlFallbackCommand(t, big)
+		err = runImportCommand(cmd, nil, docImportFlowConfig())
+		if err == nil || !strings.Contains(err.Error(), "exceeds 20MB limit") {
+			t.Fatalf("runImportCommand() error = %v, want 20MB limit rejection", err)
+		}
+	})
+
+	t.Run("fallback enforces the shared empty-file guard", func(t *testing.T) {
+		installScriptedCaller(t, &scriptedToolCaller{})
+		empty := filepath.Join(t.TempDir(), "empty.html")
+		if err := os.WriteFile(empty, nil, 0o600); err != nil {
+			t.Fatal(err)
+		}
+		cmd := htmlFallbackCommand(t, empty)
+		err := runImportCommand(cmd, nil, docImportFlowConfig())
+		if err == nil || !strings.Contains(err.Error(), "file is empty") {
+			t.Fatalf("runImportCommand() error = %v, want empty-file rejection", err)
+		}
+	})
+
+	t.Run("importable formats keep the conversion path", func(t *testing.T) {
+		caller := &scriptedToolCaller{format: "json", dry: true}
+		installScriptedCaller(t, caller)
+		var stdout bytes.Buffer
+		deps.Out.w = &stdout
+
+		cmd := htmlFallbackCommand(t, writeImportFixture(t, "md"))
+		if err := runImportCommand(cmd, nil, docImportFlowConfig()); err != nil {
+			t.Fatalf("runImportCommand() error = %v", err)
+		}
+		var payload map[string]any
+		if err := json.Unmarshal(stdout.Bytes(), &payload); err != nil {
+			t.Fatalf("import dry-run output is not JSON: %v\n%s", err, stdout.String())
+		}
+		if payload["operation"] != "导入本地文件为在线文档" {
+			t.Fatalf("operation = %v, want conversion path", payload["operation"])
+		}
+		if _, ok := payload["fallback"]; ok {
+			t.Fatalf("importable format must not carry fallback marker: %#v", payload)
+		}
+	})
+
+	t.Run("missing file falls through to the import required-flag error", func(t *testing.T) {
+		installScriptedCaller(t, &scriptedToolCaller{})
+		cmd := htmlFallbackCommand(t, "")
+		err := runImportCommand(cmd, nil, docImportFlowConfig())
+		if err == nil || !strings.Contains(err.Error(), "--file is required") {
+			t.Fatalf("runImportCommand() error = %v, want --file required", err)
+		}
+	})
+
+	t.Run("text dry run prints the fallback plan as key values", func(t *testing.T) {
+		caller := &scriptedToolCaller{dry: true}
+		installScriptedCaller(t, caller)
+		var stdout bytes.Buffer
+		deps.Out.w = &stdout
+
+		cmd := htmlFallbackCommand(t, writeImportFixture(t, "html"))
+		if err := runImportCommand(cmd, nil, docImportFlowConfig()); err != nil {
+			t.Fatalf("runImportCommand() error = %v", err)
+		}
+		if caller.calls != 0 {
+			t.Fatalf("dry run must not call MCP, calls = %d", caller.calls)
+		}
+		if !strings.Contains(stdout.String(), "doc import 回退") || !strings.Contains(stdout.String(), "sales.html") {
+			t.Fatalf("text dry-run must print the fallback plan, got %q", stdout.String())
+		}
+	})
+
+	t.Run("upload chain errors propagate", func(t *testing.T) {
+		cases := []struct {
+			name    string
+			steps   []scriptedToolStep
+			putErr  error
+			wantErr string
+		}{
+			{name: "upload info request fails", steps: []scriptedToolStep{{err: errUploadInfo}}, wantErr: "upload info failed"},
+			{name: "upload credentials incomplete", steps: []scriptedToolStep{{text: `{"resourceUrl":""}`}}, wantErr: "incomplete upload credentials"},
+			{name: "http put fails", steps: []scriptedToolStep{{text: `{"resourceUrl":"https://upload.example.test/object","uploadKey":"key-1"}`}}, putErr: errUploadPut, wantErr: "put failed"},
+		}
+		for _, tc := range cases {
+			t.Run(tc.name, func(t *testing.T) {
+				installScriptedCaller(t, &scriptedToolCaller{steps: tc.steps})
+				SetHTTPPutFile(func(context.Context, string, map[string]string, string, int64) error { return tc.putErr })
+				t.Cleanup(func() { SetHTTPPutFile(nil) })
+
+				cmd := htmlFallbackCommand(t, writeImportFixture(t, "html"))
+				err := runImportCommand(cmd, nil, docImportFlowConfig())
+				if err == nil || !strings.Contains(err.Error(), tc.wantErr) {
+					t.Fatalf("runImportCommand() error = %v, want %q", err, tc.wantErr)
+				}
+			})
+		}
+	})
+
+	t.Run("commit responses that cannot prove success are rejected", func(t *testing.T) {
+		cases := []struct {
+			name    string
+			commit  string
+			wantErr string
+		}{
+			{name: "empty legacy ack", commit: "  ", wantErr: "响应为空"},
+			{name: "non-json text", commit: "commit-ok-plain-text", wantErr: "无法解析为 JSON"},
+			{name: "missing file identity", commit: `{"ok":true}`, wantErr: "缺少文件标识"},
+		}
+		for _, tc := range cases {
+			t.Run(tc.name, func(t *testing.T) {
+				installScriptedCaller(t, &scriptedToolCaller{format: "json", steps: []scriptedToolStep{
+					{text: `{"resourceUrl":"https://upload.example.test/object","uploadKey":"key-1"}`},
+					{text: tc.commit},
+				}})
+				SetHTTPPutFile(func(context.Context, string, map[string]string, string, int64) error { return nil })
+				t.Cleanup(func() { SetHTTPPutFile(nil) })
+
+				cmd := htmlFallbackCommand(t, writeImportFixture(t, "html"))
+				err := runImportCommand(cmd, nil, docImportFlowConfig())
+				if err == nil || !strings.Contains(err.Error(), tc.wantErr) {
+					t.Fatalf("runImportCommand() error = %v, want %q", err, tc.wantErr)
+				}
+			})
+		}
+	})
+
+	t.Run("nested result envelope yields the dentry id", func(t *testing.T) {
+		caller := &scriptedToolCaller{format: "json", steps: []scriptedToolStep{
+			{text: `{"resourceUrl":"https://upload.example.test/object","uploadKey":"key-1"}`},
+			{text: `{"result":{"dentryUuid":"nested-node-9"}}`},
+		}}
+		installScriptedCaller(t, caller)
+		var stdout bytes.Buffer
+		deps.Out.w = &stdout
+		SetHTTPPutFile(func(context.Context, string, map[string]string, string, int64) error { return nil })
+		t.Cleanup(func() { SetHTTPPutFile(nil) })
+
+		cmd := htmlFallbackCommand(t, writeImportFixture(t, "html"))
+		if err := runImportCommand(cmd, nil, docImportFlowConfig()); err != nil {
+			t.Fatalf("runImportCommand() error = %v", err)
+		}
+		var payload map[string]any
+		if err := json.Unmarshal(stdout.Bytes(), &payload); err != nil {
+			t.Fatalf("fallback result must stay JSON: %v\n%s", err, stdout.String())
+		}
+		if payload["dentry_id"] != "nested-node-9" {
+			t.Fatalf("dentry_id = %v, want nested-node-9", payload["dentry_id"])
+		}
+	})
+
+	t.Run("sheet import keeps rejecting html without fallback", func(t *testing.T) {
+		installScriptedCaller(t, &scriptedToolCaller{})
+		// 无导入目标时也必须先报 unsupported（基线校验顺序：扩展名先于目标）
+		cmd := htmlFallbackCommand(t, writeImportFixture(t, "html"))
+		err := runImportCommand(cmd, nil, sheetImportFlowConfig())
+		if err == nil || !strings.Contains(err.Error(), "unsupported file format") {
+			t.Fatalf("sheet import (no target) error = %v, want unsupported file format", err)
+		}
+
+		cmd = htmlFallbackCommand(t, writeImportFixture(t, "html"))
+		if err := cmd.Flags().Set("workspace", "ws-1"); err != nil {
+			t.Fatal(err)
+		}
+		err = runImportCommand(cmd, nil, sheetImportFlowConfig())
+		if err == nil || !strings.Contains(err.Error(), "unsupported file format") {
+			t.Fatalf("sheet import (with target) error = %v, want unsupported file format", err)
+		}
+	})
+}

internal/helpers/leaf_policy_test.go

@@ -143,39 +143,42 @@ func leafSpecCompositeLits(t *testing.T) []leafSpecLit {
 	}
 	dir := filepath.Dir(file)
 	fset := token.NewFileSet()
-	pkgs, err := parser.ParseDir(fset, dir, func(info os.FileInfo) bool {
-		name := info.Name()
-		return strings.HasSuffix(name, ".go") && !strings.HasSuffix(name, "_test.go")
-	}, 0)
+	entries, err := os.ReadDir(dir)
 	if err != nil {
-		t.Fatalf("parse helpers: %v", err)
+		t.Fatalf("read helpers dir: %v", err)
 	}
 	var out []leafSpecLit
-	for _, pkg := range pkgs {
-		for _, f := range pkg.Files {
-			ast.Inspect(f, func(n ast.Node) bool {
-				call, ok := n.(*ast.CallExpr)
-				if !ok {
-					return true
-				}
-				fun, ok := call.Fun.(*ast.Ident)
-				if !ok || fun.Name != "NewLeafCommand" || len(call.Args) != 1 {
-					return true
-				}
-				lit, ok := call.Args[0].(*ast.CompositeLit)
-				if !ok {
-					return true
-				}
-				if typ, ok := lit.Type.(*ast.Ident); !ok || typ.Name != "LeafSpec" {
-					return true
-				}
-				out = append(out, leafSpecLit{
-					pos: fset.Position(lit.Pos()).String(),
-					lit: lit,
-				})
-				return true
-			})
+	for _, entry := range entries {
+		name := entry.Name()
+		if entry.IsDir() || !strings.HasSuffix(name, ".go") || strings.HasSuffix(name, "_test.go") {
+			continue
 		}
+		parsed, err := parser.ParseFile(fset, filepath.Join(dir, name), nil, 0)
+		if err != nil {
+			t.Fatalf("parse helpers file %s: %v", name, err)
+		}
+		ast.Inspect(parsed, func(n ast.Node) bool {
+			call, ok := n.(*ast.CallExpr)
+			if !ok {
+				return true
+			}
+			fun, ok := call.Fun.(*ast.Ident)
+			if !ok || fun.Name != "NewLeafCommand" || len(call.Args) != 1 {
+				return true
+			}
+			lit, ok := call.Args[0].(*ast.CompositeLit)
+			if !ok {
+				return true
+			}
+			if typ, ok := lit.Type.(*ast.Ident); !ok || typ.Name != "LeafSpec" {
+				return true
+			}
+			out = append(out, leafSpecLit{
+				pos: fset.Position(lit.Pos()).String(),
+				lit: lit,
+			})
+			return true
+		})
 	}
 	if len(out) == 0 {
 		t.Fatal("no NewLeafCommand(LeafSpec{...}) literals found")

internal/helpers/sheet.go

@@ -44,6 +44,7 @@ func newSheetCommand() *cobra.Command {
 
 命令结构:
   dws sheet create                              创建钉钉表格文档
+  dws sheet create-with-data                     创建表格文档并写入初始数据（可选样式）
   dws sheet list                                获取全部工作表列表
   dws sheet info                                获取指定工作表详情
   dws sheet new                                 新建工作表
@@ -110,6 +111,7 @@ func newSheetCommand() *cobra.Command {
   dws sheet chart update                         更新浮动图表
   dws sheet chart delete                         删除浮动图表
   dws sheet export                              导出表格为 xlsx（异步任务一站式：提交→轮询→可选下载）
+  dws sheet export-csv                          导出单个工作表为纯 CSV（同步，可落盘）
   dws sheet import                              导入 xlsx/xls 为在线电子表格
   dws sheet template list                       获取表格模板列表
   dws sheet template search                     搜索表格模板
@@ -159,6 +161,10 @@ func newSheetCommand() *cobra.Command {
 			},
 		},
 	})
+	// 建表带初始数据、csv 同步导出都是与既有叶子不同的接口种类（一条 composite
+	// 编排、一条 get_range_as_csv 直连），各自独立成叶子，不挂到 create / export 上。
+	createWithDataCmd := newSheetCreateWithDataCmd()
+	exportCsvCmd := newSheetExportCsvCmd()
 	importCmd := newSheetImportCmd()
 	templateCmd := newSheetTemplateCmd()
 	tableCmds := newTableCmds()
@@ -238,7 +244,7 @@ func newSheetCommand() *cobra.Command {
 	standaloneCmds = append(standaloneCmds, mediaCmds...)
 	standaloneCmds = append(standaloneCmds, floatImageCmds...)
 	standaloneCmds = append(standaloneCmds, tableCmds...)
-	standaloneCmds = append(standaloneCmds, exportCmd, importCmd, batchUpdateCmd)
+	standaloneCmds = append(standaloneCmds, exportCmd, exportCsvCmd, importCmd, batchUpdateCmd, createWithDataCmd)
 
 	// Register cross-product aliases
 	for _, cmd := range standaloneCmds {

internal/helpers/sheet_batch.go

@@ -517,12 +517,12 @@ func newRangeBatchClearCmd() *cobra.Command {
 			}
 			operations := make([]any, 0, len(ranges))
 			for i, rng := range ranges {
-				idx := strings.Index(rng, "!")
-				if idx <= 0 || idx == len(rng)-1 {
-					return fmt.Errorf("--ranges[%d] (%q) 必须包含工作表前缀，格式为 \"SheetName!A1:B3\"", i, rng)
+				// 与 batch-set-style 共用同一个拆分器：此前这里自己拆，且只按原始串里
+				// ! 的位置判断，" !A1:B2" 会拆出空工作表名并带着 sheetId:"" 下发。
+				sheetName, rangeAddr, err := splitSheetPrefixedRange(rng, i)
+				if err != nil {
+					return err
 				}
-				sheetName := strings.TrimSpace(rng[:idx])
-				rangeAddr := strings.TrimSpace(rng[idx+1:])
 				operations = append(operations, map[string]any{
 					"toolName": "clear_range",
 					"input": map[string]any{

internal/helpers/sheet_data.go

@@ -129,6 +129,7 @@ sheetId 支持传入工作表 ID 或工作表名称，可通过 sheet list 获
   --match-case           区分大小写（默认 false）
   --match-entire-cell    完整单元格匹配（默认 false）
   --use-regexp           使用正则表达式匹配（默认 false）
+  --match-formula        在公式文本中查找替换（默认 false，替换公式源码而非显示值）
   --include-hidden       包含隐藏行/列（默认 false）
   --range                限定替换范围，A1 表示法（不传时在整个工作表中替换）
 
@@ -166,6 +167,8 @@ sheetId 支持传入工作表 ID 或工作表名称，可通过 sheet list 获
 			toolArgs["matchEntireCell"] = matchEntireCell
 			useRegExp, _ := cmd.Flags().GetBool("use-regexp")
 			toolArgs["useRegExp"] = useRegExp
+			matchFormula, _ := cmd.Flags().GetBool("match-formula")
+			toolArgs["matchFormulaText"] = matchFormula
 			includeHidden, _ := cmd.Flags().GetBool("include-hidden")
 			toolArgs["includeHidden"] = includeHidden
 			return callMCPTool("replace_all", toolArgs)
@@ -198,6 +201,7 @@ sheetId 支持传入工作表 ID 或工作表名称，可通过 sheet list 获
 			},
 			Parameters: []contract.ParamDecl{
 				{Name: "find", Property: "text"},
+				{Name: "match-formula", Property: "matchFormulaText"},
 				{Name: "node", Property: "nodeId"},
 				{Name: "replacement", Property: "replaceText"},
 				{Name: "use-regexp", Property: "useRegExp"},
@@ -212,6 +216,7 @@ sheetId 支持传入工作表 ID 或工作表名称，可通过 sheet list 获
 	replaceCmd.Flags().Bool("match-case", false, "区分大小写 (默认 false)")
 	replaceCmd.Flags().Bool("match-entire-cell", false, "完整单元格匹配")
 	replaceCmd.Flags().Bool("use-regexp", false, "启用正则表达式匹配")
+	replaceCmd.Flags().Bool("match-formula", false, "在公式文本中查找替换（默认 false）")
 	replaceCmd.Flags().Bool("include-hidden", false, "包含隐藏行/列")
 
 	appendCmd := &cobra.Command{

internal/helpers/sheet_dimension.go

@@ -3,12 +3,21 @@ package helpers
 import (
 	"encoding/json"
 	"fmt"
+	"strconv"
 	"strings"
 
 	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/corecmd/contract"
 	"github.com/spf13/cobra"
 )
 
+// sizeTypeEnumHint 返回该维度支持的尺寸模式列表（行高含 auto，列宽不含，与飞书一致）。
+func sizeTypeEnumHint(dimension string) string {
+	if dimension == "ROWS" {
+		return "pixel / standard / auto"
+	}
+	return "pixel / standard"
+}
+
 // newDimensionCmds creates dimension-related commands: insert/delete/update/move/add-dimension,
 // merge-cells, unmerge-cells, and dropdown commands (set/get/delete-dropdown).
 func newDimensionCmds() []*cobra.Command {
@@ -43,8 +52,11 @@ func newDimensionCmds() []*cobra.Command {
 				return fmt.Errorf("--dimension 必须为 ROWS 或 COLUMNS，当前值: %s", dimension)
 			}
 			lengthStr := mustGetFlag(cmd, "length")
-			var length int
-			if _, err := fmt.Sscanf(lengthStr, "%d", &length); err != nil || length < 1 {
+			// strconv.Atoi 要求整个字符串都是合法整数。fmt.Sscanf("%d") 只消费前缀
+			// 数字，会把 "3x" / "3foo" 静默当成 3，从而对错误的行列数执行插入/删除/
+			// 调整——删除方向不可回滚。
+			length, lengthErr := strconv.Atoi(lengthStr)
+			if lengthErr != nil || length < 1 {
 				return fmt.Errorf("--length 必须为正整数（>= 1），当前值: %s", lengthStr)
 			}
 			if length > 5000 {
@@ -428,8 +440,11 @@ sheetId 支持传入工作表 ID 或工作表名称，可通过 sheet list 获
 				return fmt.Errorf("--dimension 必须为 ROWS 或 COLUMNS，当前值: %s", dimension)
 			}
 			lengthStr := mustGetFlag(cmd, "length")
-			var length int
-			if _, err := fmt.Sscanf(lengthStr, "%d", &length); err != nil || length < 1 {
+			// strconv.Atoi 要求整个字符串都是合法整数。fmt.Sscanf("%d") 只消费前缀
+			// 数字，会把 "3x" / "3foo" 静默当成 3，从而对错误的行列数执行插入/删除/
+			// 调整——删除方向不可回滚。
+			length, lengthErr := strconv.Atoi(lengthStr)
+			if lengthErr != nil || length < 1 {
 				return fmt.Errorf("--length 必须为正整数（>= 1），当前值: %s", lengthStr)
 			}
 			if length > 5000 {
@@ -490,9 +505,15 @@ sheetId 支持传入工作表 ID 或工作表名称，可通过 sheet list 获
   --dimension COLUMNS 时，--start-index 为列字母，如 "A" 表示从 A 列开始、"AB" 表示从 AB 列开始
 
 支持在 --start-index 中携带工作表前缀（如 "Sheet1!3" / "Sheet1!A"），此时将忽略 --sheet-id。
---hidden 与 --pixel-size 至少必须提供一个。当同时提供时，将先应用尺寸再应用显隐，任一失败整体失败。
+--hidden 与 --pixel-size 至少必须提供一个（或用 --size-type standard/auto 让服务端决定尺寸）。
+当同时提供时，将先应用尺寸再应用显隐，任一失败整体失败。
 --pixel-size 单位为像素，dimension=ROWS 时表示行高、dimension=COLUMNS 时表示列宽。
 
+尺寸模式（--size-type，对齐飞书）:
+  pixel     默认，按 --pixel-size 指定的像素值设置
+  standard  恢复默认行高/列宽，无需 --pixel-size
+  auto      按内容自适应行高，仅 ROWS 支持（列宽不提供自适应），无需 --pixel-size
+
 常见场景：隐藏/显示指定连续行或列、批量调整行高/列宽、在同一次调用中同时修改尺寸与显隐。`,
 		Example: `  # 隐藏第 3~4 行
   dws sheet update-dimension --node NODE_ID --sheet-id SHEET_ID --dimension ROWS --start-index "3" --length 2 --hidden
@@ -517,8 +538,11 @@ sheetId 支持传入工作表 ID 或工作表名称，可通过 sheet list 获
 				return fmt.Errorf("--dimension 必须为 ROWS 或 COLUMNS，当前值: %s", dimension)
 			}
 			lengthStr := mustGetFlag(cmd, "length")
-			var length int
-			if _, err := fmt.Sscanf(lengthStr, "%d", &length); err != nil || length < 1 {
+			// strconv.Atoi 要求整个字符串都是合法整数。fmt.Sscanf("%d") 只消费前缀
+			// 数字，会把 "3x" / "3foo" 静默当成 3，从而对错误的行列数执行插入/删除/
+			// 调整——删除方向不可回滚。
+			length, lengthErr := strconv.Atoi(lengthStr)
+			if lengthErr != nil || length < 1 {
 				return fmt.Errorf("--length 必须为正整数（>= 1），当前值: %s", lengthStr)
 			}
 			if length > 5000 {
@@ -526,8 +550,30 @@ sheetId 支持传入工作表 ID 或工作表名称，可通过 sheet list 获
 			}
 			hiddenChanged := cmd.Flags().Changed("hidden")
 			pixelSizeChanged := cmd.Flags().Changed("pixel-size")
-			if !hiddenChanged && !pixelSizeChanged {
-				return fmt.Errorf("--hidden 与 --pixel-size 至少必须提供一个")
+			sizeType, _ := cmd.Flags().GetString("size-type")
+			sizeType = strings.ToLower(strings.TrimSpace(sizeType))
+			// 尺寸模式枚举按维度区分（与飞书一致）：行高有 auto，列宽只有 pixel / standard
+			switch {
+			case sizeType == "" || sizeType == "pixel" || sizeType == "standard":
+			case sizeType == "auto" && dimension == "ROWS":
+			case sizeType == "auto":
+				return fmt.Errorf("--size-type 对 COLUMNS 仅支持 pixel / standard（列宽不提供自适应）")
+			default:
+				return fmt.Errorf("--size-type 必须为 %s，当前值: %s", sizeTypeEnumHint(dimension), sizeType)
+			}
+			// standard/auto 由服务端决定尺寸，无需 --pixel-size
+			sizeTypeDrivesSize := sizeType == "standard" || sizeType == "auto"
+			if !hiddenChanged && !pixelSizeChanged && !sizeTypeDrivesSize {
+				return fmt.Errorf("--hidden 与 --pixel-size 至少必须提供一个（或用 --size-type standard/auto 让服务端决定尺寸）")
+			}
+			// standard/auto 说"尺寸交给服务端"，--pixel-size 又指定固定像素，
+			// 两者语义直接冲突；同时下发会得到依赖服务端实现的结果，故先拒。
+			if sizeTypeDrivesSize && pixelSizeChanged {
+				return fmt.Errorf("--size-type %s 表示尺寸由服务端决定，不能同时指定 --pixel-size；要指定固定像素请用 --size-type pixel（或省略 --size-type）", sizeType)
+			}
+			// 反向：显式声明 pixel 模式却不给像素值，同样是不完整的请求。
+			if sizeType == "pixel" && !pixelSizeChanged {
+				return fmt.Errorf("--size-type pixel 必须配合 --pixel-size 指定像素值；若本次只改显隐，请省略 --size-type")
 			}
 			toolArgs := map[string]any{
 				"nodeId":     mustGetFlag(cmd, "node"),
@@ -547,6 +593,9 @@ sheetId 支持传入工作表 ID 或工作表名称，可通过 sheet list 获
 				}
 				toolArgs["pixelSize"] = pixelSize
 			}
+			if sizeType != "" {
+				toolArgs["sizeType"] = sizeType
+			}
 			return callMCPTool("update_dimension", toolArgs)
 		},
 	}
@@ -587,6 +636,7 @@ sheetId 支持传入工作表 ID 或工作表名称，可通过 sheet list 获
 	updateDimensionCmd.Flags().String("length", "", "更新数量，正整数 (必填)，最大 5000")
 	updateDimensionCmd.Flags().Bool("hidden", false, "是否隐藏 (true=隐藏, false=显示)")
 	updateDimensionCmd.Flags().Int("pixel-size", 0, "行高或列宽（像素），ROWS 时为行高，COLUMNS 时为列宽")
+	updateDimensionCmd.Flags().String("size-type", "", "尺寸模式（对齐飞书）: pixel(默认,用 --pixel-size) / standard(恢复默认行高列宽) / auto(按内容自适应行高，仅 ROWS；列宽无此选项)")
 
 	groupDimensionCmd := &cobra.Command{
 		Use:   "group-dimension",

internal/helpers/sheet_dimension_coverage_more_test.go

@@ -2,6 +2,7 @@ package helpers
 
 import (
 	"io"
+	"strings"
 	"testing"
 
 	"github.com/spf13/cobra"
@@ -74,6 +75,84 @@ func TestCrossPlatformCoverageDimensionValidationRemainingCoverage(t *testing.T)
 	}
 }
 
+// --length 必须整个值都是合法正整数。改用 strconv.Atoi 之前走的是
+// fmt.Sscanf("%d")，只消费前缀数字，"2x" 被静默当成 2 并对错误的行列数执行操作
+// ——delete 方向不可回滚。这是对既有命令的用户可见行为变更（已记入 CHANGELOG），
+// 用测试钉住，避免哪天改回宽松解析没人发现。
+func TestDimensionLengthRejectsTrailingCharacters(t *testing.T) {
+	installScriptedCaller(t, &scriptedToolCaller{dry: true})
+	common := []string{"--node", "node", "--sheet-id", "sheet", "--dimension", "ROWS"}
+
+	for _, cmdName := range []string{"insert-dimension", "delete-dimension", "update-dimension"} {
+		locator := []string{"--position", "1"}
+		if cmdName == "update-dimension" {
+			locator = []string{"--start-index", "1", "--hidden"}
+		}
+		for _, bad := range []string{"2x", "3foo", "1 2", "0x10", "abc", ""} {
+			args := append(append(append([]string{}, common...), locator...), "--length", bad)
+			err := executeDimensionCoverage(t, cmdName, args...)
+			if err == nil {
+				t.Errorf("%s --length %q 未报错，畸形长度被静默接受", cmdName, bad)
+				continue
+			}
+			if !strings.Contains(err.Error(), "--length") {
+				t.Errorf("%s --length %q: err = %v, want 指明 --length", cmdName, bad, err)
+			}
+		}
+	}
+}
+
+// --size-type 的枚举按维度区分：行高有 auto，列宽只有 pixel / standard（对齐飞书）。
+func TestUpdateDimensionSizeTypeEnumIsDimensionScoped(t *testing.T) {
+	installScriptedCaller(t, &scriptedToolCaller{dry: true})
+	common := []string{"--node", "node", "--sheet-id", "sheet", "--start-index", "1", "--length", "2"}
+
+	for _, tc := range []struct {
+		name      string
+		dimension string
+		sizeType  string
+		wantErr   string
+	}{
+		// standard / auto 由服务端决定尺寸，不再要求 --hidden 或 --pixel-size
+		{"rows-standard", "ROWS", "standard", ""},
+		{"rows-auto", "ROWS", "auto", ""},
+		{"columns-standard", "COLUMNS", "standard", ""},
+		{"columns-auto-rejected", "COLUMNS", "auto", "仅支持 pixel / standard"},
+		// 非法枚举的提示文案必须按维度给出不同的合法值列表
+		{"rows-bogus-hint", "ROWS", "bogus", "必须为 pixel / standard / auto"},
+		{"columns-bogus-hint", "COLUMNS", "bogus", "必须为 pixel / standard"},
+	} {
+		t.Run(tc.name, func(t *testing.T) {
+			args := append(append([]string{}, common...), "--dimension", tc.dimension, "--size-type", tc.sizeType)
+			err := executeDimensionCoverage(t, "update-dimension", args...)
+			if tc.wantErr == "" {
+				if err != nil {
+					t.Fatalf("err = %v, want nil", err)
+				}
+				return
+			}
+			if err == nil || !strings.Contains(err.Error(), tc.wantErr) {
+				t.Fatalf("err = %v, want contains %q", err, tc.wantErr)
+			}
+		})
+	}
+
+	// pixel 仍需 --hidden 或 --pixel-size 之一
+	args := append(append([]string{}, common...), "--dimension", "ROWS", "--size-type", "pixel")
+	if err := executeDimensionCoverage(t, "update-dimension", args...); err == nil {
+		t.Fatal("size-type=pixel 且未给 --pixel-size/--hidden 时应报错")
+	}
+}
+
+func TestSizeTypeEnumHint(t *testing.T) {
+	if got := sizeTypeEnumHint("ROWS"); got != "pixel / standard / auto" {
+		t.Fatalf("ROWS hint = %q", got)
+	}
+	if got := sizeTypeEnumHint("COLUMNS"); got != "pixel / standard" {
+		t.Fatalf("COLUMNS hint = %q", got)
+	}
+}
+
 func TestCrossPlatformCoverageDropdownValidationRemainingCoverage(t *testing.T) {
 	installScriptedCaller(t, &scriptedToolCaller{dry: true})
 	base := []string{"--node", "node", "--sheet-id", "sheet", "--range", "A1"}
@@ -86,3 +165,80 @@ func TestCrossPlatformCoverageDropdownValidationRemainingCoverage(t *testing.T)
 		t.Fatalf("multi-select dropdown: %v", err)
 	}
 }
+
+// --size-type standard/auto 与 --pixel-size 语义冲突（一个说交给服务端、一个
+// 指定固定像素），必须在发请求前拒；反向 pixel 模式也必须带像素值。
+func TestUpdateDimensionRejectsSizeTypeAndPixelSizeConflict(t *testing.T) {
+	for _, tc := range []struct {
+		name  string
+		flags [][2]string
+		want  string
+	}{
+		{
+			"standard-with-pixel-size",
+			[][2]string{{"dimension", "ROWS"}, {"size-type", "standard"}, {"pixel-size", "40"}},
+			"不能同时指定 --pixel-size",
+		},
+		{
+			"auto-with-pixel-size",
+			[][2]string{{"dimension", "ROWS"}, {"size-type", "auto"}, {"pixel-size", "40"}},
+			"不能同时指定 --pixel-size",
+		},
+		{
+			"explicit-pixel-without-size",
+			[][2]string{{"dimension", "ROWS"}, {"size-type", "pixel"}, {"hidden", "true"}},
+			"--size-type pixel 必须配合 --pixel-size",
+		},
+	} {
+		t.Run(tc.name, func(t *testing.T) {
+			caller := &scriptedToolCaller{}
+			flags := append([][2]string{{"node", "N"}, {"sheet-id", "S"}, {"start-index", "1"}, {"length", "2"}}, tc.flags...)
+			err := runUpdateDimensionForTest(t, caller, flags)
+			if err == nil || !strings.Contains(err.Error(), tc.want) {
+				t.Fatalf("err = %v, want contains %q", err, tc.want)
+			}
+			if caller.calls != 0 {
+				t.Fatalf("calls = %d, want 0（冲突必须在发请求前拒）", caller.calls)
+			}
+		})
+	}
+
+	// 合法组合仍然放行
+	for _, tc := range []struct {
+		name  string
+		flags [][2]string
+	}{
+		{"standard-alone", [][2]string{{"dimension", "ROWS"}, {"size-type", "standard"}}},
+		{"auto-alone", [][2]string{{"dimension", "ROWS"}, {"size-type", "auto"}}},
+		{"pixel-with-size", [][2]string{{"dimension", "ROWS"}, {"size-type", "pixel"}, {"pixel-size", "40"}}},
+		{"pixel-size-without-size-type", [][2]string{{"dimension", "COLUMNS"}, {"pixel-size", "120"}}},
+		{"hidden-only", [][2]string{{"dimension", "ROWS"}, {"hidden", "true"}}},
+	} {
+		t.Run("ok/"+tc.name, func(t *testing.T) {
+			caller := &scriptedToolCaller{steps: []scriptedToolStep{{text: `{"success":true}`}}}
+			flags := append([][2]string{{"node", "N"}, {"sheet-id", "S"}, {"start-index", "1"}, {"length", "2"}}, tc.flags...)
+			if err := runUpdateDimensionForTest(t, caller, flags); err != nil {
+				t.Fatalf("合法组合被拒: %v", err)
+			}
+		})
+	}
+}
+
+func runUpdateDimensionForTest(t *testing.T, caller *scriptedToolCaller, flags [][2]string) error {
+	t.Helper()
+	installScriptedCaller(t, caller)
+	installSheetProductArgs(t)
+	for _, c := range newDimensionCmds() {
+		if c.Name() != "update-dimension" {
+			continue
+		}
+		for _, kv := range flags {
+			if err := c.Flags().Set(kv[0], kv[1]); err != nil {
+				t.Fatalf("set --%s: %v", kv[0], err)
+			}
+		}
+		return c.RunE(c, nil)
+	}
+	t.Fatal("update-dimension not found")
+	return nil
+}

internal/helpers/sheet_export_csv.go

@@ -0,0 +1,241 @@
+package helpers
+
+import (
+	"context"
+	"encoding/json"
+	"fmt"
+	"os"
+	"path/filepath"
+	"strings"
+
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/corecmd/contract"
+	"github.com/spf13/cobra"
+)
+
+// ============================================================================
+// sheet export-csv：把单个工作表同步导出为纯 RFC4180 CSV
+//
+// 独立成一条命令而不是给 sheet export 加 --export-format csv：sheet export 的叶子
+// 契约是 interface_mode=mcp + interface_ref=submit_export_job（xlsx 异步任务）。
+// csv 是互斥分支，走的是 get_range_as_csv，submit_export_job 根本不执行；挂在同一
+// 条叶子上会让 Agent 的接口审计、参数映射和后端能力判断在 csv 场景下全都拿到错误
+// 信息。本叶子如实声明 interface_ref=get_range_as_csv，sheet export 则保持原样。
+//
+// 与 sheet csv-get 的分工：csv-get 面向 Agent 阅读，输出带 [row=N] 行号前缀并按
+// -f 渲染，只回 stdout；本命令面向落盘，输出纯 CSV、可写文件，且截断时 fail-closed。
+// ============================================================================
+
+func newSheetExportCsvCmd() *cobra.Command {
+	cmd := &cobra.Command{
+		Use:   "export-csv",
+		Short: "导出单个工作表为纯 CSV（同步）",
+		Long: `将钉钉在线电子表格的单个工作表同步导出为纯 RFC4180 CSV。
+
+整篇表格导出为 xlsx 请用 dws sheet export（异步任务：提交→轮询→可选下载）。
+
+参数说明:
+  --node                  表格文档 ID 或链接 URL，系统自动识别（必填）
+  --sheet-id              要导出的工作表 ID 或名称（不传则第一个工作表）
+  --range                 导出范围，A1 表示法（不传则整表；大表可用此分块导出）
+  --value-render-option   取值模式: formatted_value(默认) / raw_value / formula
+  --output                本地保存路径（可选）。可为文件路径或目录（目录时存为
+                          sheet-export.csv）；不传则把 CSV 打到 stdout
+  --allow-truncated       允许数据被截断时仍然导出
+
+截断行为:
+  表格超出单次读取上限时默认直接报错并且不写文件，避免不完整数据被当成完整导出、
+  或覆盖掉已有的完整文件。要接受不完整结果须显式加 --allow-truncated。
+
+落盘保证:
+  写文件走同目录临时文件 + rename 原子替换，写入中途失败（磁盘满、配额、I/O 错误）
+  不会破坏已存在的目标文件。父目录不存在时报错，不会替你创建。
+
+权限要求:
+  当前用户对目标表格具备可查看权限。`,
+		Example: `  # 导出第一个工作表到 stdout（可管道处理）
+  dws sheet export-csv --node NODE_ID
+
+  # 导出指定工作表为本地 CSV 文件
+  dws sheet export-csv --node NODE_ID --sheet-id SHEET_ID --output ./data.csv
+
+  # 只导出某个范围，取原始值
+  dws sheet export-csv --node NODE_ID --range A1:Z1000 --value-render-option raw_value`,
+		RunE: runSheetExportCsv,
+	}
+	DeclareLeafMetadata(cmd, LeafSpec{
+		Safety: contract.SafetySpec{
+			Effect: "read", Risk: "low",
+			Confirmation: "not_required", Idempotency: "idempotent",
+		},
+		Contract: LeafContract{
+			Identity: contract.ToolIdentitySpec{
+				ProductID:      "sheet",
+				Name:           "export_csv",
+				CanonicalPath:  "sheet.export_csv",
+				CLIPath:        "sheet export-csv",
+				PrimaryCLIPath: "sheet export-csv",
+			},
+			Description: "同步导出单个工作表为纯 RFC4180 CSV（可落盘，截断即报错）。",
+			DryRun:      &contract.DryRunSpec{PreviewKind: "plan", RemoteReads: false},
+			Interface: &contract.InterfaceSpec{
+				Mode:         "mcp",
+				Availability: "available",
+				Ref:          &contract.InterfaceRefSpec{ProductID: "sheet", RPCName: "get_range_as_csv"},
+			},
+			Selection: contract.SelectionSpec{
+				AgentSummary: "同步导出单个工作表为纯 RFC4180 CSV（可落盘，截断即报错）。",
+				UseWhen:      []string{"需要单个工作表的纯 CSV（写本地文件或管道处理）时"},
+				AvoidWhen:    []string{"要整篇表格的 xlsx 用 sheet export；只是让 Agent 读内容用 sheet csv-get（带 [row=N] 行号）"},
+				Examples: []string{
+					"dws sheet export-csv --node <NODE_ID> --sheet-id <SHEET_ID> --output ./data.csv",
+					"dws sheet export-csv --node <NODE_ID> --range A1:Z1000",
+				},
+			},
+			Parameters: []contract.ParamDecl{
+				{Name: "node", Property: "nodeId"},
+				{Name: "range", Property: "range"},
+				{Name: "sheet-id", Property: "sheetId"},
+				{Name: "value-render-option", Property: "valueRenderOption"},
+			},
+		},
+	})
+	cmd.Flags().String("node", "", "表格文档 ID 或 URL (必填)")
+	cmd.Flags().String("sheet-id", "", "工作表 ID 或名称（不传则第一个工作表）")
+	cmd.Flags().String("range", "", "导出范围，A1 表示法（不传则整表；大表可用此分块导出）")
+	cmd.Flags().String("value-render-option", "", "取值模式: formatted_value(默认) / raw_value / formula")
+	cmd.Flags().String("output", "", "本地保存路径（可选，支持文件路径或目录）；不传则输出到 stdout")
+	cmd.Flags().Bool("allow-truncated", false, "允许数据被截断时仍然导出。默认截断即报错并不写文件，避免不完整数据被当成完整导出")
+	return cmd
+}
+
+// valueRenderOptionEnum 是 --value-render-option 的合法取值。
+var valueRenderOptionEnum = map[string]bool{
+	"formatted_value": true, "raw_value": true, "formula": true,
+}
+
+// runSheetExportCsv 导出单个工作表为纯 CSV（同步，复用 get_range_as_csv，annotateRowNumbers=false）。
+func runSheetExportCsv(cmd *cobra.Command, _ []string) error {
+	nodeID := mustGetFlag(cmd, "node")
+	if nodeID == "" {
+		return fmt.Errorf("flag --node is required")
+	}
+	sheetID, _ := cmd.Flags().GetString("sheet-id")
+	rangeAddr, _ := cmd.Flags().GetString("range")
+	valueRenderOption, _ := cmd.Flags().GetString("value-render-option")
+	valueRenderOption = strings.ToLower(strings.TrimSpace(valueRenderOption))
+	if valueRenderOption != "" && !valueRenderOptionEnum[valueRenderOption] {
+		return fmt.Errorf("--value-render-option 必须为 formatted_value / raw_value / formula，当前值: %s", valueRenderOption)
+	}
+	outputPath, _ := cmd.Flags().GetString("output")
+	allowTruncated, _ := cmd.Flags().GetBool("allow-truncated")
+
+	if deps.Caller.DryRun() {
+		deps.Out.PrintKeyValue("操作", "导出工作表为 CSV")
+		deps.Out.PrintKeyValue("节点", nodeID)
+		if sheetID != "" {
+			deps.Out.PrintKeyValue("工作表", sheetID)
+		}
+		if outputPath != "" {
+			deps.Out.PrintKeyValue("输出", outputPath)
+		}
+		return nil
+	}
+
+	ctx := context.Background()
+	toolArgs := map[string]any{
+		"nodeId":             nodeID,
+		"annotateRowNumbers": false,
+	}
+	if sheetID != "" {
+		toolArgs["sheetId"] = sheetID
+	}
+	if rangeAddr != "" {
+		toolArgs["range"] = rangeAddr
+	}
+	if valueRenderOption != "" {
+		toolArgs["valueRenderOption"] = valueRenderOption
+	}
+
+	// CSV 正文走 stdout，进度/警告一律不能污染它。
+	text, err := callMCPToolReturnText(ctx, "get_range_as_csv", toolArgs)
+	if err != nil {
+		return fmt.Errorf("读取 CSV 失败: %w", err)
+	}
+
+	csvContent, hasMore, err := parseGetRangeAsCsvResult(text)
+	if err != nil {
+		return err
+	}
+
+	// 截断必须 fail-closed：只打 stderr 警告然后照常落盘 + 报"导出完成" + 退出码 0，
+	// 会让自动化调用方（和没留意 stderr 的人）把不完整文件当成完整导出，且若目标
+	// 文件已存在还会被截断数据覆盖。默认在写文件/输出之前就失败，要接受不完整结果
+	// 必须显式加 --allow-truncated。
+	if hasMore && !allowTruncated {
+		return fmt.Errorf("表格数据超出单次读取上限，CSV 会被截断，已中止导出（未写入 %s）；"+
+			"请用 --range 分块导出（如 --range A1:Z1000、A1001:Z2000 ...）、改用 dws sheet export 导出完整表格的 xlsx，"+
+			"或确认可接受不完整数据后加 --allow-truncated",
+			firstNonEmpty(outputPath, "stdout"))
+	}
+	if hasMore {
+		deps.Out.PrintWarning("表格数据超出单次读取上限，CSV 已被截断（--allow-truncated 已显式放行）。" +
+			"请用 --range 分块导出（如 --range A1:Z1000、A1001:Z2000 ...），或改用 dws sheet export 导出完整表格的 xlsx。")
+	}
+
+	if outputPath == "" {
+		deps.Out.PrintRaw(csvContent)
+		return nil
+	}
+	if fi, statErr := os.Stat(outputPath); statErr == nil && fi.IsDir() {
+		outputPath = filepath.Join(outputPath, "sheet-export.csv")
+	}
+	// 必须原子替换：os.WriteFile 会先把已存在的 CSV 截断，写入中途失败（磁盘满、
+	// 配额、I/O 错误）就把用户的原文件毁掉了。AtomicWrite 写同目录临时文件再 rename，
+	// 失败时原文件保持不变。
+	// AtomicWrite 会 MkdirAll 父目录，先探一次，保持与 sheet export 一致的「父目录
+	// 不存在即报错」语义，避免把拼错的路径悄悄建成目录。
+	if _, statErr := os.Stat(filepath.Dir(outputPath)); statErr != nil {
+		return fmt.Errorf("写入 CSV 文件失败: %w", statErr)
+	}
+	if err := AtomicWrite(outputPath, []byte(csvContent), 0o644); err != nil {
+		return fmt.Errorf("写入 CSV 文件失败: %w", err)
+	}
+	if hasMore {
+		deps.Out.PrintInfo(fmt.Sprintf("导出完成（数据已截断，不是完整表格）: %s", outputPath))
+		return nil
+	}
+	deps.Out.PrintInfo(fmt.Sprintf("导出完成: %s", outputPath))
+	return nil
+}
+
+// parseGetRangeAsCsvResult 从 get_range_as_csv 的 MCP 响应中提取 csv 文本与 hasMore 标志。
+//
+// csv 字段缺失或类型不对，必须报错而不是当成空表：调用方会把空内容写进
+// --output，用 0 字节覆盖已有文件并打印"导出完成"，等于静默数据丢失。
+// 「字段存在且为空串」是合法的（真的空区域），与「字段缺失」区分开。
+func parseGetRangeAsCsvResult(text string) (csv string, hasMore bool, err error) {
+	var data map[string]any
+	if e := json.Unmarshal([]byte(text), &data); e != nil {
+		return "", false, fmt.Errorf("解析 get_range_as_csv 响应失败: %w", e)
+	}
+	if raw, wrapped := data["result"]; wrapped {
+		result, ok := raw.(map[string]any)
+		if !ok {
+			return "", false, fmt.Errorf("解析 get_range_as_csv 响应失败: result 不是对象，响应: %s", text)
+		}
+		data = result
+	}
+	raw, exists := data["csv"]
+	if !exists {
+		return "", false, fmt.Errorf("解析 get_range_as_csv 响应失败: 缺少 csv 字段，响应: %s", text)
+	}
+	csvVal, ok := raw.(string)
+	if !ok {
+		return "", false, fmt.Errorf("解析 get_range_as_csv 响应失败: csv 字段不是字符串（%T），响应: %s", raw, text)
+	}
+	csv = csvVal
+	if hm, ok := data["hasMore"].(bool); ok {
+		hasMore = hm
+	}
+	return csv, hasMore, nil
+}

internal/helpers/sheet_export_csv_test.go

@@ -0,0 +1,353 @@
+// 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 helpers
+
+import (
+	"bytes"
+	"errors"
+	"os"
+	"path/filepath"
+	"reflect"
+	"sort"
+	"strings"
+	"testing"
+
+	"github.com/spf13/pflag"
+)
+
+func executeSheetExportCsv(t *testing.T, caller *scriptedToolCaller, args ...string) error {
+	t.Helper()
+	installScriptedCaller(t, caller)
+	oldArgs := os.Args
+	os.Args = []string{"dws", "sheet"}
+	t.Cleanup(func() { os.Args = oldArgs })
+	cmd := newSheetExportCsvCmd()
+	for index := 0; index < len(args); index += 2 {
+		if err := cmd.Flags().Set(args[index], args[index+1]); err != nil {
+			t.Fatalf("set %s: %v", args[index], err)
+		}
+	}
+	return runSheetExportCsv(cmd, nil)
+}
+
+// export-csv 是独立叶子，不能和 xlsx 的 sheet export 共用一套 flag：混在一条命令里
+// 时，自动化漏写 --export-format csv 会让 --range 被静默丢掉、导出整篇工作簿而仍报
+// 成功。这里钉住两条命令的 flag 面互不渗透。
+func TestSheetExportAndExportCsvFlagsDoNotLeak(t *testing.T) {
+	xlsxOnly := []string{}
+	newExportCmd().Flags().VisitAll(func(f *pflag.Flag) { xlsxOnly = append(xlsxOnly, f.Name) })
+	sort.Strings(xlsxOnly)
+	if want := []string{"node", "output"}; !reflect.DeepEqual(xlsxOnly, want) {
+		t.Errorf("sheet export flags = %v, want %v（csv 专属 flag 不能回到 xlsx 命令上）", xlsxOnly, want)
+	}
+
+	csvFlags := map[string]bool{}
+	newSheetExportCsvCmd().Flags().VisitAll(func(f *pflag.Flag) { csvFlags[f.Name] = true })
+	for _, name := range []string{"node", "output", "sheet-id", "range", "value-render-option", "allow-truncated"} {
+		if !csvFlags[name] {
+			t.Errorf("sheet export-csv 缺少 flag --%s", name)
+		}
+	}
+	// 不能出现格式路由 flag：格式由命令名决定，避免又回到单命令多分支的老问题。
+	if csvFlags["export-format"] {
+		t.Error("sheet export-csv 不应有 --export-format（格式已由命令名区分）")
+	}
+}
+
+func TestSheetExportCsvRequiresNode(t *testing.T) {
+	caller := &scriptedToolCaller{}
+	err := executeSheetExportCsv(t, caller)
+	if err == nil || !strings.Contains(err.Error(), "--node is required") {
+		t.Fatalf("err = %v, want required node", err)
+	}
+	if caller.calls != 0 {
+		t.Fatalf("calls = %d, want 0", caller.calls)
+	}
+}
+
+// --value-render-option 的枚举必须在发请求之前校验，否则非法取值会被静默透传。
+func TestSheetExportCsvValidatesValueRenderOption(t *testing.T) {
+	caller := &scriptedToolCaller{}
+	err := executeSheetExportCsv(t, caller, "node", "NODE", "value-render-option", "pretty")
+	if err == nil ||
+		!strings.Contains(err.Error(), "--value-render-option 必须为 formatted_value / raw_value / formula") {
+		t.Fatalf("err = %v, want enum rejection", err)
+	}
+	if caller.calls != 0 {
+		t.Fatalf("calls = %d, want 0（校验必须在发请求之前）", caller.calls)
+	}
+
+	for _, option := range []string{"formatted_value", "raw_value", "formula", "FORMULA", " raw_value "} {
+		ok := &scriptedToolCaller{steps: []scriptedToolStep{{text: `{"csv":"a,b\n"}`}}}
+		if err := executeSheetExportCsv(t, ok, "node", "NODE", "value-render-option", option); err != nil {
+			t.Fatalf("value-render-option=%q 应被接受: %v", option, err)
+		}
+	}
+}
+
+func TestSheetExportCsvDryRunPreviewsSelectors(t *testing.T) {
+	caller := &scriptedToolCaller{dry: true}
+	if err := executeSheetExportCsv(t, caller,
+		"node", "NODE", "sheet-id", "SHEET_1", "output", "out.csv"); err != nil {
+		t.Fatalf("csv dry run: %v", err)
+	}
+	if caller.calls != 0 {
+		t.Fatalf("dry run 不应发请求, calls = %d", caller.calls)
+	}
+}
+
+func TestSheetExportCsvForwardsSelectorsAndWarnsOnTruncation(t *testing.T) {
+	caller := &scriptedToolCaller{steps: []scriptedToolStep{
+		{text: `{"csv":"a,b\n","hasMore":true}`},
+	}}
+	// 截断默认失败，所以要显式放行才能走到落盘/输出。
+	if err := executeSheetExportCsv(t, caller,
+		"node", "NODE", "range", "A1:Z1000", "sheet-id", "SHEET_1",
+		"value-render-option", "raw_value", "allow-truncated", "true"); err != nil {
+		t.Fatalf("csv export: %v", err)
+	}
+	for _, kv := range [][2]string{
+		{"range", "A1:Z1000"},
+		{"sheetId", "SHEET_1"},
+		{"valueRenderOption", "raw_value"},
+	} {
+		if got := caller.args[kv[0]]; got != kv[1] {
+			t.Errorf("%s 未透传: %#v", kv[0], got)
+		}
+	}
+	// csv 正文不带行号前缀，annotateRowNumbers 必须显式关掉。
+	if caller.args["annotateRowNumbers"] != false {
+		t.Fatalf("annotateRowNumbers = %#v, want false", caller.args["annotateRowNumbers"])
+	}
+}
+
+// 截断必须 fail-closed：默认报错、不写文件、不报“导出完成”。否则自动化调用方会把
+// 不完整文件当成完整导出，且已存在的目标文件会被截断数据覆盖。
+func TestSheetExportCsvFailsOnTruncationWithoutOptIn(t *testing.T) {
+	dir := t.TempDir()
+	out := filepath.Join(dir, "out.csv")
+	if err := os.WriteFile(out, []byte("PREEXISTING\n"), 0o644); err != nil {
+		t.Fatal(err)
+	}
+	caller := &scriptedToolCaller{steps: []scriptedToolStep{
+		{text: `{"csv":"a,b\n","hasMore":true}`},
+	}}
+	err := executeSheetExportCsv(t, caller, "node", "NODE", "output", out)
+	if err == nil {
+		t.Fatal("截断时未放行应报错，而非静默成功")
+	}
+	for _, want := range []string{"截断", "已中止导出", "--allow-truncated"} {
+		if !strings.Contains(err.Error(), want) {
+			t.Errorf("错误信息缺少 %q: %v", want, err)
+		}
+	}
+	// 关键：既有文件必须保持原样，不能被截断数据覆盖。
+	body, readErr := os.ReadFile(out)
+	if readErr != nil {
+		t.Fatalf("read output: %v", readErr)
+	}
+	if string(body) != "PREEXISTING\n" {
+		t.Fatalf("目标文件被截断数据覆盖了: %q", string(body))
+	}
+}
+
+// 显式放行后允许落盘，但成功信息必须点明数据不完整。
+func TestSheetExportCsvAllowTruncatedWritesAndFlagsIncompleteness(t *testing.T) {
+	dir := t.TempDir()
+	out := filepath.Join(dir, "out.csv")
+	caller := &scriptedToolCaller{steps: []scriptedToolStep{
+		{text: `{"csv":"a,b\n","hasMore":true}`},
+	}}
+	if err := executeSheetExportCsv(t, caller,
+		"node", "NODE", "output", out, "allow-truncated", "true"); err != nil {
+		t.Fatalf("allow-truncated 应允许导出: %v", err)
+	}
+	body, err := os.ReadFile(out)
+	if err != nil {
+		t.Fatalf("read output: %v", err)
+	}
+	if string(body) != "a,b\n" {
+		t.Fatalf("csv 内容 = %q", string(body))
+	}
+}
+
+// 未截断时不受影响：不需要 --allow-truncated，成功信息也不提不完整。
+func TestSheetExportCsvCompleteReadNeedsNoOptIn(t *testing.T) {
+	dir := t.TempDir()
+	out := filepath.Join(dir, "out.csv")
+	caller := &scriptedToolCaller{steps: []scriptedToolStep{{text: `{"csv":"a,b\n","hasMore":false}`}}}
+	if err := executeSheetExportCsv(t, caller, "node", "NODE", "output", out); err != nil {
+		t.Fatalf("完整读取不应要求放行: %v", err)
+	}
+	body, err := os.ReadFile(out)
+	if err != nil || string(body) != "a,b\n" {
+		t.Fatalf("csv 内容 = %q, err = %v", string(body), err)
+	}
+}
+
+func TestSheetExportCsvWritesIntoDirectory(t *testing.T) {
+	dir := t.TempDir()
+	caller := &scriptedToolCaller{steps: []scriptedToolStep{{text: `{"csv":"a,b\n"}`}}}
+	if err := executeSheetExportCsv(t, caller, "node", "NODE", "output", dir); err != nil {
+		t.Fatalf("csv export: %v", err)
+	}
+	body, err := os.ReadFile(filepath.Join(dir, "sheet-export.csv"))
+	if err != nil {
+		t.Fatalf("read exported csv: %v", err)
+	}
+	if string(body) != "a,b\n" {
+		t.Fatalf("csv 内容 = %q", string(body))
+	}
+}
+
+// 落盘失败必须保住已有文件：os.WriteFile 会先截断目标，写到一半失败（磁盘满、
+// 配额、I/O 错误）用户的原文件就没了。走 AtomicWrite 后失败只丢临时文件。
+func TestSheetExportCsvWriteFailureKeepsExistingFile(t *testing.T) {
+	out := filepath.Join(t.TempDir(), "existing.csv")
+	const original = "原有重要数据\n1,2,3\n"
+	if err := os.WriteFile(out, []byte(original), 0o644); err != nil {
+		t.Fatal(err)
+	}
+
+	// 模拟"临时文件写完、替换目标时失败"，这是 os.WriteFile 会毁掉原文件的时刻。
+	previousRename := atomicRename
+	t.Cleanup(func() { atomicRename = previousRename })
+	atomicRename = func(string, string) error { return errors.New("no space left on device") }
+
+	caller := &scriptedToolCaller{steps: []scriptedToolStep{{text: `{"csv":"a,b\n"}`}}}
+	err := executeSheetExportCsv(t, caller, "node", "NODE", "output", out)
+	if err == nil || !strings.Contains(err.Error(), "写入 CSV 文件失败") {
+		t.Fatalf("err = %v, want write failure", err)
+	}
+
+	body, readErr := os.ReadFile(out)
+	if readErr != nil {
+		t.Fatalf("read output: %v", readErr)
+	}
+	if string(body) != original {
+		t.Fatalf("写入失败后原文件被破坏为 %q，应保持不变", string(body))
+	}
+
+	// 临时文件不能留在目录里冒充导出产物。
+	entries, readDirErr := os.ReadDir(filepath.Dir(out))
+	if readDirErr != nil {
+		t.Fatalf("read dir: %v", readDirErr)
+	}
+	if len(entries) != 1 || entries[0].Name() != filepath.Base(out) {
+		names := make([]string, 0, len(entries))
+		for _, entry := range entries {
+			names = append(names, entry.Name())
+		}
+		t.Fatalf("目录残留临时文件: %v", names)
+	}
+}
+
+func TestSheetExportCsvReportsWriteFailure(t *testing.T) {
+	caller := &scriptedToolCaller{steps: []scriptedToolStep{{text: `{"csv":"a,b\n"}`}}}
+	missing := filepath.Join(t.TempDir(), "no-such-dir", "out.csv")
+	err := executeSheetExportCsv(t, caller, "node", "NODE", "output", missing)
+	if err == nil || !strings.Contains(err.Error(), "写入 CSV 文件失败") {
+		t.Fatalf("err = %v, want write failure", err)
+	}
+}
+
+func TestSheetExportCsvReportsReadFailure(t *testing.T) {
+	caller := &scriptedToolCaller{steps: []scriptedToolStep{{text: `{"errorCode":"NoPermission"}`}}}
+	err := executeSheetExportCsv(t, caller, "node", "NODE")
+	if err == nil || !strings.Contains(err.Error(), "读取 CSV 失败") {
+		t.Fatalf("err = %v, want read failure", err)
+	}
+}
+
+func TestParseGetRangeAsCsvResult(t *testing.T) {
+	// 兼容裸响应与 result 包装两种形状
+	for _, body := range []string{
+		`{"csv":"a,b\n","hasMore":true}`,
+		`{"result":{"csv":"a,b\n","hasMore":true}}`,
+	} {
+		csv, hasMore, err := parseGetRangeAsCsvResult(body)
+		if err != nil || csv != "a,b\n" || !hasMore {
+			t.Fatalf("parseGetRangeAsCsvResult(%s) = (%q,%v,%v)", body, csv, hasMore, err)
+		}
+	}
+	// 字段存在但为空串是合法的（真的空区域）
+	csv, hasMore, err := parseGetRangeAsCsvResult(`{"csv":""}`)
+	if err != nil || csv != "" || hasMore {
+		t.Fatalf(`csv:"" = (%q,%v,%v), want ("",false,nil)`, csv, hasMore, err)
+	}
+}
+
+// csv 字段缺失或类型不对必须报错。此前会被当成空表，配合 --output 会用
+// 0 字节覆盖已有文件并打印"导出完成"，属于静默数据丢失。
+func TestParseGetRangeAsCsvResultRejectsMissingOrBadCsvField(t *testing.T) {
+	for _, tc := range []struct{ body, want string }{
+		{"not json", "解析 get_range_as_csv 响应失败"},
+		{`{"message":"something odd"}`, "缺少 csv 字段"},
+		{`{"csv":123}`, "csv 字段不是字符串"},
+		{`{"csv":null}`, "csv 字段不是字符串"},
+		{`{"result":"not-an-object"}`, "result 不是对象"},
+		{`{"result":{"message":"odd"}}`, "缺少 csv 字段"},
+	} {
+		if _, _, err := parseGetRangeAsCsvResult(tc.body); err == nil ||
+			!strings.Contains(err.Error(), tc.want) {
+			t.Errorf("parseGetRangeAsCsvResult(%s) err = %v, want contains %q", tc.body, err, tc.want)
+		}
+	}
+}
+
+// 端到端：响应缺 csv 字段时，绝不能覆盖 --output 指向的已有文件。
+func TestSheetExportCsvNeverTruncatesOutputOnBadResponse(t *testing.T) {
+	out := filepath.Join(t.TempDir(), "existing.csv")
+	const original = "原有重要数据\n1,2,3\n"
+	if err := os.WriteFile(out, []byte(original), 0o644); err != nil {
+		t.Fatal(err)
+	}
+	caller := &scriptedToolCaller{steps: []scriptedToolStep{{text: `{"message":"something odd"}`}}}
+	err := executeSheetExportCsv(t, caller, "node", "NODE", "output", out)
+	if err == nil || !strings.Contains(err.Error(), "缺少 csv 字段") {
+		t.Fatalf("err = %v, want missing csv field", err)
+	}
+	body, readErr := os.ReadFile(out)
+	if readErr != nil {
+		t.Fatal(readErr)
+	}
+	if string(body) != original {
+		t.Fatalf("已有文件被改写为 %q，原内容应保持不变", string(body))
+	}
+}
+
+// CSV 正文走 stdout，截断警告必须只走 stderr：否则管道/重定向拿到的文件里会
+// 混入 [WARN] 文本，不再是合法 RFC4180 CSV。而大表恰恰最容易触发这个分支。
+func TestSheetExportCsvKeepsStdoutPureAndWarnsOnStderr(t *testing.T) {
+	previousDeps := deps
+	t.Cleanup(func() { deps = previousDeps })
+
+	caller := &scriptedToolCaller{steps: []scriptedToolStep{{text: `{"csv":"a,b\nc,d\n","hasMore":true}`}}}
+	installScriptedCaller(t, caller)
+	installSheetProductArgs(t)
+	var stdout, stderr bytes.Buffer
+	deps.Out.w, deps.Out.errW = &stdout, &stderr
+
+	cmd := newSheetExportCsvCmd()
+	// 截断默认 fail-closed；本用例验证的是放行后 stdout 仍然纯净、警告只走 stderr。
+	for _, kv := range [][2]string{{"node", "NODE"}, {"allow-truncated", "true"}} {
+		if err := cmd.Flags().Set(kv[0], kv[1]); err != nil {
+			t.Fatalf("set --%s: %v", kv[0], err)
+		}
+	}
+	if err := runSheetExportCsv(cmd, nil); err != nil {
+		t.Fatalf("csv export: %v", err)
+	}
+
+	if got := stdout.String(); got != "a,b\nc,d\n" {
+		t.Fatalf("stdout = %q, want 纯 CSV 正文", got)
+	}
+	if strings.Contains(stdout.String(), "WARN") || strings.Contains(stdout.String(), "截断") {
+		t.Fatalf("stdout 混入了警告文本: %q", stdout.String())
+	}
+	if !strings.Contains(stderr.String(), "CSV 已被截断") {
+		t.Fatalf("stderr = %q, want 截断提示", stderr.String())
+	}
+}

internal/helpers/sheet_style_batch_budget_test.go

@@ -0,0 +1,376 @@
+// 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 helpers
+
+import (
+	"encoding/json"
+	"fmt"
+	"io"
+	"os"
+	"path/filepath"
+	"strings"
+	"testing"
+
+	"github.com/spf13/cobra"
+	"github.com/spf13/pflag"
+)
+
+// 跨 range 累计单元格预算（对齐飞书 checkBatchStampBudget）。
+//
+// 每个 "S!A1:Z1000" 是 26×1000=26000 格：单独看都在 buildStyleCells 的
+// 单区域上限（rows≤1000、rows×cols≤30000）之内，所以只有累计预算能拦住它们。
+// 7 个 = 182000 通过，8 个 = 208000 越过 200000 上限。
+const budgetTestRange = "A1:Z1000"
+
+func budgetTestRanges(n int) []string {
+	out := make([]string, 0, n)
+	for i := 0; i < n; i++ {
+		out = append(out, fmt.Sprintf("Sheet%d!%s", i+1, budgetTestRange))
+	}
+	return out
+}
+
+func newBatchStyleCmdWithStyle(t *testing.T) *cobra.Command {
+	t.Helper()
+	cmd := newRangeBatchSetStyleCmd()
+	if err := cmd.Flags().Set("bg-color", "#FFF2CC"); err != nil {
+		t.Fatalf("set --bg-color: %v", err)
+	}
+	return cmd
+}
+
+func TestBatchStyleRangesRespectAggregateCellBudget(t *testing.T) {
+	// 7 个区域刚好在预算内：必须成功，且每个区域各产出一个 op。
+	within, err := json.Marshal(budgetTestRanges(7))
+	if err != nil {
+		t.Fatal(err)
+	}
+	ops, err := buildBatchStyleOpsFromRanges(newBatchStyleCmdWithStyle(t), string(within))
+	if err != nil {
+		t.Fatalf("7 个区域（累计 182000 格）应在预算内: %v", err)
+	}
+	if len(ops) != 7 {
+		t.Fatalf("operations = %d, want 7", len(ops))
+	}
+
+	// 第 8 个把累计推到 208000，越过 200000 上限。
+	over, err := json.Marshal(budgetTestRanges(8))
+	if err != nil {
+		t.Fatal(err)
+	}
+	_, err = buildBatchStyleOpsFromRanges(newBatchStyleCmdWithStyle(t), string(over))
+	if err == nil {
+		t.Fatal("8 个区域（累计 208000 格）必须被累计预算拒绝")
+	}
+	for _, want := range []string{"--ranges[7]", "208000", "200000"} {
+		if !strings.Contains(err.Error(), want) {
+			t.Fatalf("error = %v, want contains %q", err, want)
+		}
+	}
+}
+
+func TestBatchStyleFileRespectsAggregateCellBudgetAndItemCap(t *testing.T) {
+	writeBatch := func(t *testing.T, items []map[string]any) string {
+		t.Helper()
+		data, err := json.Marshal(items)
+		if err != nil {
+			t.Fatal(err)
+		}
+		path := filepath.Join(t.TempDir(), "styles.json")
+		if err := os.WriteFile(path, data, 0o600); err != nil {
+			t.Fatalf("write batch fixture: %v", err)
+		}
+		return path
+	}
+	item := func(i int) map[string]any {
+		return map[string]any{
+			"sheetId":    fmt.Sprintf("Sheet%d", i+1),
+			"range":      budgetTestRange,
+			"fontWeight": "bold",
+		}
+	}
+
+	// 累计预算：8 条 × 26000 = 208000 > 200000。
+	overBudget := make([]map[string]any, 0, 8)
+	for i := 0; i < 8; i++ {
+		overBudget = append(overBudget, item(i))
+	}
+	_, err := buildBatchStyleOpsFromFile(writeBatch(t, overBudget))
+	if err == nil {
+		t.Fatal("--batch 累计 208000 格必须被拒绝")
+	}
+	for _, want := range []string{"第 8/8 条", "208000", "200000"} {
+		if !strings.Contains(err.Error(), want) {
+			t.Fatalf("error = %v, want contains %q", err, want)
+		}
+	}
+
+	// 条数上限与 --ranges 对齐：101 条应在展开任何矩阵之前就被拒。
+	tooMany := make([]map[string]any, 0, maxBatchStyleRanges+1)
+	for i := 0; i < maxBatchStyleRanges+1; i++ {
+		tooMany = append(tooMany, map[string]any{
+			"sheetId": "Sheet1", "range": "A1", "fontWeight": "bold",
+		})
+	}
+	_, err = buildBatchStyleOpsFromFile(writeBatch(t, tooMany))
+	if err == nil || !strings.Contains(err.Error(), "--batch 最多 100 条") {
+		t.Fatalf("error = %v, want --batch item cap", err)
+	}
+
+	// 预算内的正常配置仍然照常展开。
+	ops, err := buildBatchStyleOpsFromFile(writeBatch(t, []map[string]any{item(0), item(1)}))
+	if err != nil {
+		t.Fatalf("2 条应在预算内: %v", err)
+	}
+	if len(ops) != 2 {
+		t.Fatalf("operations = %d, want 2", len(ops))
+	}
+}
+
+func TestAddBatchStyleCellsAccumulatesAcrossCalls(t *testing.T) {
+	var total int64
+	if err := addBatchStyleCells(&total, 1000, 26); err != nil {
+		t.Fatalf("first range: %v", err)
+	}
+	if total != 26000 {
+		t.Fatalf("total = %d, want 26000", total)
+	}
+	// 单次就越界也必须拦住（累计器同时覆盖单区域巨量输入）。
+	var single int64
+	if err := addBatchStyleCells(&single, 1000, 1000); err == nil {
+		t.Fatal("1000x1000=1000000 必须超出累计预算")
+	}
+}
+
+// styleFlagNames 是手工维护的，必须与 bindStyleFlags 实际绑定的 flag 严格一致：
+// 漏登记会让 --batch 的互斥检查放过一个静默失效的 flag。
+func TestStyleFlagNamesMatchBoundFlags(t *testing.T) {
+	probe := &cobra.Command{Use: "probe"}
+	bindStyleFlags(probe)
+
+	bound := map[string]bool{}
+	probe.Flags().VisitAll(func(f *pflag.Flag) { bound[f.Name] = true })
+
+	listed := map[string]bool{}
+	for _, name := range styleFlagNames {
+		if listed[name] {
+			t.Fatalf("styleFlagNames 重复登记 %q", name)
+		}
+		listed[name] = true
+		if !bound[name] {
+			t.Errorf("styleFlagNames 登记了 %q，但 bindStyleFlags 没有绑定它", name)
+		}
+	}
+	for name := range bound {
+		if !listed[name] {
+			t.Errorf("bindStyleFlags 绑定了 %q，但 styleFlagNames 漏登记（--batch 互斥检查会放过它）", name)
+		}
+	}
+}
+
+func TestBatchSetStyleRejectsStyleFlagsInBatchMode(t *testing.T) {
+	previousDeps := deps
+	t.Cleanup(func() { deps = previousDeps })
+
+	batchPath := filepath.Join(t.TempDir(), "styles.json")
+	if err := os.WriteFile(batchPath, []byte(`[{"sheetId":"Sheet1","range":"A1:B2","fontWeight":"bold"}]`), 0o600); err != nil {
+		t.Fatalf("write batch fixture: %v", err)
+	}
+
+	caller := &sheetStyleDryRunCaller{format: "json"}
+	InitDeps(caller)
+
+	cmd := newRangeBatchSetStyleCmd()
+	cmd.SetOut(io.Discard)
+	cmd.SetErr(io.Discard)
+	cmd.SilenceErrors = true
+	cmd.SilenceUsage = true
+	cmd.SetArgs([]string{"--node", "NODE_ID", "--batch", batchPath, "--bg-color", "#FFF2CC", "--font-family", "Arial"})
+	err := cmd.Execute()
+	if err == nil {
+		t.Fatal("--batch 搭配命令行样式 flag 必须报错，而不是静默忽略")
+	}
+	for _, want := range []string{"--batch 模式下样式来自配置文件", "--bg-color", "--font-family"} {
+		if !strings.Contains(err.Error(), want) {
+			t.Fatalf("error = %v, want contains %q", err, want)
+		}
+	}
+	if caller.calls != 0 {
+		t.Fatalf("remote CallTool count = %d, want 0", caller.calls)
+	}
+}
+
+// buildBatchStyleOpsFromRanges 的每条前置校验都必须在组装 operations 之前拦住。
+func TestBatchStyleRangesRejectsInvalidInput(t *testing.T) {
+	tooMany := make([]string, 0, maxBatchStyleRanges+1)
+	for i := 0; i <= maxBatchStyleRanges; i++ {
+		tooMany = append(tooMany, `"Sheet1!A1"`)
+	}
+
+	for _, tc := range []struct {
+		name   string
+		ranges string
+		style  bool
+		want   string
+	}{
+		{name: "bad-json", ranges: `[`, style: true, want: "--ranges JSON 解析失败"},
+		{name: "empty-array", ranges: `[]`, style: true, want: "--ranges 不能为空数组"},
+		{
+			name:   "over-range-cap",
+			ranges: "[" + strings.Join(tooMany, ",") + "]",
+			style:  true,
+			want:   "--ranges 最多 100 项",
+		},
+		{
+			name:   "missing-sheet-prefix",
+			ranges: `["A1:B2"]`,
+			style:  true,
+			want:   "必须包含工作表前缀",
+		},
+		{
+			name:   "blank-sheet-prefix",
+			ranges: `[" !A1:B2"]`,
+			style:  true,
+			want:   "必须包含工作表前缀",
+		},
+		{
+			name:   "blank-range-after-prefix",
+			ranges: `["Sheet1! "]`,
+			style:  true,
+			want:   "必须包含工作表前缀",
+		},
+		{
+			name:   "bad-range-address",
+			ranges: `["Sheet1!zz"]`,
+			style:  true,
+			want:   "解析失败",
+		},
+		{
+			name:   "no-style-flag",
+			ranges: `["Sheet1!A1:B2"]`,
+			style:  false,
+			want:   "至少需要指定一个样式参数",
+		},
+	} {
+		t.Run(tc.name, func(t *testing.T) {
+			cmd := newRangeBatchSetStyleCmd()
+			if tc.style {
+				if err := cmd.Flags().Set("bg-color", "#FFF2CC"); err != nil {
+					t.Fatalf("set --bg-color: %v", err)
+				}
+			}
+			_, err := buildBatchStyleOpsFromRanges(cmd, tc.ranges)
+			if err == nil || !strings.Contains(err.Error(), tc.want) {
+				t.Fatalf("err = %v, want contains %q", err, tc.want)
+			}
+		})
+	}
+}
+
+// 空白工作表标识必须在下发之前被拒绝 —— 三条入口共用同一条不变量。
+//
+// 只按原始串里 ! 的位置判断，" !A1:B2" 会拆出空工作表名，操作却照样带着
+// sheetId:"" 提交：服务端要么让整批 batch_update 失败，要么落到默认工作表而不是
+// 用户指定的那张表（后者更糟，因为命令报成功）。--batch 的纯空白 sheetId 同理，
+// 它此前只挡 == ""。三条路径都必须在发请求之前失败。
+func TestBlankSheetIdentifierIsRejectedBeforeAnyRemoteCall(t *testing.T) {
+	previousDeps := deps
+	t.Cleanup(func() { deps = previousDeps })
+
+	batchDir := t.TempDir()
+	writeBatch := func(t *testing.T, name, body string) string {
+		t.Helper()
+		p := filepath.Join(batchDir, name)
+		if err := os.WriteFile(p, []byte(body), 0o600); err != nil {
+			t.Fatalf("write batch fixture: %v", err)
+		}
+		return p
+	}
+
+	for _, tc := range []struct {
+		name string
+		cmd  func() *cobra.Command
+		args func(t *testing.T) []string
+		want string
+	}{
+		{
+			name: "set-style-ranges-blank-sheet",
+			cmd:  newRangeBatchSetStyleCmd,
+			args: func(*testing.T) []string {
+				return []string{"--node", "NODE_ID", "--ranges", `[" !A1:B2"]`, "--bg-color", "#FFF2CC"}
+			},
+			want: "必须包含工作表前缀",
+		},
+		{
+			name: "set-style-batch-blank-sheet-id",
+			cmd:  newRangeBatchSetStyleCmd,
+			args: func(t *testing.T) []string {
+				p := writeBatch(t, "blank-sheet.json", `[{"sheetId":"   ","range":"A1:B2","fontWeight":"bold"}]`)
+				return []string{"--node", "NODE_ID", "--batch", p}
+			},
+			want: "缺少 sheetId 或 range",
+		},
+		{
+			name: "set-style-batch-blank-range",
+			cmd:  newRangeBatchSetStyleCmd,
+			args: func(t *testing.T) []string {
+				p := writeBatch(t, "blank-range.json", `[{"sheetId":"Sheet1","range":"  ","fontWeight":"bold"}]`)
+				return []string{"--node", "NODE_ID", "--batch", p}
+			},
+			want: "缺少 sheetId 或 range",
+		},
+		{
+			name: "batch-clear-ranges-blank-sheet",
+			cmd:  rangeBatchClearCoverageCommand,
+			args: func(*testing.T) []string {
+				return []string{"--node", "NODE_ID", "--ranges", `[" !A1:B2"]`}
+			},
+			want: "必须包含工作表前缀",
+		},
+	} {
+		t.Run(tc.name, func(t *testing.T) {
+			caller := &sheetStyleDryRunCaller{format: "json"}
+			InitDeps(caller)
+			deps.Out.w = io.Discard
+			deps.Out.errW = io.Discard
+
+			cmd := tc.cmd()
+			cmd.SetArgs(tc.args(t))
+			cmd.SilenceUsage, cmd.SilenceErrors = true, true
+			err := cmd.Execute()
+			if err == nil || !strings.Contains(err.Error(), tc.want) {
+				t.Fatalf("err = %v, want contains %q", err, tc.want)
+			}
+			if caller.calls != 0 {
+				t.Fatalf("remote CallTool count = %d, want 0", caller.calls)
+			}
+		})
+	}
+}
+
+// 判空看修剪后的值，但下发仍用原值：sheetId 可以是工作表**名**，名字允许带首尾
+// 空格，替用户修剪会指向另一张表。--batch 是需要精确名字时的入口，必须原样透传。
+func TestBatchStyleSheetIDIsSentVerbatimNotTrimmed(t *testing.T) {
+	p := filepath.Join(t.TempDir(), "padded.json")
+	if err := os.WriteFile(p, []byte(`[{"sheetId":" Sheet1 ","range":" A1:B2 ","fontWeight":"bold"}]`), 0o600); err != nil {
+		t.Fatalf("write batch fixture: %v", err)
+	}
+
+	ops, err := buildBatchStyleOpsFromFile(p)
+	if err != nil {
+		t.Fatalf("buildBatchStyleOpsFromFile = %v, want accepted", err)
+	}
+	if len(ops) != 1 {
+		t.Fatalf("operations = %d, want 1", len(ops))
+	}
+	op, _ := ops[0].(map[string]any)
+	input, _ := op["input"].(map[string]any)
+	if input["sheetId"] != " Sheet1 " {
+		t.Fatalf("sheetId = %#v, want %q (verbatim)", input["sheetId"], " Sheet1 ")
+	}
+	if input["rangeAddress"] != " A1:B2 " {
+		t.Fatalf("rangeAddress = %#v, want %q (verbatim)", input["rangeAddress"], " A1:B2 ")
+	}
+}

internal/helpers/sheet_style_coverage_more_test.go

@@ -12,17 +12,16 @@ func TestCrossPlatformCoverageSheetStyleRangeAndFontCoverage(t *testing.T) {
 	for _, raw := range []string{"", "Sheet1!A1:B2", "B2:A1", "A1", "A", "1", "A0", "A-1", "A1:bad"} {
 		_, _, _ = parseA1Range(raw)
 	}
-	_ = fillStringMatrix(2, 2, "x")
-	_ = fillIntMatrix(2, 2, 1)
-	for _, spec := range []*styleSpec{
-		{FontSize: 1, FontSizesJSON: `[[1]]`},
-		{FontSize: -1},
-		{FontSize: 12},
-		{FontSizesJSON: `{`},
-		{FontSizesJSON: `[[1,2]]`},
-		{FontSizesJSON: `[[1]]`},
+	for _, tc := range []struct {
+		scalar  int
+		jsonStr string
+	}{
+		{1, `[[1]]`}, {-1, ""}, {12, ""}, {0, `{`}, {0, `[[1,2]]`}, {0, `[[1]]`}, {0, ""},
 	} {
-		_ = applyFontSize(spec, 1, 1, map[string]any{})
+		if get, err := intGrid(tc.scalar, tc.jsonStr, "font-size", 1, 1); err == nil && get != nil {
+			_, _ = get(0, 0)
+			_, _ = get(9, 9)
+		}
 	}
 	for _, tc := range []struct {
 		scalar, raw string
@@ -32,7 +31,17 @@ func TestCrossPlatformCoverageSheetStyleRangeAndFontCoverage(t *testing.T) {
 		{"", "", nil}, {"", `{`, nil}, {"", `[["x","y"]]`, nil},
 		{"", `[[""]]`, hAlignEnum}, {"", `[["bad"]]`, hAlignEnum}, {"", `[["left"]]`, hAlignEnum},
 	} {
-		_ = apply2DString(tc.scalar, tc.raw, 1, 1, "align", "alignments", tc.enum, map[string]any{})
+		if get, err := strGrid(tc.scalar, tc.raw, "align", tc.enum, 1, 1); err == nil && get != nil {
+			_, _ = get(0, 0)
+			_, _ = get(9, 9)
+		}
+	}
+	for _, raw := range []string{"", `{`, `{}`, `{"nope":{"style":"solid"}}`, `{"top":{}}`,
+		`{"top":{"style":"bogus"}}`, `{"top":{"style":"solid"}}`, `{"top":{"style":"solid","color":"#000"}}`} {
+		_, _ = parseBorderStyles(raw)
+	}
+	for _, raw := range []string{"Sheet1!A1:B2", "A1:B2", "Sheet1!", "!A1"} {
+		_, _, _ = splitSheetPrefixedRange(raw, 0)
 	}
 	_ = maxColLenStr([][]string{{}, {"a", "b"}})
 	_ = maxColLen2D([][]int{{}, {1, 2}})
@@ -46,8 +55,16 @@ func TestCrossPlatformCoverageSheetStyleRangeAndFontCoverage(t *testing.T) {
 		{styleSpec{}, 0, 1}, {styleSpec{}, 1001, 1}, {styleSpec{}, 1000, 31},
 		{styleSpec{WordWrap: "invalid"}, 1, 1}, {styleSpec{WordWrap: "clip"}, 1, 1},
 		{styleSpec{NumberFormat: "General"}, 1, 1}, {styleSpec{}, 1, 1},
+		{styleSpec{FontStyle: "bogus"}, 1, 1}, {styleSpec{FontStyle: "italic"}, 1, 1},
+		{styleSpec{FontLine: "bogus"}, 1, 1}, {styleSpec{FontLine: "underline"}, 1, 1},
+		{styleSpec{FontLine: "line-through"}, 1, 1}, {styleSpec{FontLine: "none"}, 1, 1},
+		{styleSpec{FontFamily: "Arial"}, 1, 1},
+		{styleSpec{BorderStylesJSON: `{"top":{"style":"solid"}}`}, 1, 1},
+		{styleSpec{BorderStylesJSON: `{`}, 1, 1},
+		{styleSpec{BgColor: "#FFF", FontSize: 12, HAlign: "center", VAlign: "top",
+			FontColor: "#000", FontWeight: "bold"}, 2, 2},
 	} {
-		_ = applyStyleSpec(&tc.spec, tc.rows, tc.cols, map[string]any{})
+		_, _ = buildStyleCells(&tc.spec, tc.rows, tc.cols)
 	}
 }
 

internal/helpers/sheet_style_test.go

@@ -62,16 +62,22 @@ func TestCrossPlatformCoverageRangeBatchSetStyleDryRunNeverCallsRemote(t *testin
 				if err := json.Unmarshal(output.Bytes(), &payload); err != nil {
 					t.Fatalf("JSON dry-run stdout must be one document: %v\n%s", err, preview)
 				}
-				results, _ := payload["results"].([]any)
-				if len(results) != 1 {
-					t.Fatalf("JSON dry-run results = %#v", payload["results"])
+				// batch-set-style 现在组装为一次 batch_update 原子提交，
+				// dry-run 因此是单条标准记录，而不是本地循环产生的 results 数组。
+				if payload["tool"] != "batch_update" {
+					t.Fatalf("JSON dry-run tool = %#v, want batch_update", payload["tool"])
 				}
-				entry, _ := results[0].(map[string]any)
-				if entry["dryRun"] != true || entry["tool"] != "update_range" {
-					t.Fatalf("JSON dry-run result = %#v", entry)
+				args, _ := payload["arguments"].(map[string]any)
+				ops, _ := args["operations"].([]any)
+				if len(ops) != 1 {
+					t.Fatalf("JSON dry-run operations = %#v, want 1 item", args["operations"])
+				}
+				op, _ := ops[0].(map[string]any)
+				if op["toolName"] != "set_cell_range" {
+					t.Fatalf("JSON dry-run operation toolName = %#v", op["toolName"])
 				}
 			} else {
-				for _, want := range []string{"Tool:", "update_range", "Arguments:"} {
+				for _, want := range []string{"Tool:", "batch_update", "set_cell_range", "Arguments:"} {
 					if !strings.Contains(preview, want) {
 						t.Fatalf("dry-run preview missing %q:\n%s", want, preview)
 					}
@@ -104,3 +110,104 @@ func TestRangeBatchSetStylePropagatesJSONWriteFailure(t *testing.T) {
 		t.Fatalf("remote CallTool count = %d, want 0", caller.calls)
 	}
 }
+
+// 边框每条边只认 style / color。写错的键（colour）或类型不对的 color 此前被静默
+// 忽略：命令报成功，却画出一条没有颜色的边框。这与 --sheets / --styles 拒绝未知键
+// 是同一条不变量，三条使用路径（单区域 flag、batch、create-with-data --styles）共用
+// parseBorderStyles，所以核心在这里一次验完，再逐路径确认拒绝发生在发请求之前。
+func TestParseBorderStylesRejectsUnknownAndMistypedEdgeFields(t *testing.T) {
+	for _, tc := range []struct {
+		name string
+		json string
+		want string
+	}{
+		{"near-miss-color-key", `{"top":{"style":"solid","colour":"#f00"}}`, `未知字段 "colour"`},
+		{"unknown-key", `{"top":{"style":"solid","width":2}}`, `未知字段 "width"`},
+		{"camel-case-alias-hinted", `{"top":{"style":"solid","Color":"#f00"}}`, `应为 "color"`},
+		{"color-not-string", `{"top":{"style":"solid","color":123}}`, "color 必须是字符串，实际是 float64"},
+		{"color-empty-string", `{"top":{"style":"solid","color":""}}`, "color 不能为空字符串"},
+		{"style-not-string", `{"top":{"style":123}}`, "style 必须是字符串，实际是 float64"},
+		{"style-empty-string", `{"top":{"style":""}}`, "缺少 style"},
+		{"style-missing", `{"top":{"color":"#f00"}}`, "缺少 style"},
+		{"style-null", `{"top":{"style":null}}`, "缺少 style"},
+	} {
+		t.Run(tc.name, func(t *testing.T) {
+			_, err := parseBorderStyles(tc.json)
+			if err == nil || !strings.Contains(err.Error(), tc.want) {
+				t.Fatalf("err = %v, want contains %q", err, tc.want)
+			}
+			if err != nil && !strings.Contains(err.Error(), "--border-styles-json.top") {
+				t.Fatalf("err = %v, want the edge name in the message", err)
+			}
+		})
+	}
+
+	// 合法形态不能被误拒：只给 style、给 style+color、显式 null color（等同省略）
+	for _, tc := range []struct {
+		name      string
+		json      string
+		wantColor any
+	}{
+		{"style-only", `{"top":{"style":"solid"}}`, nil},
+		{"style-and-color", `{"top":{"style":"solid","color":"#FF0000"}}`, "#FF0000"},
+		{"explicit-null-color", `{"top":{"style":"solid","color":null}}`, nil},
+	} {
+		t.Run(tc.name, func(t *testing.T) {
+			out, err := parseBorderStyles(tc.json)
+			if err != nil {
+				t.Fatalf("parseBorderStyles(%s) = %v, want accepted", tc.json, err)
+			}
+			edge, _ := out["top"].(map[string]any)
+			if edge["style"] != "solid" {
+				t.Fatalf("style = %#v, want solid", edge["style"])
+			}
+			if edge["color"] != tc.wantColor {
+				t.Fatalf("color = %#v, want %#v", edge["color"], tc.wantColor)
+			}
+		})
+	}
+}
+
+// 单区域与 batch 两条路径：非法边框配置必须在组装/下发之前失败。
+func TestBorderStyleRejectionHappensBeforeAnyRemoteCall(t *testing.T) {
+	const badBorder = `{"top":{"style":"solid","colour":"#f00"}}`
+	previousDeps := deps
+	t.Cleanup(func() { deps = previousDeps })
+
+	batchPath := filepath.Join(t.TempDir(), "styles.json")
+	fixture := `[{"sheetId":"Sheet1","range":"A1:B2","borderStylesJson":"{\"top\":{\"style\":\"solid\",\"colour\":\"#f00\"}}"}]`
+	if err := os.WriteFile(batchPath, []byte(fixture), 0o600); err != nil {
+		t.Fatalf("write batch fixture: %v", err)
+	}
+
+	for _, tc := range []struct {
+		name string
+		args []string
+	}{
+		{"single-range", []string{"--node", "NODE_ID", "--sheet-id", "S", "--range", "A1:B2", "--border-styles-json", badBorder}},
+		{"batch-ranges", []string{"--node", "NODE_ID", "--ranges", `["Sheet1!A1:B2"]`, "--border-styles-json", badBorder}},
+		{"batch-file", []string{"--node", "NODE_ID", "--batch", batchPath}},
+	} {
+		t.Run(tc.name, func(t *testing.T) {
+			caller := &sheetStyleDryRunCaller{format: "json"}
+			InitDeps(caller)
+			var output bytes.Buffer
+			deps.Out.w = &output
+			deps.Out.errW = &output
+
+			cmd := newRangeSetStyleCmd()
+			if tc.name != "single-range" {
+				cmd = newRangeBatchSetStyleCmd()
+			}
+			cmd.SetArgs(tc.args)
+			cmd.SilenceUsage, cmd.SilenceErrors = true, true
+			err := cmd.Execute()
+			if err == nil || !strings.Contains(err.Error(), `未知字段 "colour"`) {
+				t.Fatalf("err = %v, want the unknown border field rejected", err)
+			}
+			if caller.calls != 0 {
+				t.Fatalf("remote CallTool count = %d, want 0", caller.calls)
+			}
+		})
+	}
+}

internal/helpers/small_remaining_coverage_test.go

@@ -12,7 +12,7 @@ import (
 )
 
 func TestCrossPlatformCoverageSheetAndMinutesSmallRemainingBranches(t *testing.T) {
-	if err := applyStyleSpec(&styleSpec{FontColorsJSON: "{"}, 1, 1, map[string]any{}); err == nil {
+	if _, err := buildStyleCells(&styleSpec{FontColorsJSON: "{"}, 1, 1); err == nil {
 		t.Fatal("invalid font colors JSON returned nil")
 	}
 

internal/interfacesnapshot/compare.go

@@ -73,6 +73,10 @@ func CompareAll(current Snapshot, references map[string]Snapshot) Report {
 //     become required; an existing path may not gain a new required flag,
 //   - new commands and flags are allowed.
 //
+// The single exception to the type rule is an individually reviewed migration
+// listed in reviewedFlagTypeChanges, and only when nothing else about the flag
+// changed. See reviewed.go.
+//
 // Comparing the effective local + inherited set is intentional. It catches a
 // persistent flag whose scope is accidentally narrowed to its declaring
 // command, while allowing a local flag to move to an ancestor without breaking
@@ -183,13 +187,20 @@ func compareEffectiveFlags(result *Comparison, acceptedPath string, oldCommand,
 			continue
 		}
 		if oldFlag.Type != newFlag.Type {
-			result.Blocking = append(result.Blocking, Change{
-				Kind:   "flag_type_changed",
-				Path:   acceptedPath,
-				Flag:   name,
-				Before: oldFlag.Type,
-				After:  newFlag.Type,
-			})
+			// Resolve the exemption against the canonical path, not acceptedPath.
+			// An aliased command reaches this function once per accepted spelling,
+			// so keying on acceptedPath would let every alias spelling re-report a
+			// reviewed migration.
+			if !reviewedFlagTypeChange(newCommand.Path, name, oldFlag.Type, newFlag.Type) ||
+				flagContractOtherwiseChanged(oldFlag, newFlag) {
+				result.Blocking = append(result.Blocking, Change{
+					Kind:   "flag_type_changed",
+					Path:   acceptedPath,
+					Flag:   name,
+					Before: oldFlag.Type,
+					After:  newFlag.Type,
+				})
+			}
 		}
 		if !oldFlag.Required && newFlag.Required {
 			result.Blocking = append(result.Blocking, Change{
@@ -218,6 +229,15 @@ func compareEffectiveFlags(result *Comparison, acceptedPath string, oldCommand,
 				After:  newFlag.NoOpt,
 			})
 		}
+		if oldFlag.AliasOf != "" && newFlag.AliasOf != oldFlag.AliasOf {
+			result.Blocking = append(result.Blocking, Change{
+				Kind:   "flag_alias_target_changed",
+				Path:   acceptedPath,
+				Flag:   name,
+				Before: oldFlag.AliasOf,
+				After:  newFlag.AliasOf,
+			})
+		}
 		if !oldFlag.Hidden && newFlag.Hidden {
 			result.Blocking = append(result.Blocking, Change{
 				Kind:   "flag_became_hidden",

internal/interfacesnapshot/compare_test.go

@@ -13,7 +13,11 @@
 
 package interfacesnapshot
 
-import "testing"
+import (
+	"os"
+	"regexp"
+	"testing"
+)
 
 func TestCrossPlatformCoverageCompareAdmissionPolicy(t *testing.T) {
 	base := testSnapshot(
@@ -247,6 +251,40 @@ func TestCrossPlatformCoverageCompareBlocksCallableMetadataRegressions(t *testin
 	}
 }
 
+func TestCrossPlatformCoverageCompareBlocksFlagAliasTargetRegression(t *testing.T) {
+	baseLegacy := testFlag("legacy-id", "string", false)
+	baseLegacy.AliasOf = "message-id"
+	base := testSnapshot(
+		testCommand("dws"),
+		testCommand("dws send", baseLegacy, testFlag("message-id", "string", false)),
+	)
+
+	for _, target := range []string{"", "other-id"} {
+		t.Run("target_"+target, func(t *testing.T) {
+			currentLegacy := baseLegacy
+			currentLegacy.AliasOf = target
+			current := testSnapshot(
+				testCommand("dws"),
+				testCommand(
+					"dws send",
+					currentLegacy,
+					testFlag("message-id", "string", false),
+					testFlag("other-id", "string", false),
+				),
+			)
+			comparison := Compare(current, base, "base")
+			if comparison.Compatible || !hasFlagChange(
+				comparison.Blocking,
+				"flag_alias_target_changed",
+				"dws send",
+				"legacy-id",
+			) {
+				t.Fatalf("alias target regression was not blocked: %#v", comparison)
+			}
+		})
+	}
+}
+
 func TestCrossPlatformCoverageCompareUsesEffectiveFlagsAtEveryCommandPath(t *testing.T) {
 	base := testSnapshot(
 		testCommand("dws", testFlag("profile", "string", false)),
@@ -310,6 +348,511 @@ func TestCrossPlatformCoverageCompareAllRequiresBothReferencesToPass(t *testing.
 	}
 }
 
+func TestCrossPlatformCoverageCompareAllWithFlagMigrationsAuthorizesOnlyTheExactTransition(t *testing.T) {
+	before := testFlagMigrationSnapshot(false, true)
+	authority := testFlagMigrationManifest(FlagMigrationPending)
+	candidate := testFlagMigrationManifest(FlagMigrationConsumed)
+
+	report, err := CompareAllWithFlagMigrations(
+		testFlagMigrationSnapshot(true, true),
+		map[string]Snapshot{
+			"merge-base": before,
+			"stable":     before,
+		},
+		authority,
+		candidate,
+	)
+	if err != nil {
+		t.Fatalf("exact base-owned migration was rejected: %v", err)
+	}
+	if !report.Compatible {
+		t.Fatalf("exact migration remained incompatible: %#v", report.Comparisons)
+	}
+	for _, comparison := range report.Comparisons {
+		if len(comparison.Blocking) != 0 {
+			t.Fatalf("reference %q retained blocking changes: %#v", comparison.Reference, comparison.Blocking)
+		}
+	}
+}
+
+func TestCrossPlatformCoverageCompareAllWithFlagMigrationsComposesWithReviewedTypeChange(t *testing.T) {
+	registerReviewedFixture(t, flagTypeChange{CommandPath: "dws search", Flag: "query", From: "string", To: "int"})
+
+	before := testFlagMigrationSnapshot(false, true)
+	before.Commands = append(before.Commands, testCommand("dws search", testFlag("query", "string", false)))
+	after := testFlagMigrationSnapshot(true, true)
+	after.Commands = append(after.Commands, testCommand("dws search", testFlag("query", "int", false)))
+
+	report, err := CompareAllWithFlagMigrations(
+		after,
+		map[string]Snapshot{
+			"merge-base": before,
+			"stable":     before,
+		},
+		testFlagMigrationManifest(FlagMigrationPending),
+		testFlagMigrationManifest(FlagMigrationConsumed),
+	)
+	if err != nil {
+		t.Fatalf("composed reviewed migrations were rejected: %v", err)
+	}
+	if !report.Compatible {
+		t.Fatalf("composed reviewed migrations remained incompatible: %#v", report.Comparisons)
+	}
+}
+
+// registerReviewedFixture puts a fixture command in the reviewed table for the
+// duration of one test, so the behaviour tests exercise the real lookup instead
+// of depending on whichever production entries happen to exist.
+func registerReviewedFixture(t *testing.T, change flagTypeChange) {
+	t.Helper()
+	if _, exists := reviewedFlagTypeChanges[change]; exists {
+		t.Fatalf("fixture %+v collides with a production entry", change)
+	}
+	reviewedFlagTypeChanges[change] = struct{}{}
+	t.Cleanup(func() { delete(reviewedFlagTypeChanges, change) })
+}
+
+func TestCrossPlatformCoverageCompareAcceptsReviewedFlagTypeChange(t *testing.T) {
+	base := testSnapshot(
+		testCommand("dws"),
+		testCommand("dws search", testFlag("query", "string", false)),
+	)
+	current := testSnapshot(
+		testCommand("dws"),
+		testCommand("dws search", testFlag("query", "int", false)),
+	)
+
+	// Without an entry the migration is still a blocking change.
+	if comparison := Compare(current, base, "base"); !hasChangeKind(comparison.Blocking, "flag_type_changed") {
+		t.Fatalf("an unreviewed type change should block: %#v", comparison.Blocking)
+	}
+
+	registerReviewedFixture(t, flagTypeChange{CommandPath: "dws search", Flag: "query", From: "string", To: "int"})
+	if comparison := Compare(current, base, "base"); !comparison.Compatible {
+		t.Fatalf("a reviewed string->int migration should pass: %#v", comparison.Blocking)
+	}
+}
+
+func TestCrossPlatformCoverageCompareReviewedFlagTypeChangeIsDirectionSensitive(t *testing.T) {
+	registerReviewedFixture(t, flagTypeChange{CommandPath: "dws search", Flag: "query", From: "string", To: "int"})
+
+	// The reverse migration shares command and flag but not direction, so the
+	// string->int entry must not admit it.
+	base := testSnapshot(
+		testCommand("dws"),
+		testCommand("dws search", testFlag("query", "int", false)),
+	)
+	current := testSnapshot(
+		testCommand("dws"),
+		testCommand("dws search", testFlag("query", "string", false)),
+	)
+	if comparison := Compare(current, base, "base"); comparison.Compatible {
+		t.Fatal("int->string must not be admitted by a string->int entry")
+	}
+}
+
+func TestCrossPlatformCoverageCompareReviewedFlagTypeChangeRejectsOtherCommandsAndFlags(t *testing.T) {
+	base := testSnapshot(
+		testCommand("dws"),
+		testCommand("dws search", testFlag("query", "string", false)),
+	)
+	current := testSnapshot(
+		testCommand("dws"),
+		testCommand("dws search", testFlag("query", "int", false)),
+	)
+
+	// Same flag and direction, different command.
+	registerReviewedFixture(t, flagTypeChange{CommandPath: "dws status", Flag: "query", From: "string", To: "int"})
+	if comparison := Compare(current, base, "base"); comparison.Compatible {
+		t.Fatal("an entry for another command must not admit this migration")
+	}
+
+	// Same command and direction, different flag.
+	registerReviewedFixture(t, flagTypeChange{CommandPath: "dws search", Flag: "elsewhere", From: "string", To: "int"})
+	if comparison := Compare(current, base, "base"); comparison.Compatible {
+		t.Fatal("an entry for another flag must not admit this migration")
+	}
+}
+
+// A reviewed migration is only accepted when the rest of the flag's contract
+// held still. Each case bundles one unrelated regression with the reviewed type
+// change and requires the type failure to reappear, which also anchors
+// flagContractOtherwiseChanged against the checks it mirrors: a condition that
+// drifts out of sync fails here instead of silently widening the exemption.
+func TestCrossPlatformCoverageCompareReviewedFlagTypeChangeRejectsBundledRegression(t *testing.T) {
+	registerReviewedFixture(t, flagTypeChange{CommandPath: "dws search", Flag: "query", From: "string", To: "int"})
+
+	tests := []struct {
+		name string
+		base Flag
+		next Flag
+		kind string
+	}{
+		{
+			name: "became required",
+			base: Flag{Name: "query", Type: "string"},
+			next: Flag{Name: "query", Type: "int", Required: true},
+			kind: "flag_became_required",
+		},
+		{
+			name: "lost shorthand",
+			base: Flag{Name: "query", Type: "string", Shorthand: "q"},
+			next: Flag{Name: "query", Type: "int"},
+			kind: "flag_shorthand_changed",
+		},
+		{
+			name: "changed no-opt",
+			base: Flag{Name: "query", Type: "string", NoOpt: "all"},
+			next: Flag{Name: "query", Type: "int", NoOpt: "1"},
+			kind: "flag_no_opt_changed",
+		},
+		{
+			name: "became hidden",
+			base: Flag{Name: "query", Type: "string"},
+			next: Flag{Name: "query", Type: "int", Hidden: true},
+			kind: "flag_became_hidden",
+		},
+		{
+			name: "changed alias target",
+			base: Flag{Name: "query", Type: "string", AliasOf: "old-query"},
+			next: Flag{Name: "query", Type: "int", AliasOf: "new-query"},
+			kind: "flag_alias_target_changed",
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			base := testSnapshot(testCommand("dws"), testCommand("dws search", test.base))
+			current := testSnapshot(testCommand("dws"), testCommand("dws search", test.next))
+			comparison := Compare(current, base, "base")
+			if !hasChangeKind(comparison.Blocking, test.kind) {
+				t.Fatalf("blocking=%#v, want kind %q", comparison.Blocking, test.kind)
+			}
+			if !hasChangeKind(comparison.Blocking, "flag_type_changed") {
+				t.Fatalf("a bundled regression must re-report the type change: %#v", comparison.Blocking)
+			}
+		})
+	}
+}
+
+// The exemption is resolved against the canonical Command.Path, never the
+// alias-expanded accepted path. An aliased command enters compareEffectiveFlags
+// once per accepted spelling, so keying on the accepted path would let every
+// alias spelling re-report a reviewed migration.
+func TestCrossPlatformCoverageCompareResolvesReviewedFlagTypeChangeByCanonicalPath(t *testing.T) {
+	base := testSnapshot(
+		testCommand("dws"),
+		testCommandWithAliases("dws search", []string{"find"}, testFlag("query", "string", false)),
+	)
+	current := testSnapshot(
+		testCommand("dws"),
+		testCommandWithAliases("dws search", []string{"find"}, testFlag("query", "int", false)),
+	)
+
+	// The canonical path admits the migration for every accepted spelling.
+	canonical := flagTypeChange{CommandPath: "dws search", Flag: "query", From: "string", To: "int"}
+	registerReviewedFixture(t, canonical)
+	if comparison := Compare(current, base, "base"); !comparison.Compatible {
+		t.Fatalf("canonical path should admit the migration on every alias: %#v", comparison.Blocking)
+	}
+	delete(reviewedFlagTypeChanges, canonical)
+
+	// An entry spelled with the alias path matches nothing, because the lookup
+	// never sees the accepted spelling.
+	registerReviewedFixture(t, flagTypeChange{CommandPath: "dws find", Flag: "query", From: "string", To: "int"})
+	if comparison := Compare(current, base, "base"); comparison.Compatible {
+		t.Fatal("an alias-spelled entry must not admit the migration")
+	}
+}
+
+// A string -> int migration always changes the recorded default ("" to "0").
+// Defaults are not part of what this gate enforces, and folding them into
+// flagContractOtherwiseChanged would make every table entry dead on arrival.
+func TestCrossPlatformCoverageCompareReviewedFlagTypeChangeToleratesDefaultDrift(t *testing.T) {
+	registerReviewedFixture(t, flagTypeChange{CommandPath: "dws search", Flag: "query", From: "string", To: "int"})
+
+	base := testSnapshot(
+		testCommand("dws"),
+		testCommand("dws search", Flag{Name: "query", Type: "string", Default: ""}),
+	)
+	current := testSnapshot(
+		testCommand("dws"),
+		testCommand("dws search", Flag{Name: "query", Type: "int", Default: "0"}),
+	)
+	if comparison := Compare(current, base, "base"); !comparison.Compatible {
+		t.Fatalf("default drift must not defeat a reviewed migration: %#v", comparison.Blocking)
+	}
+}
+
+// reviewedEntryPattern matches one table entry as gofmt writes it on a single
+// line. It is used to read the sibling copy of the table out of the policy
+// helper, which cannot be imported: that directory is copied into a worktree at
+// a historical revision and built there, so it must not depend on this package.
+var reviewedEntryPattern = regexp.MustCompile(
+	`\{CommandPath:\s*"([^"]*)",\s*Flag:\s*"([^"]*)",\s*From:\s*"([^"]*)",\s*To:\s*"([^"]*)"\}`)
+
+const interfaceBaselineReviewedPath = "../../scripts/policy/interface-baseline/reviewed.go"
+
+// The modern authority and legacy smoke helper keep mirrored tables so local
+// smoke checks cannot disagree with the authoritative comparison.
+func TestCrossPlatformCoverageReviewedFlagTypeTableMatchesInterfaceBaseline(t *testing.T) {
+	source, err := os.ReadFile(interfaceBaselineReviewedPath)
+	if err != nil {
+		t.Fatalf("读取 %s 失败: %v", interfaceBaselineReviewedPath, err)
+	}
+
+	sibling := make(map[flagTypeChange]struct{})
+	for _, match := range reviewedEntryPattern.FindAllStringSubmatch(string(source), -1) {
+		sibling[flagTypeChange{CommandPath: match[1], Flag: match[2], From: match[3], To: match[4]}] = struct{}{}
+	}
+
+	// Without this the whole guard degrades into comparing two empty sets the
+	// moment the pattern stops matching, which is exactly how a duplicated
+	// allowlist goes stale unnoticed.
+	if len(sibling) == 0 {
+		t.Fatalf("未能从 %s 提取到任何条目：正则或表的书写形态已漂移，这条守卫已失效", interfaceBaselineReviewedPath)
+	}
+
+	for change := range reviewedFlagTypeChanges {
+		if _, ok := sibling[change]; !ok {
+			t.Errorf("%+v 只存在于 interfacesnapshot，缺在 %s", change, interfaceBaselineReviewedPath)
+		}
+	}
+	for change := range sibling {
+		if _, ok := reviewedFlagTypeChanges[change]; !ok {
+			t.Errorf("%+v 只存在于 %s，缺在 interfacesnapshot", change, interfaceBaselineReviewedPath)
+		}
+	}
+}
+
+func TestCrossPlatformCoverageCompareAllWithFlagMigrationsRejectsInvalidLifecycleTransitions(t *testing.T) {
+	empty := testEmptyFlagMigrationManifest()
+	pending := testFlagMigrationManifest(FlagMigrationPending)
+	consumed := testFlagMigrationManifest(FlagMigrationConsumed)
+	before := testFlagMigrationSnapshot(false, true)
+	after := testFlagMigrationSnapshot(true, true)
+
+	partial := testFlagMigrationSnapshot(false, true)
+	partial.Commands[1].LocalFlags[0].Hidden = true
+	partial.Commands[1].LocalFlags[0].Required = false
+	partial.Commands[1].LocalFlags[0].AliasOf = "message-id"
+
+	typeDrift := testFlagMigrationSnapshot(true, true)
+	typeDrift.Commands[1].LocalFlags[0].Type = "stringSlice"
+	noOptDrift := testFlagMigrationSnapshot(true, true)
+	noOptDrift.Commands[1].LocalFlags[0].NoOpt = ""
+	shorthandDrift := testFlagMigrationSnapshot(true, true)
+	shorthandDrift.Commands[1].LocalFlags[0].Shorthand = "x"
+
+	tests := []struct {
+		name      string
+		current   Snapshot
+		authority FlagMigrationManifest
+		candidate FlagMigrationManifest
+	}{
+		{
+			name:      "candidate-added pending record cannot approve its own surface change",
+			current:   after,
+			authority: empty,
+			candidate: pending,
+		},
+		{
+			name:      "candidate-added consumed record cannot approve its own surface change",
+			current:   after,
+			authority: empty,
+			candidate: consumed,
+		},
+		{
+			name:      "partial migration is rejected",
+			current:   partial,
+			authority: pending,
+			candidate: consumed,
+		},
+		{
+			name:      "pending record cannot be falsely consumed before the surface changes",
+			current:   before,
+			authority: pending,
+			candidate: consumed,
+		},
+		{
+			name:      "completed surface change must consume its pending record",
+			current:   after,
+			authority: pending,
+			candidate: pending,
+		},
+		{
+			name:      "legacy type drift is outside the approval",
+			current:   typeDrift,
+			authority: pending,
+			candidate: consumed,
+		},
+		{
+			name:      "legacy no-opt drift is outside the approval",
+			current:   noOptDrift,
+			authority: pending,
+			candidate: consumed,
+		},
+		{
+			name:      "legacy shorthand drift is outside the approval",
+			current:   shorthandDrift,
+			authority: pending,
+			candidate: consumed,
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			report, err := CompareAllWithFlagMigrations(
+				test.current,
+				map[string]Snapshot{
+					"merge-base": before,
+					"stable":     before,
+				},
+				test.authority,
+				test.candidate,
+			)
+			if err == nil {
+				t.Fatalf("invalid transition passed: %#v", report)
+			}
+		})
+	}
+}
+
+func TestCrossPlatformCoverageCompareAllWithFlagMigrationsAllowsApprovalCreationBeforeSurfaceChange(t *testing.T) {
+	before := testFlagMigrationSnapshot(false, true)
+	report, err := CompareAllWithFlagMigrations(
+		before,
+		map[string]Snapshot{
+			"merge-base": before,
+			"stable":     before,
+		},
+		testEmptyFlagMigrationManifest(),
+		testFlagMigrationManifest(FlagMigrationPending),
+	)
+	if err != nil {
+		t.Fatalf("pending approval creation was rejected before the surface changed: %v", err)
+	}
+	if !report.Compatible {
+		t.Fatalf("pending approval creation changed compatibility: %#v", report.Comparisons)
+	}
+}
+
+func TestCrossPlatformCoverageCompareAllWithFlagMigrationsKeepsUnrelatedBreakageBlocking(t *testing.T) {
+	before := testFlagMigrationSnapshot(false, true)
+	pending := testFlagMigrationManifest(FlagMigrationPending)
+	consumed := testFlagMigrationManifest(FlagMigrationConsumed)
+
+	tests := []struct {
+		name   string
+		mutate func(*Snapshot)
+		kind   string
+		flag   string
+	}{
+		{
+			name: "unrelated flag removal",
+			mutate: func(snapshot *Snapshot) {
+				snapshot.Commands[1].LocalFlags = snapshot.Commands[1].LocalFlags[:2]
+			},
+			kind: "flag_removed",
+			flag: "format",
+		},
+		{
+			name: "unrelated flag type change",
+			mutate: func(snapshot *Snapshot) {
+				snapshot.Commands[1].LocalFlags[2].Type = "stringSlice"
+			},
+			kind: "flag_type_changed",
+			flag: "format",
+		},
+		{
+			name: "unrelated flag no-opt change",
+			mutate: func(snapshot *Snapshot) {
+				snapshot.Commands[1].LocalFlags[2].NoOpt = ""
+			},
+			kind: "flag_no_opt_changed",
+			flag: "format",
+		},
+		{
+			name: "unrelated flag shorthand change",
+			mutate: func(snapshot *Snapshot) {
+				snapshot.Commands[1].LocalFlags[2].Shorthand = ""
+			},
+			kind: "flag_shorthand_changed",
+			flag: "format",
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			current := testFlagMigrationSnapshot(true, true)
+			test.mutate(&current)
+			report, err := CompareAllWithFlagMigrations(
+				current,
+				map[string]Snapshot{
+					"merge-base": before,
+					"stable":     before,
+				},
+				pending,
+				consumed,
+			)
+			if err != nil {
+				t.Fatalf("valid migration with unrelated breakage returned a lifecycle error: %v", err)
+			}
+			if report.Compatible {
+				t.Fatalf("unrelated breakage was hidden by migration approval: %#v", report)
+			}
+			for _, comparison := range report.Comparisons {
+				if !hasFlagChange(comparison.Blocking, test.kind, "dws chat send", test.flag) {
+					t.Fatalf("reference %q blocking=%#v, want %s for --%s", comparison.Reference, comparison.Blocking, test.kind, test.flag)
+				}
+			}
+		})
+	}
+}
+
+func TestCrossPlatformCoverageCompareAllWithFlagMigrationsRetainsConsumedReceiptUntilStableCatchesUp(t *testing.T) {
+	before := testFlagMigrationSnapshot(false, true)
+	after := testFlagMigrationSnapshot(true, true)
+	consumed := testFlagMigrationManifest(FlagMigrationConsumed)
+
+	report, err := CompareAllWithFlagMigrations(
+		after,
+		map[string]Snapshot{
+			"merge-base": after,
+			"stable":     before,
+		},
+		consumed,
+		consumed,
+	)
+	if err != nil {
+		t.Fatalf("consumed receipt was rejected while stable still needed it: %v", err)
+	}
+	if !report.Compatible {
+		t.Fatalf("stable comparison did not honor the consumed receipt: %#v", report.Comparisons)
+	}
+}
+
+func TestCrossPlatformCoverageCompareAllWithFlagMigrationsRequiresCleanupAfterAllReferencesCatchUp(t *testing.T) {
+	after := testFlagMigrationSnapshot(true, true)
+	consumed := testFlagMigrationManifest(FlagMigrationConsumed)
+	references := map[string]Snapshot{
+		"merge-base": after,
+		"stable":     after,
+	}
+
+	if report, err := CompareAllWithFlagMigrations(after, references, consumed, consumed); err == nil {
+		t.Fatalf("stale consumed receipt was accepted after every reference caught up: %#v", report)
+	}
+
+	report, err := CompareAllWithFlagMigrations(after, references, consumed, testEmptyFlagMigrationManifest())
+	if err != nil {
+		t.Fatalf("cleanup of stale consumed receipt was rejected: %v", err)
+	}
+	if !report.Compatible {
+		t.Fatalf("cleanup-only candidate was incompatible: %#v", report.Comparisons)
+	}
+}
+
 func testSnapshot(commands ...Command) Snapshot {
 	return Snapshot{
 		SchemaVersion: SchemaVersion,
@@ -348,6 +891,92 @@ func testFlag(name, flagType string, required bool) Flag {
 	return Flag{Name: name, Type: flagType, Required: required}
 }
 
+func testFlagMigrationSnapshot(after, includeUnrelated bool) Snapshot {
+	legacy := Flag{
+		Name:      "legacy-id",
+		Shorthand: "l",
+		Type:      "string",
+		NoOpt:     "auto",
+		Required:  true,
+	}
+	flags := []Flag{legacy}
+	if after {
+		legacy.Required = false
+		legacy.Hidden = true
+		legacy.AliasOf = "message-id"
+		flags = []Flag{
+			legacy,
+			{
+				Name:     "message-id",
+				Type:     "string",
+				Required: true,
+			},
+		}
+	}
+	if includeUnrelated {
+		flags = append(flags, Flag{
+			Name:      "format",
+			Shorthand: "f",
+			Type:      "string",
+			NoOpt:     "json",
+		})
+	}
+	return testSnapshot(
+		testCommand("dws"),
+		testCommand("dws chat send", flags...),
+	)
+}
+
+func testFlagMigrationManifest(state string) FlagMigrationManifest {
+	return FlagMigrationManifest{
+		Version: FlagMigrationManifestVersion,
+		Migrations: []FlagMigration{
+			{
+				Command: "dws chat send",
+				Legacy: FlagMigrationSide{
+					Name: "legacy-id",
+					Before: FlagMigrationState{
+						Present:   true,
+						Type:      "string",
+						Required:  true,
+						Shorthand: "l",
+						NoOpt:     "auto",
+						Scope:     "local",
+					},
+					After: FlagMigrationState{
+						Present:   true,
+						Type:      "string",
+						Hidden:    true,
+						Shorthand: "l",
+						NoOpt:     "auto",
+						Scope:     "local",
+						AliasOf:   "message-id",
+					},
+				},
+				Canonical: FlagMigrationSide{
+					Name:   "message-id",
+					Before: FlagMigrationState{},
+					After: FlagMigrationState{
+						Present:  true,
+						Type:     "string",
+						Required: true,
+						Scope:    "local",
+					},
+				},
+				State:  state,
+				Reason: "rename the public flag while preserving the executable legacy alias",
+			},
+		},
+	}
+}
+
+func testEmptyFlagMigrationManifest() FlagMigrationManifest {
+	return FlagMigrationManifest{
+		Version:    FlagMigrationManifestVersion,
+		Migrations: []FlagMigration{},
+	}
+}
+
 func hasChangeKind(changes []Change, kind string) bool {
 	for _, change := range changes {
 		if change.Kind == kind {

internal/interfacesnapshot/migrations.go

@@ -0,0 +1,696 @@
+// Copyright 2026 Alibaba Group
+// Licensed under the Apache License, Version 2.0 (the "License");
+
+package interfacesnapshot
+
+import (
+	"bytes"
+	"encoding/json"
+	"fmt"
+	"io"
+	"reflect"
+	"strings"
+)
+
+const FlagMigrationManifestVersion = 1
+
+const (
+	FlagMigrationPending  = "pending"
+	FlagMigrationConsumed = "consumed"
+)
+
+type FlagMigrationManifest struct {
+	Version    int             `json:"version"`
+	Migrations []FlagMigration `json:"migrations"`
+}
+
+type FlagMigration struct {
+	Command   string            `json:"command"`
+	Legacy    FlagMigrationSide `json:"legacy"`
+	Canonical FlagMigrationSide `json:"canonical"`
+	State     string            `json:"state"`
+	Reason    string            `json:"reason"`
+}
+
+type FlagMigrationSide struct {
+	Name   string             `json:"name"`
+	Before FlagMigrationState `json:"before"`
+	After  FlagMigrationState `json:"after"`
+}
+
+type FlagMigrationState struct {
+	Present   bool   `json:"present"`
+	Type      string `json:"type,omitempty"`
+	Required  bool   `json:"required,omitempty"`
+	Hidden    bool   `json:"hidden,omitempty"`
+	Shorthand string `json:"shorthand,omitempty"`
+	NoOpt     string `json:"no_opt,omitempty"`
+	Scope     string `json:"scope,omitempty"`
+	AliasOf   string `json:"alias_of,omitempty"`
+}
+
+func ReadFlagMigrationManifest(r io.Reader) (FlagMigrationManifest, error) {
+	data, err := io.ReadAll(r)
+	if err != nil {
+		return FlagMigrationManifest{}, fmt.Errorf("read flag migration manifest: %w", err)
+	}
+	var manifest FlagMigrationManifest
+	decoder := json.NewDecoder(bytes.NewReader(data))
+	decoder.DisallowUnknownFields()
+	if err := decoder.Decode(&manifest); err != nil {
+		return FlagMigrationManifest{}, err
+	}
+	if err := decoder.Decode(&struct{}{}); err != io.EOF {
+		if err == nil {
+			return FlagMigrationManifest{}, fmt.Errorf("flag migration manifest contains trailing multiple JSON values")
+		}
+		return FlagMigrationManifest{}, fmt.Errorf("read trailing flag migration manifest data: %w", err)
+	}
+	if err := validateFlagMigrationJSONSchema(data); err != nil {
+		return FlagMigrationManifest{}, err
+	}
+	if err := manifest.Validate(); err != nil {
+		return FlagMigrationManifest{}, err
+	}
+	return manifest, nil
+}
+
+func validateFlagMigrationJSONSchema(data []byte) error {
+	decoder := json.NewDecoder(bytes.NewReader(data))
+	decoder.UseNumber()
+	if err := validateMigrationJSONValue(decoder, "$", reflect.TypeOf(FlagMigrationManifest{})); err != nil {
+		return err
+	}
+	if token, err := decoder.Token(); err != io.EOF {
+		if err == nil {
+			return fmt.Errorf("flag migration manifest contains trailing JSON value %v", token)
+		}
+		return fmt.Errorf("read trailing flag migration manifest data: %w", err)
+	}
+	return nil
+}
+
+func validateMigrationJSONValue(decoder *json.Decoder, path string, schema reflect.Type) error {
+	token, err := decoder.Token()
+	if err != nil {
+		return fmt.Errorf("read flag migration manifest value at %s: %w", path, err)
+	}
+
+	switch schema.Kind() {
+	case reflect.Struct:
+		if delimiter, ok := token.(json.Delim); !ok || delimiter != '{' {
+			return migrationJSONTypeError(path, schema, token)
+		}
+		fields := migrationJSONFields(schema)
+		seen := make(map[string]bool, len(fields))
+		for decoder.More() {
+			keyToken, keyErr := decoder.Token()
+			if keyErr != nil {
+				return fmt.Errorf("read flag migration manifest field at %s: %w", path, keyErr)
+			}
+			// encoding/json guarantees object member names are string tokens.
+			key := keyToken.(string)
+			fieldSchema, exists := fields[key]
+			if !exists {
+				for canonical := range fields {
+					if strings.EqualFold(key, canonical) {
+						return fmt.Errorf(
+							"flag migration manifest contains non-canonical field %q at %s (want %q)",
+							key,
+							path,
+							canonical,
+						)
+					}
+				}
+				return fmt.Errorf("flag migration manifest contains unknown field %q at %s", key, path)
+			}
+			if seen[key] {
+				return fmt.Errorf("flag migration manifest contains duplicate field %q at %s", key, path)
+			}
+			seen[key] = true
+			if err := validateMigrationJSONValue(decoder, path+"."+key, fieldSchema); err != nil {
+				return err
+			}
+		}
+		if _, closeErr := decoder.Token(); closeErr != nil {
+			return fmt.Errorf("close flag migration manifest object at %s: %w", path, closeErr)
+		}
+		return nil
+	case reflect.Slice:
+		if delimiter, ok := token.(json.Delim); !ok || delimiter != '[' {
+			return migrationJSONTypeError(path, schema, token)
+		}
+		for index := 0; decoder.More(); index++ {
+			if err := validateMigrationJSONValue(decoder, fmt.Sprintf("%s[%d]", path, index), schema.Elem()); err != nil {
+				return err
+			}
+		}
+		if _, closeErr := decoder.Token(); closeErr != nil {
+			return fmt.Errorf("close flag migration manifest array at %s: %w", path, closeErr)
+		}
+		return nil
+	case reflect.String:
+		if _, ok := token.(string); !ok {
+			return migrationJSONTypeError(path, schema, token)
+		}
+		return nil
+	case reflect.Int:
+		if _, ok := token.(json.Number); !ok {
+			return migrationJSONTypeError(path, schema, token)
+		}
+		return nil
+	case reflect.Bool:
+		if _, ok := token.(bool); !ok {
+			return migrationJSONTypeError(path, schema, token)
+		}
+		return nil
+	default:
+		return fmt.Errorf("flag migration manifest value at %s has unsupported Go schema type %s", path, schema)
+	}
+}
+
+func migrationJSONFields(schema reflect.Type) map[string]reflect.Type {
+	fields := make(map[string]reflect.Type, schema.NumField())
+	for index := 0; index < schema.NumField(); index++ {
+		field := schema.Field(index)
+		name := field.Tag.Get("json")
+		if comma := strings.IndexByte(name, ','); comma >= 0 {
+			name = name[:comma]
+		}
+		if name == "-" {
+			continue
+		}
+		if name == "" {
+			name = field.Name
+		}
+		fields[name] = field.Type
+	}
+	return fields
+}
+
+func migrationJSONTypeError(path string, want reflect.Type, token json.Token) error {
+	return fmt.Errorf(
+		"flag migration manifest value at %s must be %s, got %s",
+		path,
+		migrationJSONKindDescription(want),
+		migrationJSONTokenDescription(token),
+	)
+}
+
+func migrationJSONKindDescription(schema reflect.Type) string {
+	switch schema.Kind() {
+	case reflect.Struct:
+		return "an object"
+	case reflect.Slice:
+		return "an array"
+	case reflect.String:
+		return "a string"
+	case reflect.Int:
+		return "a number"
+	case reflect.Bool:
+		return "a boolean"
+	default:
+		return "the declared JSON type"
+	}
+}
+
+func migrationJSONTokenDescription(token json.Token) string {
+	if token == nil {
+		return "null"
+	}
+	if delimiter, ok := token.(json.Delim); ok {
+		switch delimiter {
+		case '{':
+			return "object"
+		case '[':
+			return "array"
+		default:
+			return fmt.Sprintf("delimiter %q", delimiter)
+		}
+	}
+	switch token.(type) {
+	case string:
+		return "string"
+	case json.Number, float64:
+		return "number"
+	case bool:
+		return "boolean"
+	default:
+		return fmt.Sprintf("%T", token)
+	}
+}
+
+func (m FlagMigrationManifest) Validate() error {
+	if m.Version != FlagMigrationManifestVersion {
+		return fmt.Errorf(
+			"unsupported flag migration manifest version %d (want %d)",
+			m.Version,
+			FlagMigrationManifestVersion,
+		)
+	}
+	if m.Migrations == nil {
+		return fmt.Errorf("flag migration manifest migrations must be an array")
+	}
+	seen := make(map[string]bool, len(m.Migrations))
+	legacyTargets := make(map[string]string, len(m.Migrations))
+	for index, migration := range m.Migrations {
+		if err := migration.validate(); err != nil {
+			return fmt.Errorf("flag migration %d: %w", index, err)
+		}
+		key := migration.key()
+		if seen[key] {
+			return fmt.Errorf("flag migration %d duplicates %s", index, key)
+		}
+		seen[key] = true
+		legacyKey := migration.Command + "\x00" + migration.Legacy.Name
+		if canonical, exists := legacyTargets[legacyKey]; exists && canonical != migration.Canonical.Name {
+			return fmt.Errorf(
+				"flag migration %d maps %s --%s to both --%s and --%s",
+				index,
+				migration.Command,
+				migration.Legacy.Name,
+				canonical,
+				migration.Canonical.Name,
+			)
+		}
+		legacyTargets[legacyKey] = migration.Canonical.Name
+	}
+	return nil
+}
+
+func (m FlagMigration) validate() error {
+	if !isExactCommandPath(m.Command) {
+		return fmt.Errorf("command must be an exact command path rooted at dws: %q", m.Command)
+	}
+	if !isExactFlagName(m.Legacy.Name) {
+		return fmt.Errorf("legacy name must be an exact legacy flag: %q", m.Legacy.Name)
+	}
+	if !isExactFlagName(m.Canonical.Name) {
+		return fmt.Errorf("canonical name must be an exact canonical flag: %q", m.Canonical.Name)
+	}
+	if m.Legacy.Name == m.Canonical.Name {
+		return fmt.Errorf("legacy and canonical flags must differ: --%s", m.Legacy.Name)
+	}
+	if strings.TrimSpace(m.Reason) == "" {
+		return fmt.Errorf("migration must include a non-empty reason")
+	}
+	if m.Reason != strings.TrimSpace(m.Reason) {
+		return fmt.Errorf("migration reason must be trimmed")
+	}
+	if m.State != FlagMigrationPending && m.State != FlagMigrationConsumed {
+		return fmt.Errorf("invalid state %q", m.State)
+	}
+	if err := m.Legacy.Before.validate("legacy before"); err != nil {
+		return err
+	}
+	if err := m.Legacy.After.validate("legacy after"); err != nil {
+		return err
+	}
+	if err := m.Canonical.Before.validate("canonical before"); err != nil {
+		return err
+	}
+	if err := m.Canonical.After.validate("canonical after"); err != nil {
+		return err
+	}
+	if !m.Legacy.Before.Present || !m.Legacy.After.Present {
+		return fmt.Errorf("legacy flag must remain present before and after migration")
+	}
+	if m.Legacy.Before.Hidden || !m.Legacy.After.Hidden {
+		return fmt.Errorf("legacy flag must migrate exactly from visible to hidden")
+	}
+	if !m.Canonical.After.Present {
+		return fmt.Errorf("canonical flag must be present after migration")
+	}
+	if m.Canonical.After.Hidden {
+		return fmt.Errorf("canonical flag must remain visible")
+	}
+	if !m.Canonical.After.Required {
+		return fmt.Errorf("canonical flag must be required after migration")
+	}
+	if m.Canonical.Before.Present && m.Canonical.Before.Required {
+		return fmt.Errorf("canonical flag must be absent or optional before migration")
+	}
+	if m.Legacy.After.AliasOf != m.Canonical.Name {
+		return fmt.Errorf(
+			"legacy flag after state must declare alias_of %q",
+			m.Canonical.Name,
+		)
+	}
+	if m.Legacy.Before.AliasOf != "" || m.Canonical.Before.AliasOf != "" || m.Canonical.After.AliasOf != "" {
+		return fmt.Errorf("alias_of is only valid on the legacy after state")
+	}
+	if m.Canonical.Before.Present {
+		if m.Canonical.Before.Type != m.Canonical.After.Type {
+			return fmt.Errorf("canonical flag type must remain unchanged")
+		}
+		if m.Canonical.Before.Shorthand != m.Canonical.After.Shorthand {
+			return fmt.Errorf("canonical flag shorthand must remain unchanged")
+		}
+		if m.Canonical.Before.NoOpt != m.Canonical.After.NoOpt {
+			return fmt.Errorf("canonical flag no_opt must remain unchanged")
+		}
+		if m.Canonical.Before.Scope != m.Canonical.After.Scope {
+			return fmt.Errorf("canonical flag scope must remain unchanged")
+		}
+	}
+	if m.Legacy.Before.Type != m.Legacy.After.Type || m.Legacy.After.Type != m.Canonical.After.Type {
+		return fmt.Errorf("legacy and canonical flag types must match exactly")
+	}
+	if m.Legacy.Before.Shorthand != m.Legacy.After.Shorthand {
+		return fmt.Errorf("legacy flag shorthand must remain unchanged")
+	}
+	if m.Legacy.Before.NoOpt != m.Legacy.After.NoOpt {
+		return fmt.Errorf("legacy flag no_opt must remain unchanged")
+	}
+	if m.Legacy.Before.Scope != m.Legacy.After.Scope {
+		return fmt.Errorf("legacy flag scope must remain unchanged")
+	}
+	return nil
+}
+
+func (s FlagMigrationState) validate(label string) error {
+	if !s.Present {
+		if s.Type != "" || s.Required || s.Hidden || s.Shorthand != "" || s.NoOpt != "" || s.Scope != "" || s.AliasOf != "" {
+			return fmt.Errorf("%s absent state must not declare flag attributes", label)
+		}
+		return nil
+	}
+	if strings.TrimSpace(s.Type) == "" {
+		return fmt.Errorf("%s present state requires type", label)
+	}
+	if s.Scope != "local" && s.Scope != "inherited" {
+		return fmt.Errorf("%s present state has invalid scope %q", label, s.Scope)
+	}
+	return nil
+}
+
+func (m FlagMigration) key() string {
+	return m.Command + "\x00" + m.Legacy.Name + "\x00" + m.Canonical.Name
+}
+
+func isExactCommandPath(path string) bool {
+	return path != "" &&
+		path == strings.TrimSpace(path) &&
+		strings.Join(strings.Fields(path), " ") == path &&
+		(path == "dws" || strings.HasPrefix(path, "dws ")) &&
+		!strings.ContainsAny(path, "*?[]{}")
+}
+
+func isExactFlagName(name string) bool {
+	return name != "" &&
+		name == strings.TrimSpace(name) &&
+		!strings.HasPrefix(name, "-") &&
+		!strings.ContainsAny(name, "*?[]{} /\t\r\n")
+}
+
+// CompareAllWithFlagMigrations applies the ordinary compatibility policy and
+// then consumes only exact, merge-base-owned flag migrations. Candidate-owned
+// records participate in the lifecycle check, but never authorize their own
+// interface change.
+func CompareAllWithFlagMigrations(
+	current Snapshot,
+	references map[string]Snapshot,
+	authority FlagMigrationManifest,
+	candidate FlagMigrationManifest,
+) (Report, error) {
+	authorizations, err := AuthorizeFlagMigrations(current, references, authority, candidate)
+	if err != nil {
+		return Report{}, err
+	}
+	report := CompareAll(current, references)
+	if len(authorizations) == 0 {
+		return report, nil
+	}
+
+	for index := range report.Comparisons {
+		comparison := &report.Comparisons[index]
+		reference := references[comparison.Reference]
+		filtered := comparison.Blocking[:0]
+		for _, change := range comparison.Blocking {
+			if flagMigrationAuthorizesChange(current, reference, change, authorizations) {
+				continue
+			}
+			filtered = append(filtered, change)
+		}
+		comparison.Blocking = filtered
+		comparison.Compatible = len(filtered) == 0
+	}
+	report.Compatible = true
+	for _, comparison := range report.Comparisons {
+		if !comparison.Compatible {
+			report.Compatible = false
+			break
+		}
+	}
+	return report, nil
+}
+
+// AuthorizeFlagMigrations validates both snapshots and manifests, enforces the
+// base-owned migration lifecycle, and returns only approvals that may authorize
+// the current exact interface transition. A non-empty lifecycle requires both a
+// main/merge-base authority reference and a stable reference. Candidate-added
+// records never appear in the returned authorization set.
+func AuthorizeFlagMigrations(
+	current Snapshot,
+	references map[string]Snapshot,
+	authority FlagMigrationManifest,
+	candidate FlagMigrationManifest,
+) ([]FlagMigration, error) {
+	if err := current.Validate(); err != nil {
+		return nil, fmt.Errorf("validate current interface snapshot: %w", err)
+	}
+	for label, snapshot := range references {
+		if err := snapshot.Validate(); err != nil {
+			return nil, fmt.Errorf("validate %s interface snapshot: %w", label, err)
+		}
+	}
+	if err := authority.Validate(); err != nil {
+		return nil, fmt.Errorf("validate approved flag migrations: %w", err)
+	}
+	if err := candidate.Validate(); err != nil {
+		return nil, fmt.Errorf("validate candidate flag migrations: %w", err)
+	}
+	return evaluateFlagMigrationLifecycle(current, references, authority, candidate)
+}
+
+type flagMigrationPhase string
+
+const (
+	flagMigrationBefore  flagMigrationPhase = "before"
+	flagMigrationAfter   flagMigrationPhase = "after"
+	flagMigrationPartial flagMigrationPhase = "partial"
+)
+
+func evaluateFlagMigrationLifecycle(
+	current Snapshot,
+	references map[string]Snapshot,
+	authority FlagMigrationManifest,
+	candidate FlagMigrationManifest,
+) ([]FlagMigration, error) {
+	if len(authority.Migrations) == 0 && len(candidate.Migrations) == 0 {
+		return nil, nil
+	}
+	if _, ok := references["stable"]; !ok {
+		return nil, fmt.Errorf("flag migration lifecycle requires a stable reference")
+	}
+	mergeBase, label, ok := flagMigrationAuthoritySnapshot(references)
+	if !ok {
+		return nil, fmt.Errorf("flag migration lifecycle requires a main or merge-base reference")
+	}
+
+	authorityByKey := flagMigrationIndex(authority)
+	candidateByKey := flagMigrationIndex(candidate)
+	authorizations := make([]FlagMigration, 0, len(authority.Migrations))
+
+	for _, approved := range authority.Migrations {
+		basePhase := matchFlagMigrationPhase(mergeBase, approved)
+		wantBasePhase := flagMigrationBefore
+		if approved.State == FlagMigrationConsumed {
+			wantBasePhase = flagMigrationAfter
+		}
+		if basePhase != wantBasePhase {
+			return nil, fmt.Errorf(
+				"approved flag migration %s is %s in %s, want exact %s state for %s",
+				approved.displayKey(),
+				basePhase,
+				label,
+				wantBasePhase,
+				approved.State,
+			)
+		}
+
+		proposed, exists := candidateByKey[approved.key()]
+		if exists && !sameFlagMigrationApproval(approved, proposed) {
+			return nil, fmt.Errorf("candidate modified base-owned flag migration %s", approved.displayKey())
+		}
+
+		currentPhase := matchFlagMigrationPhase(current, approved)
+		switch approved.State {
+		case FlagMigrationPending:
+			if !exists {
+				return nil, fmt.Errorf("candidate removed pending flag migration %s", approved.displayKey())
+			}
+			switch currentPhase {
+			case flagMigrationBefore:
+				if proposed.State != FlagMigrationPending {
+					return nil, fmt.Errorf("candidate falsely consumed unchanged flag migration %s", approved.displayKey())
+				}
+			case flagMigrationAfter:
+				if proposed.State != FlagMigrationConsumed {
+					return nil, fmt.Errorf("candidate completed flag migration %s without marking it consumed", approved.displayKey())
+				}
+				authorizations = append(authorizations, approved)
+			default:
+				return nil, fmt.Errorf("candidate partially applied flag migration %s", approved.displayKey())
+			}
+		case FlagMigrationConsumed:
+			if currentPhase != flagMigrationAfter {
+				return nil, fmt.Errorf("candidate drifted from consumed flag migration %s", approved.displayKey())
+			}
+			allReferencesAfter := true
+			for _, reference := range references {
+				if matchFlagMigrationPhase(reference, approved) != flagMigrationAfter {
+					allReferencesAfter = false
+					break
+				}
+			}
+			if allReferencesAfter {
+				if exists {
+					return nil, fmt.Errorf("consumed flag migration %s is stale after all references reached the after state", approved.displayKey())
+				}
+				continue
+			}
+			if !exists {
+				return nil, fmt.Errorf("candidate removed consumed flag migration %s before every reference reached the after state", approved.displayKey())
+			}
+			if proposed.State != FlagMigrationConsumed {
+				return nil, fmt.Errorf("candidate changed consumed flag migration %s back to pending", approved.displayKey())
+			}
+			authorizations = append(authorizations, approved)
+		}
+	}
+
+	for _, proposed := range candidate.Migrations {
+		if _, exists := authorityByKey[proposed.key()]; exists {
+			continue
+		}
+		if proposed.State != FlagMigrationPending {
+			return nil, fmt.Errorf("candidate-added flag migration %s must start pending", proposed.displayKey())
+		}
+		if matchFlagMigrationPhase(mergeBase, proposed) != flagMigrationBefore {
+			return nil, fmt.Errorf("candidate-added flag migration %s does not match the merge-base before state", proposed.displayKey())
+		}
+		if matchFlagMigrationPhase(current, proposed) != flagMigrationBefore {
+			return nil, fmt.Errorf("candidate-added flag migration %s cannot authorize its own interface change", proposed.displayKey())
+		}
+	}
+
+	return authorizations, nil
+}
+
+func flagMigrationAuthoritySnapshot(references map[string]Snapshot) (Snapshot, string, bool) {
+	for _, label := range []string{"merge-base", "main"} {
+		if snapshot, ok := references[label]; ok {
+			return snapshot, label, true
+		}
+	}
+	return Snapshot{}, "", false
+}
+
+func flagMigrationIndex(manifest FlagMigrationManifest) map[string]FlagMigration {
+	index := make(map[string]FlagMigration, len(manifest.Migrations))
+	for _, migration := range manifest.Migrations {
+		index[migration.key()] = migration
+	}
+	return index
+}
+
+func sameFlagMigrationApproval(left, right FlagMigration) bool {
+	left.State = ""
+	right.State = ""
+	return reflect.DeepEqual(left, right)
+}
+
+func matchFlagMigrationPhase(snapshot Snapshot, migration FlagMigration) flagMigrationPhase {
+	command, exists := commandIndex(snapshot)[migration.Command]
+	if !exists {
+		return flagMigrationPartial
+	}
+	legacy := flagMigrationStateForCommand(command, migration.Legacy.Name)
+	canonical := flagMigrationStateForCommand(command, migration.Canonical.Name)
+	if legacy == migration.Legacy.Before && canonical == migration.Canonical.Before {
+		return flagMigrationBefore
+	}
+	if legacy == migration.Legacy.After && canonical == migration.Canonical.After {
+		return flagMigrationAfter
+	}
+	return flagMigrationPartial
+}
+
+func flagMigrationStateForCommand(command Command, name string) FlagMigrationState {
+	for _, flag := range command.LocalFlags {
+		if flag.Name == name {
+			return flagMigrationState(flag, "local")
+		}
+	}
+	for _, flag := range command.InheritedFlags {
+		if flag.Name == name {
+			return flagMigrationState(flag, "inherited")
+		}
+	}
+	return FlagMigrationState{}
+}
+
+func flagMigrationState(flag Flag, scope string) FlagMigrationState {
+	return FlagMigrationState{
+		Present:   true,
+		Type:      flag.Type,
+		Required:  flag.Required,
+		Hidden:    flag.Hidden,
+		Shorthand: flag.Shorthand,
+		NoOpt:     flag.NoOpt,
+		Scope:     scope,
+		AliasOf:   flag.AliasOf,
+	}
+}
+
+func flagMigrationAuthorizesChange(
+	current Snapshot,
+	reference Snapshot,
+	change Change,
+	authorizations []FlagMigration,
+) bool {
+	canonicalPath := acceptedPathIndex(reference)[change.Path]
+	if canonicalPath == "" {
+		canonicalPath = change.Path
+	}
+	for _, migration := range authorizations {
+		if canonicalPath != migration.Command ||
+			matchFlagMigrationPhase(reference, migration) != flagMigrationBefore ||
+			matchFlagMigrationPhase(current, migration) != flagMigrationAfter {
+			continue
+		}
+		if change.Flag == migration.Legacy.Name && change.Kind == "flag_became_hidden" {
+			return true
+		}
+		if change.Flag != migration.Canonical.Name {
+			continue
+		}
+		if !migration.Canonical.Before.Present &&
+			migration.Canonical.After.Required &&
+			change.Kind == "required_flag_added" {
+			return true
+		}
+		if migration.Canonical.Before.Present &&
+			!migration.Canonical.Before.Required &&
+			migration.Canonical.After.Required &&
+			change.Kind == "flag_became_required" {
+			return true
+		}
+	}
+	return false
+}
+
+func (m FlagMigration) displayKey() string {
+	return fmt.Sprintf("%q --%s -> --%s", m.Command, m.Legacy.Name, m.Canonical.Name)
+}

internal/interfacesnapshot/migrations_coverage_test.go

@@ -0,0 +1,966 @@
+// Copyright 2026 Alibaba Group
+// Licensed under the Apache License, Version 2.0 (the "License");
+
+package interfacesnapshot
+
+import (
+	"encoding/json"
+	"reflect"
+	"strings"
+	"testing"
+)
+
+func TestCrossPlatformCoverageFlagMigrationStrictJSONSchemaErrors(t *testing.T) {
+	tests := []struct {
+		name    string
+		input   string
+		wantErr string
+	}{
+		{
+			name:    "schema validator rejects a trailing value",
+			input:   coverageManifestJSON() + ` {}`,
+			wantErr: "trailing JSON value",
+		},
+		{
+			name:    "schema validator reports malformed trailing data",
+			input:   coverageManifestJSON() + ` ?`,
+			wantErr: "read trailing flag migration manifest data",
+		},
+		{
+			name:    "schema validator reports an empty document",
+			input:   "",
+			wantErr: "read flag migration manifest value at $",
+		},
+		{
+			name:    "schema validator rejects an unknown field",
+			input:   `{"version":1,"migrations":[],"unexpected":true}`,
+			wantErr: `unknown field "unexpected"`,
+		},
+		{
+			name:    "schema validator reports a malformed field token",
+			input:   `{"`,
+			wantErr: "flag migration manifest",
+		},
+		{
+			name:    "schema validator reports an unclosed object",
+			input:   `{"version":1`,
+			wantErr: "close flag migration manifest object",
+		},
+		{
+			name:    "schema validator reports an unclosed array",
+			input:   `{"version":1,"migrations":[`,
+			wantErr: "close flag migration manifest array",
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			err := validateFlagMigrationJSONSchema([]byte(test.input))
+			if err == nil || !strings.Contains(err.Error(), test.wantErr) {
+				t.Fatalf("validateFlagMigrationJSONSchema() error = %v, want %q", err, test.wantErr)
+			}
+		})
+	}
+}
+
+func TestCrossPlatformCoverageFlagMigrationStrictJSONTypeDiagnostics(t *testing.T) {
+	tests := []struct {
+		name    string
+		input   string
+		schema  reflect.Type
+		wantErr string
+	}{
+		{
+			name:    "manifest must be an object",
+			input:   `[]`,
+			schema:  reflect.TypeOf(FlagMigrationManifest{}),
+			wantErr: "must be an object, got array",
+		},
+		{
+			name:    "migrations must be an array",
+			input:   `{}`,
+			schema:  reflect.TypeOf([]FlagMigration{}),
+			wantErr: "must be an array, got object",
+		},
+		{
+			name:    "integer rejects string",
+			input:   `"one"`,
+			schema:  reflect.TypeOf(int(0)),
+			wantErr: "must be a number, got string",
+		},
+		{
+			name:    "string rejects number",
+			input:   `1`,
+			schema:  reflect.TypeOf(""),
+			wantErr: "must be a string, got number",
+		},
+		{
+			name:    "string rejects boolean",
+			input:   `false`,
+			schema:  reflect.TypeOf(""),
+			wantErr: "must be a string, got boolean",
+		},
+		{
+			name:    "unsupported schema fails closed",
+			input:   `1`,
+			schema:  reflect.TypeOf(float64(0)),
+			wantErr: "unsupported Go schema type float64",
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			decoder := json.NewDecoder(strings.NewReader(test.input))
+			decoder.UseNumber()
+			err := validateMigrationJSONValue(decoder, "$", test.schema)
+			if err == nil || !strings.Contains(err.Error(), test.wantErr) {
+				t.Fatalf("validateMigrationJSONValue() error = %v, want %q", err, test.wantErr)
+			}
+		})
+	}
+}
+
+func TestCrossPlatformCoverageFlagMigrationStrictJSONSchemaHelpers(t *testing.T) {
+	type fieldFixture struct {
+		DefaultName string
+		Ignored     string `json:"-"`
+		Tagged      string `json:"tagged,omitempty"`
+	}
+	fields := migrationJSONFields(reflect.TypeOf(fieldFixture{}))
+	if _, exists := fields["Ignored"]; exists {
+		t.Fatal("migrationJSONFields() retained a json:- field")
+	}
+	if fields["DefaultName"] != reflect.TypeOf("") || fields["tagged"] != reflect.TypeOf("") {
+		t.Fatalf("migrationJSONFields() = %#v, want default and tagged string fields", fields)
+	}
+
+	if got := migrationJSONKindDescription(reflect.TypeOf(float64(0))); got != "the declared JSON type" {
+		t.Fatalf("migrationJSONKindDescription(float64) = %q", got)
+	}
+	for _, test := range []struct {
+		name  string
+		token json.Token
+		want  string
+	}{
+		{name: "closing delimiter", token: json.Delim('}'), want: `delimiter "}"`},
+		{name: "unsupported token", token: struct{}{}, want: "struct {}"},
+	} {
+		t.Run(test.name, func(t *testing.T) {
+			if got := migrationJSONTokenDescription(test.token); got != test.want {
+				t.Fatalf("migrationJSONTokenDescription() = %q, want %q", got, test.want)
+			}
+		})
+	}
+}
+
+func TestCrossPlatformCoverageFlagMigrationManifestParserEdges(t *testing.T) {
+	tests := []struct {
+		name    string
+		input   string
+		wantErr string
+	}{
+		{
+			name:    "initial malformed JSON",
+			input:   `{`,
+			wantErr: "unexpected EOF",
+		},
+		{
+			name:    "multiple JSON values",
+			input:   coverageManifestJSON() + ` {}`,
+			wantErr: "trailing multiple JSON values",
+		},
+		{
+			name:    "malformed trailing JSON",
+			input:   coverageManifestJSON() + ` {`,
+			wantErr: "read trailing flag migration manifest data",
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			_, err := ReadFlagMigrationManifest(strings.NewReader(test.input))
+			if err == nil || !strings.Contains(err.Error(), test.wantErr) {
+				t.Fatalf("ReadFlagMigrationManifest() error = %v, want %q", err, test.wantErr)
+			}
+		})
+	}
+}
+
+func TestCrossPlatformCoverageFlagMigrationManifestRejectsEveryContractDrift(t *testing.T) {
+	optionalCanonical := func() FlagMigrationManifest {
+		manifest := coverageManifest(FlagMigrationPending)
+		manifest.Migrations[0].Canonical.Before = FlagMigrationState{
+			Present: true,
+			Type:    "string",
+			Scope:   "local",
+		}
+		return manifest
+	}
+
+	tests := []struct {
+		name    string
+		make    func() FlagMigrationManifest
+		mutate  func(*FlagMigrationManifest)
+		wantErr string
+	}{
+		{
+			name:    "canonical flag name is exact",
+			make:    func() FlagMigrationManifest { return coverageManifest(FlagMigrationPending) },
+			mutate:  func(m *FlagMigrationManifest) { m.Migrations[0].Canonical.Name = "--message-id" },
+			wantErr: "canonical name must be an exact canonical flag",
+		},
+		{
+			name:    "legacy and canonical names differ",
+			make:    func() FlagMigrationManifest { return coverageManifest(FlagMigrationPending) },
+			mutate:  func(m *FlagMigrationManifest) { m.Migrations[0].Canonical.Name = m.Migrations[0].Legacy.Name },
+			wantErr: "legacy and canonical flags must differ",
+		},
+		{
+			name:    "legacy before state validates",
+			make:    func() FlagMigrationManifest { return coverageManifest(FlagMigrationPending) },
+			mutate:  func(m *FlagMigrationManifest) { m.Migrations[0].Legacy.Before.Scope = "global" },
+			wantErr: "legacy before present state has invalid scope",
+		},
+		{
+			name:    "legacy after state validates",
+			make:    func() FlagMigrationManifest { return coverageManifest(FlagMigrationPending) },
+			mutate:  func(m *FlagMigrationManifest) { m.Migrations[0].Legacy.After.Type = "" },
+			wantErr: "legacy after present state requires type",
+		},
+		{
+			name:    "canonical before state validates",
+			make:    optionalCanonical,
+			mutate:  func(m *FlagMigrationManifest) { m.Migrations[0].Canonical.Before.Type = "" },
+			wantErr: "canonical before present state requires type",
+		},
+		{
+			name:    "canonical after state validates",
+			make:    func() FlagMigrationManifest { return coverageManifest(FlagMigrationPending) },
+			mutate:  func(m *FlagMigrationManifest) { m.Migrations[0].Canonical.After.Scope = "" },
+			wantErr: "canonical after present state has invalid scope",
+		},
+		{
+			name:    "legacy before remains present",
+			make:    func() FlagMigrationManifest { return coverageManifest(FlagMigrationPending) },
+			mutate:  func(m *FlagMigrationManifest) { m.Migrations[0].Legacy.Before = FlagMigrationState{} },
+			wantErr: "legacy flag must remain present before and after migration",
+		},
+		{
+			name:    "legacy after remains present",
+			make:    func() FlagMigrationManifest { return coverageManifest(FlagMigrationPending) },
+			mutate:  func(m *FlagMigrationManifest) { m.Migrations[0].Legacy.After = FlagMigrationState{} },
+			wantErr: "legacy flag must remain present before and after migration",
+		},
+		{
+			name:    "legacy starts visible",
+			make:    func() FlagMigrationManifest { return coverageManifest(FlagMigrationPending) },
+			mutate:  func(m *FlagMigrationManifest) { m.Migrations[0].Legacy.Before.Hidden = true },
+			wantErr: "legacy flag must migrate exactly from visible to hidden",
+		},
+		{
+			name:    "legacy ends hidden",
+			make:    func() FlagMigrationManifest { return coverageManifest(FlagMigrationPending) },
+			mutate:  func(m *FlagMigrationManifest) { m.Migrations[0].Legacy.After.Hidden = false },
+			wantErr: "legacy flag must migrate exactly from visible to hidden",
+		},
+		{
+			name:    "canonical after remains present",
+			make:    func() FlagMigrationManifest { return coverageManifest(FlagMigrationPending) },
+			mutate:  func(m *FlagMigrationManifest) { m.Migrations[0].Canonical.After = FlagMigrationState{} },
+			wantErr: "canonical flag must be present after migration",
+		},
+		{
+			name:    "legacy after declares alias target",
+			make:    func() FlagMigrationManifest { return coverageManifest(FlagMigrationPending) },
+			mutate:  func(m *FlagMigrationManifest) { m.Migrations[0].Legacy.After.AliasOf = "other-id" },
+			wantErr: `legacy flag after state must declare alias_of "message-id"`,
+		},
+		{
+			name:    "legacy before cannot declare alias",
+			make:    func() FlagMigrationManifest { return coverageManifest(FlagMigrationPending) },
+			mutate:  func(m *FlagMigrationManifest) { m.Migrations[0].Legacy.Before.AliasOf = "message-id" },
+			wantErr: "alias_of is only valid on the legacy after state",
+		},
+		{
+			name:    "canonical before cannot declare alias",
+			make:    optionalCanonical,
+			mutate:  func(m *FlagMigrationManifest) { m.Migrations[0].Canonical.Before.AliasOf = "message-id" },
+			wantErr: "alias_of is only valid on the legacy after state",
+		},
+		{
+			name:    "canonical after cannot declare alias",
+			make:    func() FlagMigrationManifest { return coverageManifest(FlagMigrationPending) },
+			mutate:  func(m *FlagMigrationManifest) { m.Migrations[0].Canonical.After.AliasOf = "message-id" },
+			wantErr: "alias_of is only valid on the legacy after state",
+		},
+		{
+			name:    "canonical type remains stable",
+			make:    optionalCanonical,
+			mutate:  func(m *FlagMigrationManifest) { m.Migrations[0].Canonical.Before.Type = "stringSlice" },
+			wantErr: "canonical flag type must remain unchanged",
+		},
+		{
+			name:    "canonical shorthand remains stable",
+			make:    optionalCanonical,
+			mutate:  func(m *FlagMigrationManifest) { m.Migrations[0].Canonical.Before.Shorthand = "m" },
+			wantErr: "canonical flag shorthand must remain unchanged",
+		},
+		{
+			name:    "canonical no-opt remains stable",
+			make:    optionalCanonical,
+			mutate:  func(m *FlagMigrationManifest) { m.Migrations[0].Canonical.Before.NoOpt = "auto" },
+			wantErr: "canonical flag no_opt must remain unchanged",
+		},
+		{
+			name:    "canonical scope remains stable",
+			make:    optionalCanonical,
+			mutate:  func(m *FlagMigrationManifest) { m.Migrations[0].Canonical.Before.Scope = "inherited" },
+			wantErr: "canonical flag scope must remain unchanged",
+		},
+		{
+			name:    "legacy before type matches legacy after",
+			make:    func() FlagMigrationManifest { return coverageManifest(FlagMigrationPending) },
+			mutate:  func(m *FlagMigrationManifest) { m.Migrations[0].Legacy.Before.Type = "stringSlice" },
+			wantErr: "legacy and canonical flag types must match exactly",
+		},
+		{
+			name:    "legacy after type matches canonical after",
+			make:    func() FlagMigrationManifest { return coverageManifest(FlagMigrationPending) },
+			mutate:  func(m *FlagMigrationManifest) { m.Migrations[0].Canonical.After.Type = "stringSlice" },
+			wantErr: "legacy and canonical flag types must match exactly",
+		},
+		{
+			name:    "legacy shorthand remains stable",
+			make:    func() FlagMigrationManifest { return coverageManifest(FlagMigrationPending) },
+			mutate:  func(m *FlagMigrationManifest) { m.Migrations[0].Legacy.After.Shorthand = "x" },
+			wantErr: "legacy flag shorthand must remain unchanged",
+		},
+		{
+			name:    "legacy no-opt remains stable",
+			make:    func() FlagMigrationManifest { return coverageManifest(FlagMigrationPending) },
+			mutate:  func(m *FlagMigrationManifest) { m.Migrations[0].Legacy.After.NoOpt = "" },
+			wantErr: "legacy flag no_opt must remain unchanged",
+		},
+		{
+			name:    "legacy scope remains stable",
+			make:    func() FlagMigrationManifest { return coverageManifest(FlagMigrationPending) },
+			mutate:  func(m *FlagMigrationManifest) { m.Migrations[0].Legacy.After.Scope = "inherited" },
+			wantErr: "legacy flag scope must remain unchanged",
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			manifest := test.make()
+			test.mutate(&manifest)
+			if err := manifest.Validate(); err == nil || !strings.Contains(err.Error(), test.wantErr) {
+				t.Fatalf("Validate() error = %v, want %q", err, test.wantErr)
+			}
+		})
+	}
+}
+
+func TestCrossPlatformCoverageFlagMigrationManifestRejectsAbsentAttributesAndTargetConflicts(t *testing.T) {
+	stateMutations := []struct {
+		name   string
+		mutate func(*FlagMigrationState)
+	}{
+		{name: "type", mutate: func(s *FlagMigrationState) { s.Type = "string" }},
+		{name: "required", mutate: func(s *FlagMigrationState) { s.Required = true }},
+		{name: "hidden", mutate: func(s *FlagMigrationState) { s.Hidden = true }},
+		{name: "shorthand", mutate: func(s *FlagMigrationState) { s.Shorthand = "m" }},
+		{name: "no-opt", mutate: func(s *FlagMigrationState) { s.NoOpt = "auto" }},
+		{name: "scope", mutate: func(s *FlagMigrationState) { s.Scope = "local" }},
+		{name: "alias", mutate: func(s *FlagMigrationState) { s.AliasOf = "message-id" }},
+	}
+	for _, test := range stateMutations {
+		t.Run("absent state rejects "+test.name, func(t *testing.T) {
+			manifest := coverageManifest(FlagMigrationPending)
+			test.mutate(&manifest.Migrations[0].Canonical.Before)
+			if err := manifest.Validate(); err == nil || !strings.Contains(err.Error(), "absent state must not declare flag attributes") {
+				t.Fatalf("Validate() error = %v, want absent-state rejection", err)
+			}
+		})
+	}
+
+	manifest := coverageManifest(FlagMigrationPending)
+	second := coverageManifest(FlagMigrationPending).Migrations[0]
+	second.Canonical.Name = "canonical-two"
+	second.Legacy.After.AliasOf = second.Canonical.Name
+	manifest.Migrations = append(manifest.Migrations, second)
+	if err := manifest.Validate(); err == nil || !strings.Contains(err.Error(), "to both --message-id and --canonical-two") {
+		t.Fatalf("Validate() error = %v, want one legacy to two canonical targets rejected", err)
+	}
+}
+
+func TestCrossPlatformCoverageCompareAllWithFlagMigrationsValidatesEveryInput(t *testing.T) {
+	before := coverageMigrationSnapshot(coverageManifest(FlagMigrationPending).Migrations[0], false, false)
+	empty := coverageEmptyManifest()
+
+	tests := []struct {
+		name       string
+		current    Snapshot
+		references map[string]Snapshot
+		authority  FlagMigrationManifest
+		candidate  FlagMigrationManifest
+		wantErr    string
+	}{
+		{
+			name:       "current snapshot",
+			current:    Snapshot{},
+			references: map[string]Snapshot{"merge-base": before},
+			authority:  empty,
+			candidate:  empty,
+			wantErr:    "validate current interface snapshot",
+		},
+		{
+			name:       "reference snapshot",
+			current:    before,
+			references: map[string]Snapshot{"merge-base": {}},
+			authority:  empty,
+			candidate:  empty,
+			wantErr:    "validate merge-base interface snapshot",
+		},
+		{
+			name:       "authority manifest",
+			current:    before,
+			references: map[string]Snapshot{"merge-base": before},
+			authority:  FlagMigrationManifest{},
+			candidate:  empty,
+			wantErr:    "validate approved flag migrations",
+		},
+		{
+			name:       "candidate manifest",
+			current:    before,
+			references: map[string]Snapshot{"merge-base": before},
+			authority:  empty,
+			candidate:  FlagMigrationManifest{},
+			wantErr:    "validate candidate flag migrations",
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			_, err := CompareAllWithFlagMigrations(test.current, test.references, test.authority, test.candidate)
+			if err == nil || !strings.Contains(err.Error(), test.wantErr) {
+				t.Fatalf("CompareAllWithFlagMigrations() error = %v, want %q", err, test.wantErr)
+			}
+		})
+	}
+
+	report, err := CompareAllWithFlagMigrations(before, map[string]Snapshot{"merge-base": before}, empty, empty)
+	if err != nil || !report.Compatible {
+		t.Fatalf("empty migration lifecycle = (%#v, %v), want compatible", report, err)
+	}
+}
+
+func TestCrossPlatformCoverageCompareAllWithFlagMigrationsLifecycleErrors(t *testing.T) {
+	pending := coverageManifest(FlagMigrationPending)
+	consumed := coverageManifest(FlagMigrationConsumed)
+	empty := coverageEmptyManifest()
+	migration := pending.Migrations[0]
+	before := coverageMigrationSnapshot(migration, false, false)
+	after := coverageMigrationSnapshot(migration, true, false)
+	partial := coverageMigrationSnapshot(migration, false, false)
+	partial.Commands[len(partial.Commands)-1].LocalFlags[0].Hidden = true
+
+	modified := coverageManifest(FlagMigrationPending)
+	modified.Migrations[0].Reason = "A different reviewed reason."
+
+	tests := []struct {
+		name       string
+		current    Snapshot
+		references map[string]Snapshot
+		authority  FlagMigrationManifest
+		candidate  FlagMigrationManifest
+		wantErr    string
+	}{
+		{
+			name:       "requires merge-base or main",
+			current:    before,
+			references: map[string]Snapshot{"stable": before},
+			authority:  pending,
+			candidate:  pending,
+			wantErr:    "requires a main or merge-base reference",
+		},
+		{
+			name:       "pending authority requires before base",
+			current:    after,
+			references: map[string]Snapshot{"merge-base": after, "stable": after},
+			authority:  pending,
+			candidate:  pending,
+			wantErr:    "want exact before state for pending",
+		},
+		{
+			name:       "consumed authority requires after base",
+			current:    before,
+			references: map[string]Snapshot{"merge-base": before, "stable": before},
+			authority:  consumed,
+			candidate:  consumed,
+			wantErr:    "want exact after state for consumed",
+		},
+		{
+			name:       "missing base command is partial",
+			current:    before,
+			references: map[string]Snapshot{"merge-base": coverageRootOnlySnapshot(), "stable": before},
+			authority:  pending,
+			candidate:  pending,
+			wantErr:    "is partial in merge-base",
+		},
+		{
+			name:       "candidate cannot modify base receipt",
+			current:    before,
+			references: map[string]Snapshot{"merge-base": before, "stable": before},
+			authority:  pending,
+			candidate:  modified,
+			wantErr:    "candidate modified base-owned flag migration",
+		},
+		{
+			name:       "candidate cannot remove pending receipt",
+			current:    before,
+			references: map[string]Snapshot{"merge-base": before, "stable": before},
+			authority:  pending,
+			candidate:  empty,
+			wantErr:    "candidate removed pending flag migration",
+		},
+		{
+			name:       "candidate cannot falsely consume pending receipt",
+			current:    before,
+			references: map[string]Snapshot{"merge-base": before, "stable": before},
+			authority:  pending,
+			candidate:  consumed,
+			wantErr:    "candidate falsely consumed unchanged flag migration",
+		},
+		{
+			name:       "candidate must consume completed receipt",
+			current:    after,
+			references: map[string]Snapshot{"merge-base": before, "stable": before},
+			authority:  pending,
+			candidate:  pending,
+			wantErr:    "without marking it consumed",
+		},
+		{
+			name:       "candidate rejects partial application",
+			current:    partial,
+			references: map[string]Snapshot{"merge-base": before, "stable": before},
+			authority:  pending,
+			candidate:  consumed,
+			wantErr:    "candidate partially applied flag migration",
+		},
+		{
+			name:       "consumed receipt rejects current revert",
+			current:    before,
+			references: map[string]Snapshot{"merge-base": after, "stable": before},
+			authority:  consumed,
+			candidate:  consumed,
+			wantErr:    "candidate drifted from consumed flag migration",
+		},
+		{
+			name:       "consumed receipt cannot be removed early",
+			current:    after,
+			references: map[string]Snapshot{"merge-base": after, "stable": before},
+			authority:  consumed,
+			candidate:  empty,
+			wantErr:    "before every reference reached the after state",
+		},
+		{
+			name:       "consumed receipt cannot revert to pending",
+			current:    after,
+			references: map[string]Snapshot{"merge-base": after, "stable": before},
+			authority:  consumed,
+			candidate:  pending,
+			wantErr:    "back to pending",
+		},
+		{
+			name:       "consumed receipt becomes stale",
+			current:    after,
+			references: map[string]Snapshot{"merge-base": after, "stable": after},
+			authority:  consumed,
+			candidate:  consumed,
+			wantErr:    "is stale after all references reached the after state",
+		},
+		{
+			name:       "candidate-added receipt starts pending",
+			current:    before,
+			references: map[string]Snapshot{"merge-base": before, "stable": before},
+			authority:  empty,
+			candidate:  consumed,
+			wantErr:    "must start pending",
+		},
+		{
+			name:       "candidate-added receipt matches base before",
+			current:    before,
+			references: map[string]Snapshot{"merge-base": after, "stable": after},
+			authority:  empty,
+			candidate:  pending,
+			wantErr:    "does not match the merge-base before state",
+		},
+		{
+			name:       "candidate-added receipt cannot self-authorize",
+			current:    after,
+			references: map[string]Snapshot{"merge-base": before, "stable": before},
+			authority:  empty,
+			candidate:  pending,
+			wantErr:    "cannot authorize its own interface change",
+		},
+		{
+			name:       "missing current command is partial",
+			current:    coverageRootOnlySnapshot(),
+			references: map[string]Snapshot{"merge-base": before, "stable": before},
+			authority:  pending,
+			candidate:  pending,
+			wantErr:    "candidate partially applied flag migration",
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			_, err := CompareAllWithFlagMigrations(test.current, test.references, test.authority, test.candidate)
+			if err == nil || !strings.Contains(err.Error(), test.wantErr) {
+				t.Fatalf("CompareAllWithFlagMigrations() error = %v, want %q", err, test.wantErr)
+			}
+		})
+	}
+}
+
+func TestCrossPlatformCoverageCompareAllWithFlagMigrationsMainAuthorityAndCleanup(t *testing.T) {
+	pending := coverageManifest(FlagMigrationPending)
+	consumed := coverageManifest(FlagMigrationConsumed)
+	empty := coverageEmptyManifest()
+	migration := pending.Migrations[0]
+	before := coverageMigrationSnapshot(migration, false, false)
+	after := coverageMigrationSnapshot(migration, true, false)
+
+	report, err := CompareAllWithFlagMigrations(
+		after,
+		map[string]Snapshot{"main": before, "stable": before},
+		pending,
+		consumed,
+	)
+	if err != nil || !report.Compatible {
+		t.Fatalf("main-owned migration = (%#v, %v), want compatible", report, err)
+	}
+
+	report, err = CompareAllWithFlagMigrations(
+		after,
+		map[string]Snapshot{"merge-base": before, "main": after, "stable": before},
+		pending,
+		consumed,
+	)
+	if err != nil || !report.Compatible {
+		t.Fatalf("merge-base precedence = (%#v, %v), want compatible", report, err)
+	}
+
+	report, err = CompareAllWithFlagMigrations(
+		after,
+		map[string]Snapshot{"merge-base": after, "stable": after},
+		consumed,
+		empty,
+	)
+	if err != nil || !report.Compatible {
+		t.Fatalf("consumed receipt cleanup = (%#v, %v), want compatible", report, err)
+	}
+
+	report, err = CompareAllWithFlagMigrations(
+		before,
+		map[string]Snapshot{"merge-base": before, "stable": before},
+		empty,
+		pending,
+	)
+	if err != nil || !report.Compatible {
+		t.Fatalf("pending receipt creation = (%#v, %v), want compatible", report, err)
+	}
+}
+
+func TestCrossPlatformCoverageCompareAllWithFlagMigrationsRequiresStableForReceiptCleanup(t *testing.T) {
+	consumed := coverageManifest(FlagMigrationConsumed)
+	empty := coverageEmptyManifest()
+	after := coverageMigrationSnapshot(consumed.Migrations[0], true, false)
+
+	report, err := CompareAllWithFlagMigrations(
+		after,
+		map[string]Snapshot{"merge-base": after},
+		consumed,
+		empty,
+	)
+	if err == nil || !strings.Contains(err.Error(), "requires a stable reference") {
+		t.Fatalf("receipt cleanup without stable = (%#v, %v), want stable-reference error", report, err)
+	}
+}
+
+func TestCrossPlatformCoverageCompareAllWithFlagMigrationsInheritedAndOptionalCanonical(t *testing.T) {
+	t.Run("inherited exact state", func(t *testing.T) {
+		pending := coverageManifest(FlagMigrationPending)
+		pending.Migrations[0].Legacy.Before.Scope = "inherited"
+		pending.Migrations[0].Legacy.After.Scope = "inherited"
+		pending.Migrations[0].Canonical.After.Scope = "inherited"
+		consumed := pending
+		consumed.Migrations = append([]FlagMigration(nil), pending.Migrations...)
+		consumed.Migrations[0].State = FlagMigrationConsumed
+		before := coverageMigrationSnapshot(pending.Migrations[0], false, false)
+		after := coverageMigrationSnapshot(pending.Migrations[0], true, false)
+
+		report, err := CompareAllWithFlagMigrations(
+			after,
+			map[string]Snapshot{"merge-base": before, "stable": before},
+			pending,
+			consumed,
+		)
+		if err != nil || !report.Compatible {
+			t.Fatalf("inherited migration = (%#v, %v), want compatible", report, err)
+		}
+	})
+
+	t.Run("existing optional canonical becomes required", func(t *testing.T) {
+		pending := coverageManifest(FlagMigrationPending)
+		pending.Migrations[0].Canonical.Before = FlagMigrationState{
+			Present: true,
+			Type:    "string",
+			Scope:   "local",
+		}
+		consumed := pending
+		consumed.Migrations = append([]FlagMigration(nil), pending.Migrations...)
+		consumed.Migrations[0].State = FlagMigrationConsumed
+		before := coverageMigrationSnapshot(pending.Migrations[0], false, false)
+		after := coverageMigrationSnapshot(pending.Migrations[0], true, false)
+
+		ordinary := Compare(after, before, "merge-base")
+		if !hasFlagChange(ordinary.Blocking, "flag_became_required", pending.Migrations[0].Command, pending.Migrations[0].Canonical.Name) {
+			t.Fatalf("fixture did not create flag_became_required: %#v", ordinary.Blocking)
+		}
+		report, err := CompareAllWithFlagMigrations(
+			after,
+			map[string]Snapshot{"merge-base": before, "stable": before},
+			pending,
+			consumed,
+		)
+		if err != nil || !report.Compatible {
+			t.Fatalf("optional-to-required migration = (%#v, %v), want compatible", report, err)
+		}
+	})
+}
+
+func TestCrossPlatformCoverageCompareAllWithFlagMigrationsNormalizesAliasesAndKeepsOtherBreakage(t *testing.T) {
+	pending := coverageManifest(FlagMigrationPending)
+	consumed := coverageManifest(FlagMigrationConsumed)
+	migration := pending.Migrations[0]
+	before := coverageMigrationSnapshot(migration, false, true)
+	after := coverageMigrationSnapshot(migration, true, true)
+
+	ordinary := Compare(after, before, "merge-base")
+	if !hasFlagChange(ordinary.Blocking, "flag_became_hidden", "dws chat deliver", migration.Legacy.Name) {
+		t.Fatalf("alias fixture did not create alias-path finding: %#v", ordinary.Blocking)
+	}
+	report, err := CompareAllWithFlagMigrations(
+		after,
+		map[string]Snapshot{"merge-base": before, "stable": before},
+		pending,
+		consumed,
+	)
+	if err != nil || !report.Compatible {
+		t.Fatalf("alias-path migration = (%#v, %v), want compatible", report, err)
+	}
+
+	beforeWithOther := before
+	afterWithOther := after
+	beforeWithOther.Commands = append(beforeWithOther.Commands, coverageCommand("dws files", nil, []Flag{{Name: "format", Type: "string"}}, nil))
+	afterWithOther.Commands = append(afterWithOther.Commands, coverageCommand("dws files", nil, nil, nil))
+	report, err = CompareAllWithFlagMigrations(
+		afterWithOther,
+		map[string]Snapshot{"merge-base": beforeWithOther, "stable": beforeWithOther},
+		pending,
+		consumed,
+	)
+	if err != nil {
+		t.Fatalf("unrelated breakage returned lifecycle error: %v", err)
+	}
+	if report.Compatible || !hasFlagChange(report.Comparisons[0].Blocking, "flag_removed", "dws files", "format") {
+		t.Fatalf("unrelated breakage was hidden: %#v", report)
+	}
+}
+
+func TestCrossPlatformCoverageCompareAllWithFlagMigrationsDoesNotAuthorizePartialReferenceOrRuleDrift(t *testing.T) {
+	pending := coverageManifest(FlagMigrationPending)
+	consumed := coverageManifest(FlagMigrationConsumed)
+	migration := pending.Migrations[0]
+	before := coverageMigrationSnapshot(migration, false, false)
+	after := coverageMigrationSnapshot(migration, true, false)
+
+	partialReference := coverageMigrationSnapshot(migration, false, false)
+	partialReference.Commands[len(partialReference.Commands)-1].LocalFlags[0].Hidden = true
+	report, err := CompareAllWithFlagMigrations(
+		after,
+		map[string]Snapshot{"merge-base": before, "stable": partialReference},
+		pending,
+		consumed,
+	)
+	if err != nil {
+		t.Fatalf("partial non-authority reference returned lifecycle error: %v", err)
+	}
+	if report.Compatible {
+		t.Fatalf("partial reference was incorrectly authorized: %#v", report)
+	}
+
+	beforeWithoutRoot := coverageMigrationSnapshot(migration, false, false)
+	afterWithoutRoot := coverageMigrationSnapshot(migration, true, false)
+	beforeWithoutRoot.Commands = beforeWithoutRoot.Commands[len(beforeWithoutRoot.Commands)-1:]
+	afterWithoutRoot.Commands = afterWithoutRoot.Commands[len(afterWithoutRoot.Commands)-1:]
+	afterWithoutRoot.Rules.ExcludedFlags = []string{"help", "version"}
+	report, err = CompareAllWithFlagMigrations(
+		afterWithoutRoot,
+		map[string]Snapshot{"merge-base": beforeWithoutRoot, "stable": beforeWithoutRoot},
+		pending,
+		consumed,
+	)
+	if err != nil {
+		t.Fatalf("rule drift returned lifecycle error: %v", err)
+	}
+	if report.Compatible || !hasChangeKind(report.Comparisons[0].Blocking, "snapshot_rules_changed") {
+		t.Fatalf("snapshot rule drift was incorrectly authorized: %#v", report)
+	}
+}
+
+func coverageManifestJSON() string {
+	return `{
+  "version": 1,
+  "migrations": [
+    {
+      "command": "dws chat send",
+      "legacy": {
+        "name": "legacy-id",
+        "before": {"present": true, "type": "string", "required": true, "shorthand": "l", "no_opt": "auto", "scope": "local"},
+        "after": {"present": true, "type": "string", "hidden": true, "shorthand": "l", "no_opt": "auto", "scope": "local", "alias_of": "message-id"}
+      },
+      "canonical": {
+        "name": "message-id",
+        "before": {"present": false},
+        "after": {"present": true, "type": "string", "required": true, "scope": "local"}
+      },
+      "state": "pending",
+      "reason": "Reviewed exact flag migration."
+    }
+  ]
+}`
+}
+
+func coverageManifest(state string) FlagMigrationManifest {
+	return FlagMigrationManifest{
+		Version: FlagMigrationManifestVersion,
+		Migrations: []FlagMigration{
+			{
+				Command: "dws chat send",
+				Legacy: FlagMigrationSide{
+					Name: "legacy-id",
+					Before: FlagMigrationState{
+						Present:   true,
+						Type:      "string",
+						Required:  true,
+						Shorthand: "l",
+						NoOpt:     "auto",
+						Scope:     "local",
+					},
+					After: FlagMigrationState{
+						Present:   true,
+						Type:      "string",
+						Hidden:    true,
+						Shorthand: "l",
+						NoOpt:     "auto",
+						Scope:     "local",
+						AliasOf:   "message-id",
+					},
+				},
+				Canonical: FlagMigrationSide{
+					Name:   "message-id",
+					Before: FlagMigrationState{},
+					After: FlagMigrationState{
+						Present:  true,
+						Type:     "string",
+						Required: true,
+						Scope:    "local",
+					},
+				},
+				State:  state,
+				Reason: "Reviewed exact flag migration.",
+			},
+		},
+	}
+}
+
+func coverageEmptyManifest() FlagMigrationManifest {
+	return FlagMigrationManifest{Version: FlagMigrationManifestVersion, Migrations: []FlagMigration{}}
+}
+
+func coverageMigrationSnapshot(migration FlagMigration, after, commandAlias bool) Snapshot {
+	legacyState := migration.Legacy.Before
+	canonicalState := migration.Canonical.Before
+	if after {
+		legacyState = migration.Legacy.After
+		canonicalState = migration.Canonical.After
+	}
+	local := []Flag{}
+	inherited := []Flag{}
+	add := func(name string, state FlagMigrationState) {
+		if !state.Present {
+			return
+		}
+		flag := Flag{
+			Name:      name,
+			Shorthand: state.Shorthand,
+			Type:      state.Type,
+			NoOpt:     state.NoOpt,
+			Required:  state.Required,
+			Hidden:    state.Hidden,
+			AliasOf:   state.AliasOf,
+		}
+		if state.Scope == "inherited" {
+			inherited = append(inherited, flag)
+			return
+		}
+		local = append(local, flag)
+	}
+	add(migration.Legacy.Name, legacyState)
+	add(migration.Canonical.Name, canonicalState)
+	aliases := []string{}
+	if commandAlias {
+		aliases = []string{"deliver"}
+	}
+	return coverageSnapshot(
+		coverageCommand("dws", nil, nil, nil),
+		coverageCommand("dws chat", nil, nil, nil),
+		coverageCommand(migration.Command, aliases, local, inherited),
+	)
+}
+
+func coverageRootOnlySnapshot() Snapshot {
+	return coverageSnapshot(coverageCommand("dws", nil, nil, nil))
+}
+
+func coverageSnapshot(commands ...Command) Snapshot {
+	return Snapshot{
+		SchemaVersion: SchemaVersion,
+		Rules: Rules{
+			ExcludedCommandSubtrees: append([]string(nil), excludedCommandSubtrees...),
+			ExcludedFlags:           []string{"help"},
+		},
+		Commands: commands,
+	}
+}
+
+func coverageCommand(path string, aliases []string, local, inherited []Flag) Command {
+	if aliases == nil {
+		aliases = []string{}
+	}
+	if local == nil {
+		local = []Flag{}
+	}
+	if inherited == nil {
+		inherited = []Flag{}
+	}
+	return Command{
+		Path:           path,
+		Runnable:       true,
+		Aliases:        aliases,
+		LocalFlags:     local,
+		InheritedFlags: inherited,
+	}
+}

internal/interfacesnapshot/migrations_test.go

@@ -0,0 +1,252 @@
+// Copyright 2026 Alibaba Group
+// Licensed under the Apache License, Version 2.0 (the "License");
+
+package interfacesnapshot
+
+import (
+	"errors"
+	"reflect"
+	"strings"
+	"testing"
+)
+
+type flagMigrationErrorReader struct{}
+
+func (flagMigrationErrorReader) Read([]byte) (int, error) {
+	return 0, errors.New("injected read failure")
+}
+
+const validFlagMigrationManifestJSON = `{
+  "version": 1,
+  "migrations": [
+    {
+      "command": "dws chat message recall",
+      "legacy": {
+        "name": "msg-id",
+        "before": {"present": true, "type": "string", "required": true, "scope": "local"},
+        "after": {"present": true, "type": "string", "hidden": true, "scope": "local", "alias_of": "message-id"}
+      },
+      "canonical": {
+        "name": "message-id",
+        "before": {"present": false},
+        "after": {"present": true, "type": "string", "required": true, "scope": "local"}
+      },
+      "state": "pending",
+      "reason": "Reviewed message identifier flag migration."
+    }
+  ]
+}`
+
+func TestCrossPlatformCoverageReadFlagMigrationManifestRejectsUnknownFields(t *testing.T) {
+	_, err := ReadFlagMigrationManifest(strings.NewReader(`{
+  "version": 1,
+  "migrations": [],
+  "allow_all_chat_flags": true
+}`))
+	if err == nil || !strings.Contains(err.Error(), "unknown field") {
+		t.Fatalf("ReadFlagMigrationManifest() error = %v, want unknown field rejection", err)
+	}
+}
+
+func TestCrossPlatformCoverageReadFlagMigrationManifestRejectsDuplicateFields(t *testing.T) {
+	for _, input := range []string{
+		`{"version": 999, "version": 1, "migrations": []}`,
+		strings.Replace(validFlagMigrationManifestJSON, `"state": "pending"`, `"state": "consumed", "state": "pending"`, 1),
+	} {
+		_, err := ReadFlagMigrationManifest(strings.NewReader(input))
+		if err == nil || !strings.Contains(err.Error(), "duplicate field") {
+			t.Fatalf("ReadFlagMigrationManifest() error = %v, want duplicate field rejection", err)
+		}
+	}
+}
+
+func TestCrossPlatformCoverageReadFlagMigrationManifestRejectsNonCanonicalFields(t *testing.T) {
+	for _, input := range []string{
+		`{"Version": 1, "migrations": []}`,
+		`{"version": 1, "Version": 1, "migrations": []}`,
+		strings.Replace(validFlagMigrationManifestJSON, `"state": "pending"`, `"State": "pending"`, 1),
+	} {
+		_, err := ReadFlagMigrationManifest(strings.NewReader(input))
+		if err == nil || !strings.Contains(err.Error(), "non-canonical field") {
+			t.Fatalf("ReadFlagMigrationManifest() error = %v, want non-canonical field rejection", err)
+		}
+	}
+}
+
+func TestCrossPlatformCoverageReadFlagMigrationManifestRejectsNullScalars(t *testing.T) {
+	for _, input := range []string{
+		strings.Replace(validFlagMigrationManifestJSON, `"required": true`, `"required": null`, 1),
+		strings.Replace(validFlagMigrationManifestJSON, `"scope": "local"`, `"scope": "local", "shorthand": null`, 1),
+		`{"version": null, "migrations": []}`,
+	} {
+		_, err := ReadFlagMigrationManifest(strings.NewReader(input))
+		if err == nil || !strings.Contains(err.Error(), "must be") || !strings.Contains(err.Error(), "null") {
+			t.Fatalf("ReadFlagMigrationManifest() error = %v, want scalar null rejection", err)
+		}
+	}
+}
+
+func TestCrossPlatformCoverageReadFlagMigrationManifestRejectsWrongJSONTypes(t *testing.T) {
+	for _, input := range []string{
+		strings.Replace(validFlagMigrationManifestJSON, `"required": true`, `"required": "true"`, 1),
+		strings.Replace(validFlagMigrationManifestJSON, `"scope": "local"`, `"scope": false`, 1),
+		`{"version": "1", "migrations": []}`,
+		`{"version": 1, "migrations": {}}`,
+	} {
+		if _, err := ReadFlagMigrationManifest(strings.NewReader(input)); err == nil {
+			t.Fatal("ReadFlagMigrationManifest() accepted a field with the wrong JSON type")
+		}
+	}
+}
+
+func TestCrossPlatformCoverageReadFlagMigrationManifestReportsReaderFailure(t *testing.T) {
+	_, err := ReadFlagMigrationManifest(flagMigrationErrorReader{})
+	if err == nil || !strings.Contains(err.Error(), "injected read failure") {
+		t.Fatalf("ReadFlagMigrationManifest() error = %v, want reader failure", err)
+	}
+}
+
+func TestCrossPlatformCoverageReadFlagMigrationManifestValidatesExactEntries(t *testing.T) {
+	if _, err := ReadFlagMigrationManifest(strings.NewReader(validFlagMigrationManifestJSON)); err != nil {
+		t.Fatalf("ReadFlagMigrationManifest(valid) error = %v", err)
+	}
+
+	tests := []struct {
+		name    string
+		input   string
+		wantErr string
+	}{
+		{
+			name:    "unsupported version",
+			input:   strings.Replace(validFlagMigrationManifestJSON, `"version": 1`, `"version": 2`, 1),
+			wantErr: "unsupported flag migration manifest version",
+		},
+		{
+			name:    "missing migrations array",
+			input:   `{"version": 1}`,
+			wantErr: "migrations must be an array",
+		},
+		{
+			name:    "null migrations array",
+			input:   `{"version": 1, "migrations": null}`,
+			wantErr: "migrations must be an array",
+		},
+		{
+			name:    "wildcard command",
+			input:   strings.Replace(validFlagMigrationManifestJSON, "dws chat message recall", "dws chat *", 1),
+			wantErr: "exact command path",
+		},
+		{
+			name:    "wildcard flag",
+			input:   strings.Replace(validFlagMigrationManifestJSON, `"name": "msg-id"`, `"name": "msg-*"`, 1),
+			wantErr: "exact legacy flag",
+		},
+		{
+			name:    "empty reason",
+			input:   strings.Replace(validFlagMigrationManifestJSON, "Reviewed message identifier flag migration.", "", 1),
+			wantErr: "non-empty reason",
+		},
+		{
+			name:    "invalid state",
+			input:   strings.Replace(validFlagMigrationManifestJSON, `"state": "pending"`, `"state": "approved"`, 1),
+			wantErr: "invalid state",
+		},
+		{
+			name:    "legacy removed",
+			input:   strings.Replace(validFlagMigrationManifestJSON, `"after": {"present": true, "type": "string", "hidden": true, "scope": "local", "alias_of": "message-id"}`, `"after": {"present": false}`, 1),
+			wantErr: "legacy flag must remain present",
+		},
+		{
+			name:    "canonical hidden",
+			input:   strings.Replace(validFlagMigrationManifestJSON, `"after": {"present": true, "type": "string", "required": true, "scope": "local"}`, `"after": {"present": true, "type": "string", "required": true, "hidden": true, "scope": "local"}`, 1),
+			wantErr: "canonical flag must remain visible",
+		},
+		{
+			name: "canonical remains optional",
+			input: strings.Replace(
+				validFlagMigrationManifestJSON,
+				`"after": {"present": true, "type": "string", "required": true, "scope": "local"}`,
+				`"after": {"present": true, "type": "string", "scope": "local"}`,
+				1,
+			),
+			wantErr: "canonical flag must be required after migration",
+		},
+		{
+			name: "canonical was already required",
+			input: strings.Replace(
+				validFlagMigrationManifestJSON,
+				`"before": {"present": false}`,
+				`"before": {"present": true, "type": "string", "required": true, "scope": "local"}`,
+				1,
+			),
+			wantErr: "canonical flag must be absent or optional before migration",
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			_, err := ReadFlagMigrationManifest(strings.NewReader(test.input))
+			if err == nil || !strings.Contains(err.Error(), test.wantErr) {
+				t.Fatalf("ReadFlagMigrationManifest() error = %v, want %q", err, test.wantErr)
+			}
+		})
+	}
+}
+
+func TestCrossPlatformCoverageFlagMigrationManifestRejectsDuplicateAndInexactContracts(t *testing.T) {
+	manifest, err := ReadFlagMigrationManifest(strings.NewReader(validFlagMigrationManifestJSON))
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	duplicate := manifest
+	duplicate.Migrations = append(duplicate.Migrations, duplicate.Migrations[0])
+	if err := duplicate.Validate(); err == nil || !strings.Contains(err.Error(), "duplicates") {
+		t.Fatalf("duplicate migration error = %v", err)
+	}
+
+	whitespaceReason := manifest
+	whitespaceReason.Migrations = append([]FlagMigration(nil), manifest.Migrations...)
+	whitespaceReason.Migrations[0].Reason = "  reviewed  "
+	if err := whitespaceReason.Validate(); err == nil || !strings.Contains(err.Error(), "trimmed") {
+		t.Fatalf("whitespace reason error = %v", err)
+	}
+
+	canonicalDrift := manifest
+	canonicalDrift.Migrations = append([]FlagMigration(nil), manifest.Migrations...)
+	canonicalDrift.Migrations[0].Canonical.Before = FlagMigrationState{
+		Present: true,
+		Type:    "string",
+		Scope:   "local",
+	}
+	canonicalDrift.Migrations[0].Canonical.After.Type = "stringSlice"
+	if err := canonicalDrift.Validate(); err == nil || !strings.Contains(err.Error(), "canonical flag type") {
+		t.Fatalf("canonical type drift error = %v", err)
+	}
+
+	_, err = ReadFlagMigrationManifest(strings.NewReader(validFlagMigrationManifestJSON + ` {}`))
+	if err == nil || !strings.Contains(err.Error(), "trailing") {
+		t.Fatalf("trailing JSON error = %v", err)
+	}
+}
+
+func TestCrossPlatformCoverageAuthorizeFlagMigrationsReturnsExactBaseOwnedApproval(t *testing.T) {
+	pending := coverageManifest(FlagMigrationPending)
+	consumed := coverageManifest(FlagMigrationConsumed)
+	migration := pending.Migrations[0]
+	before := coverageMigrationSnapshot(migration, false, false)
+	after := coverageMigrationSnapshot(migration, true, false)
+
+	got, err := AuthorizeFlagMigrations(
+		after,
+		map[string]Snapshot{"merge-base": before, "stable": before},
+		pending,
+		consumed,
+	)
+	if err != nil {
+		t.Fatalf("AuthorizeFlagMigrations() error = %v", err)
+	}
+	if !reflect.DeepEqual(got, pending.Migrations) {
+		t.Fatalf("AuthorizeFlagMigrations() = %#v, want exact base-owned approval %#v", got, pending.Migrations)
+	}
+}

internal/interfacesnapshot/reviewed.go

@@ -0,0 +1,124 @@
+// 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 interfacesnapshot
+
+// flagTypeChange identifies one flag type migration exactly: which command,
+// which flag, and in which direction. Nothing is matched by wildcard — an entry
+// that differs in any of the four fields does not apply.
+//
+// CommandPath is the canonical Command.Path form, which always includes the root
+// command name ("dws minutes permission apply"). It is deliberately *not* the
+// alias-expanded accepted path: a command reachable through an alias enters
+// compareEffectiveFlags once per accepted spelling, and keying on the accepted
+// path would let every alias spelling bypass the exemption and re-report the
+// change.
+//
+// This table is duplicated in scripts/policy/interface-baseline/reviewed.go and
+// the two must stay identical; TestCrossPlatformCoverageReviewedFlagTypeTableMatchesInterfaceBaseline
+// fails if they drift. The duplication is forced rather than chosen:
+// check-authoritative-interface-baselines.sh copies the whole
+// scripts/policy/interface-baseline directory into a worktree checked out at a
+// historical revision and builds it there, so that copy cannot import a package
+// this branch adds.
+type flagTypeChange struct {
+	CommandPath string
+	Flag        string
+	From        string
+	To          string
+}
+
+// reviewedFlagTypeChanges enumerates the individually reviewed flag type
+// migrations this gate accepts. A flag type is part of the published CLI
+// contract, and a type swap cannot be proven safe from the type names alone —
+// whether old invocations still parse depends on the concrete value domain the
+// command enforces. So a migration is accepted only when this exact command,
+// flag, and old→new pair appear here. Adding an entry is a contract decision
+// and belongs in review, not in a feature change.
+//
+// Entries are direction-sensitive by construction: "string" → "int" is a
+// separate key from "int" → "string" and only the reviewed direction is
+// accepted. The reverse is not a candidate for this table — widening an int
+// flag back to string lets values the parser used to reject reach RunE, which
+// is a genuine loosening of validation rather than a type migration.
+//
+// The table alone does not admit a change: reviewedFlagTypeChange is consulted
+// only when nothing else about the flag moved. See compareEffectiveFlags.
+var reviewedFlagTypeChanges = map[flagTypeChange]struct{}{
+	// "dws minutes permission apply --policy" moved from a String flag parsed
+	// with strconv.ParseInt(v, 10, 64) in RunE to a native Int flag, keeping the
+	// same [2,4] domain check in RunE.
+	//
+	// This is admissible because the historical set of *successful* invocations
+	// is a subset of the new one, not merely similar to it:
+	//
+	//   - Old: the value had to parse as base-10 and land in [2,4]. That is
+	//     exactly "2", "3", "4" plus sign/leading-zero spellings of them
+	//     ("+3", "03", "003", ...).
+	//   - New: pflag's intValue.Set parses with strconv.ParseInt(s, 0, 64) —
+	//     base 0 — and RunE still enforces [2,4]. Every base-10 spelling above
+	//     resolves to the same number under base 0, so no historical success
+	//     becomes a failure.
+	//   - Base 0 additionally accepts "0x3" and friends, which the old parser
+	//     rejected. That widens the accepted set; it cannot break a caller who
+	//     was already succeeding.
+	//   - Values outside [2,4] and non-numeric values still fail. Only the
+	//     failure *message and timing* move, from RunE to flag parsing.
+	//
+	// The flag's default value changes from "" to "0" as an unavoidable
+	// consequence of the type. Neither gate compares defaults, and a default is
+	// not reachable by a caller here because RunE requires the flag to be
+	// explicitly set, so this is not a contract change.
+	{CommandPath: "dws minutes permission apply", Flag: "policy", From: "string", To: "int"}: {},
+}
+
+// reviewedFlagTypeChange reports whether this exact command, flag and direction
+// is a reviewed migration. Callers must additionally establish that the rest of
+// the flag's contract is unchanged before skipping the failure.
+func reviewedFlagTypeChange(commandPath, flag, from, to string) bool {
+	_, reviewed := reviewedFlagTypeChanges[flagTypeChange{
+		CommandPath: commandPath,
+		Flag:        flag,
+		From:        from,
+		To:          to,
+	}]
+	return reviewed
+}
+
+// flagContractOtherwiseChanged reports whether anything this gate checks about a
+// flag, other than its type, differs between the two revisions. A reviewed type
+// migration is only accepted when this returns false, so the exemption cannot
+// carry an unrelated regression in behind it.
+//
+// Only the fields compareEffectiveFlags actually enforces are compared. Default
+// is deliberately excluded: a string → int migration always changes it ("" to
+// "0"), so including it would make every entry in the table dead. Deprecated is
+// excluded for the same reason it is not a blocking change on its own.
+func flagContractOtherwiseChanged(oldFlag, newFlag Flag) bool {
+	if !oldFlag.Required && newFlag.Required {
+		return true
+	}
+	if oldFlag.Shorthand != "" && newFlag.Shorthand != oldFlag.Shorthand {
+		return true
+	}
+	if oldFlag.NoOpt != "" && newFlag.NoOpt != oldFlag.NoOpt {
+		return true
+	}
+	if !oldFlag.Hidden && newFlag.Hidden {
+		return true
+	}
+	if oldFlag.AliasOf != "" && newFlag.AliasOf != oldFlag.AliasOf {
+		return true
+	}
+	return false
+}

internal/interfacesnapshot/snapshot.go

@@ -22,11 +22,16 @@ import (
 	"sort"
 	"strings"
 
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/corecmd/runtimeannotate"
 	"github.com/spf13/cobra"
 	"github.com/spf13/pflag"
 )
 
-const SchemaVersion = 1
+const SchemaVersion = 2
+
+// FlagAliasOfAnnotation records a framework-originated reviewed relationship
+// between a retained compatibility flag and an exact canonical Cobra flag.
+const FlagAliasOfAnnotation = runtimeannotate.AnnotationFlagAliasOf
 
 var (
 	// Framework-owned command subtrees are deliberately excluded. They are
@@ -80,6 +85,12 @@ type Flag struct {
 	Required   bool   `json:"required"`
 	Hidden     bool   `json:"hidden"`
 	Deprecated string `json:"deprecated,omitempty"`
+	AliasOf    string `json:"alias_of,omitempty"`
+
+	// Capture cannot return an error, so it preserves malformed or non-framework
+	// relation evidence here for Validate/Write to reject. Valid snapshots never
+	// serialize this implementation detail.
+	malformedAliasEvidence bool
 }
 
 // Capture walks root without rendering help or executing any command. Both
@@ -193,6 +204,9 @@ func (s Snapshot) Validate() error {
 		if err := validateFlags(command.Path, "inherited", command.InheritedFlags); err != nil {
 			return err
 		}
+		if err := validateFlagAliases(command); err != nil {
+			return err
+		}
 	}
 	return nil
 }
@@ -206,21 +220,46 @@ func captureFlags(set *pflag.FlagSet) []Flag {
 		if flag == nil || excludedFlags[flag.Name] {
 			return
 		}
+		aliasOf, malformedAliasEvidence := captureFlagAliasOf(flag)
 		flags = append(flags, Flag{
-			Name:       flag.Name,
-			Shorthand:  flag.Shorthand,
-			Type:       flag.Value.Type(),
-			Default:    flag.DefValue,
-			NoOpt:      flag.NoOptDefVal,
-			Required:   isRequired(flag),
-			Hidden:     flag.Hidden,
-			Deprecated: strings.TrimSpace(flag.Deprecated),
+			Name:                   flag.Name,
+			Shorthand:              flag.Shorthand,
+			Type:                   flag.Value.Type(),
+			Default:                flag.DefValue,
+			NoOpt:                  flag.NoOptDefVal,
+			Required:               isRequired(flag),
+			Hidden:                 flag.Hidden,
+			Deprecated:             strings.TrimSpace(flag.Deprecated),
+			AliasOf:                aliasOf,
+			malformedAliasEvidence: malformedAliasEvidence,
 		})
 	})
 	sort.Slice(flags, func(i, j int) bool { return flags[i].Name < flags[j].Name })
 	return flags
 }
 
+func captureFlagAliasOf(flag *pflag.Flag) (string, bool) {
+	aliasOf, aliasMalformed := exactFlagAnnotation(flag, FlagAliasOfAnnotation)
+	origin, originMalformed := exactFlagAnnotation(flag, runtimeannotate.AnnotationFlagAliasOrigin)
+	if aliasOf == "" && origin == "" && !aliasMalformed && !originMalformed {
+		return "", false
+	}
+	malformed := aliasMalformed || originMalformed || aliasOf == "" ||
+		origin != runtimeannotate.FlagAliasOriginCorecmdV1
+	return aliasOf, malformed
+}
+
+func exactFlagAnnotation(flag *pflag.Flag, key string) (string, bool) {
+	values, exists := flag.Annotations[key]
+	if !exists {
+		return "", false
+	}
+	if len(values) != 1 || values[0] == "" || values[0] != strings.TrimSpace(values[0]) {
+		return "", true
+	}
+	return values[0], false
+}
+
 func isRequired(flag *pflag.Flag) bool {
 	for _, value := range flag.Annotations[cobra.BashCompOneRequiredFlag] {
 		if value == "true" {
@@ -268,3 +307,73 @@ func validateFlags(path, scope string, flags []Flag) error {
 	}
 	return nil
 }
+
+func validateFlagAliases(command Command) error {
+	flags := make(map[string]Flag, len(command.LocalFlags)+len(command.InheritedFlags))
+	ordered := make([]Flag, 0, len(command.LocalFlags)+len(command.InheritedFlags))
+	for _, flag := range command.LocalFlags {
+		flags[flag.Name] = flag
+		ordered = append(ordered, flag)
+	}
+	for _, flag := range command.InheritedFlags {
+		if _, exists := flags[flag.Name]; exists {
+			return fmt.Errorf("command %q contains flag %q in both local and inherited scopes", command.Path, flag.Name)
+		}
+		flags[flag.Name] = flag
+		ordered = append(ordered, flag)
+	}
+
+	for _, flag := range ordered {
+		if flag.malformedAliasEvidence {
+			return fmt.Errorf(
+				"command %q flag --%s must declare one exact %q value and framework origin %q",
+				command.Path,
+				flag.Name,
+				FlagAliasOfAnnotation,
+				runtimeannotate.FlagAliasOriginCorecmdV1,
+			)
+		}
+		if flag.AliasOf == "" {
+			continue
+		}
+		if !isExactFlagName(flag.AliasOf) {
+			return fmt.Errorf(
+				"command %q flag --%s has non-exact alias_of target %q",
+				command.Path,
+				flag.Name,
+				flag.AliasOf,
+			)
+		}
+		if flag.AliasOf == flag.Name {
+			return fmt.Errorf("command %q flag --%s cannot alias itself", command.Path, flag.Name)
+		}
+		target, exists := flags[flag.AliasOf]
+		if !exists {
+			return fmt.Errorf(
+				"command %q flag --%s aliases missing flag --%s",
+				command.Path,
+				flag.Name,
+				flag.AliasOf,
+			)
+		}
+		if target.AliasOf != "" {
+			return fmt.Errorf(
+				"command %q flag --%s aliases --%s, which is itself an alias",
+				command.Path,
+				flag.Name,
+				flag.AliasOf,
+			)
+		}
+		if flag.Type != target.Type {
+			return fmt.Errorf(
+				"command %q alias flags --%s and --%s have different types %q and %q",
+				command.Path,
+				flag.Name,
+				flag.AliasOf,
+				flag.Type,
+				target.Type,
+			)
+		}
+	}
+	return nil
+}

internal/interfacesnapshot/snapshot_test.go

@@ -15,9 +15,12 @@ package interfacesnapshot
 
 import (
 	"bytes"
+	"encoding/json"
 	"reflect"
 	"testing"
 
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/corecmd"
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/corecmd/runtimeannotate"
 	"github.com/spf13/cobra"
 )
 
@@ -95,7 +98,7 @@ func TestCrossPlatformCoverageCaptureUsesStableNoiseRulesAndFlagScopes(t *testin
 
 func TestCrossPlatformCoverageReadRejectsUnknownSnapshotFields(t *testing.T) {
 	input := bytes.NewBufferString(`{
-  "schema_version": 1,
+  "schema_version": 2,
   "rules": {"excluded_command_subtrees": [], "excluded_flags": []},
   "commands": [],
   "future_field": true
@@ -105,6 +108,166 @@ func TestCrossPlatformCoverageReadRejectsUnknownSnapshotFields(t *testing.T) {
 	}
 }
 
+func TestCrossPlatformCoverageCaptureRoundTripsFrameworkFlagAlias(t *testing.T) {
+	if FlagAliasOfAnnotation != runtimeannotate.AnnotationFlagAliasOf {
+		t.Fatalf(
+			"flag alias annotation key drifted: snapshot=%q runtime=%q",
+			FlagAliasOfAnnotation,
+			runtimeannotate.AnnotationFlagAliasOf,
+		)
+	}
+	root := &cobra.Command{Use: "dws"}
+	leaf := &cobra.Command{Use: "send", Run: func(*cobra.Command, []string) {}}
+	corecmd.RegisterFlags(leaf, []corecmd.FlagSpec{{
+		Name:    "message-id",
+		Usage:   "message ID",
+		Aliases: []string{"open-conversation-id"},
+	}})
+	root.AddCommand(leaf)
+
+	snapshot := Capture(root)
+	flag := flagIndex(commandIndex(snapshot)["dws send"].LocalFlags)["open-conversation-id"]
+	if flag.AliasOf != "message-id" {
+		t.Fatalf("captured alias_of = %q, want message-id", flag.AliasOf)
+	}
+
+	var encoded bytes.Buffer
+	if err := Write(&encoded, snapshot); err != nil {
+		t.Fatalf("Write: %v", err)
+	}
+	decoded, err := Read(bytes.NewReader(encoded.Bytes()))
+	if err != nil {
+		t.Fatalf("Read: %v", err)
+	}
+	decodedFlag := flagIndex(commandIndex(decoded)["dws send"].LocalFlags)["open-conversation-id"]
+	if decodedFlag.AliasOf != "message-id" {
+		t.Fatalf("round-tripped alias_of = %q, want message-id", decodedFlag.AliasOf)
+	}
+}
+
+func TestCrossPlatformCoverageCaptureRejectsUntrustedFlagAliasEvidence(t *testing.T) {
+	for _, origin := range []string{"", "handwritten.alias.v1"} {
+		t.Run(origin, func(t *testing.T) {
+			root := &cobra.Command{Use: "dws"}
+			leaf := &cobra.Command{Use: "send", Run: func(*cobra.Command, []string) {}}
+			leaf.Flags().String("message-id", "", "canonical message ID")
+			leaf.Flags().String("legacy-id", "", "legacy message ID")
+			legacy := leaf.Flags().Lookup("legacy-id")
+			legacy.Annotations = map[string][]string{
+				FlagAliasOfAnnotation: {"message-id"},
+			}
+			if origin != "" {
+				legacy.Annotations[runtimeannotate.AnnotationFlagAliasOrigin] = []string{origin}
+			}
+			root.AddCommand(leaf)
+
+			if err := Capture(root).Validate(); err == nil {
+				t.Fatal("Validate accepted alias evidence outside corecmd FlagSpec.Aliases")
+			}
+		})
+	}
+}
+
+func TestCrossPlatformCoverageCaptureRejectsMultipleNonEmptyFlagAliases(t *testing.T) {
+	root := &cobra.Command{Use: "dws"}
+	leaf := &cobra.Command{Use: "send", Run: func(*cobra.Command, []string) {}}
+	leaf.Flags().String("message-id", "", "canonical message ID")
+	leaf.Flags().String("other-id", "", "other ID")
+	leaf.Flags().String("legacy-id", "", "legacy ID")
+	legacy := leaf.Flags().Lookup("legacy-id")
+	legacy.Annotations = map[string][]string{
+		FlagAliasOfAnnotation: {"message-id", "other-id"},
+	}
+	root.AddCommand(leaf)
+
+	snapshot := Capture(root)
+	if err := snapshot.Validate(); err == nil {
+		t.Fatal("Validate accepted multiple non-empty alias annotation values")
+	}
+	var encoded bytes.Buffer
+	if err := Write(&encoded, snapshot); err == nil {
+		t.Fatal("Write accepted multiple non-empty alias annotation values")
+	}
+}
+
+func TestCrossPlatformCoverageValidateRejectsMalformedFlagAliasState(t *testing.T) {
+	tests := []struct {
+		name    string
+		aliasOf string
+	}{
+		{name: "self", aliasOf: "legacy-id"},
+		{name: "leading dashes", aliasOf: "--message-id"},
+		{name: "non exact", aliasOf: "message id"},
+		{name: "missing target", aliasOf: "missing-id"},
+	}
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			snapshot := testSnapshot(
+				testCommand("dws"),
+				testCommandWithFlagScopes(
+					"dws send",
+					[]Flag{
+						{Name: "legacy-id", Type: "string", AliasOf: test.aliasOf},
+						{Name: "message-id", Type: "string"},
+					},
+					nil,
+				),
+			)
+			if err := snapshot.Validate(); err == nil {
+				t.Fatalf("Validate accepted alias_of %q", test.aliasOf)
+			}
+			encoded, err := json.Marshal(snapshot)
+			if err != nil {
+				t.Fatalf("json.Marshal: %v", err)
+			}
+			if _, err := Read(bytes.NewReader(encoded)); err == nil {
+				t.Fatalf("Read accepted alias_of %q", test.aliasOf)
+			}
+		})
+	}
+}
+
+func TestCrossPlatformCoverageValidateRejectsFlagAliasChainTypeAndScopeConflicts(t *testing.T) {
+	tests := []struct {
+		name      string
+		local     []Flag
+		inherited []Flag
+	}{
+		{
+			name: "alias chain",
+			local: []Flag{
+				{Name: "legacy-id", Type: "string", AliasOf: "middle-id"},
+				{Name: "middle-id", Type: "string", AliasOf: "message-id"},
+				{Name: "message-id", Type: "string"},
+			},
+		},
+		{
+			name: "different types",
+			local: []Flag{
+				{Name: "legacy-id", Type: "string", AliasOf: "message-id"},
+				{Name: "message-id", Type: "int"},
+			},
+		},
+		{
+			name:      "same flag in local and inherited scopes",
+			local:     []Flag{{Name: "legacy-id", Type: "string"}},
+			inherited: []Flag{{Name: "legacy-id", Type: "string"}},
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			snapshot := testSnapshot(
+				testCommand("dws"),
+				testCommandWithFlagScopes("dws send", test.local, test.inherited),
+			)
+			if err := snapshot.Validate(); err == nil {
+				t.Fatalf("Validate accepted invalid alias contract: %#v", snapshot.Commands[1])
+			}
+		})
+	}
+}
+
 func TestCrossPlatformCoverageCompareBlocksCandidateSiblingAliasCollision(t *testing.T) {
 	base := testSnapshot(testCommand("dws"))
 	current := testSnapshot(

internal/shortcut/aitable/aitable.go

@@ -59,6 +59,51 @@ func parseJSONObject(flag, s string) (map[string]any, error) {
 	return m, nil
 }
 
+// resolveNamedList finds a list in a response envelope without conflating a
+// missing/malformed collection with a legitimate empty array. Some AITable MCP
+// versions use a direct collection key while others wrap it one level below
+// data/result, so all reviewed shapes are accepted but at least one explicit
+// JSON array must be present.
+func resolveNamedList(operation string, data map[string]any, outerKeys, innerKeys []string) ([]any, error) {
+	if data == nil {
+		return nil, fmt.Errorf("%s response is nil and does not establish a list result", operation)
+	}
+	invalid := make([]string, 0)
+	for _, key := range outerKeys {
+		value, exists := data[key]
+		if !exists {
+			continue
+		}
+		if list, ok := value.([]any); ok {
+			return list, nil
+		}
+		envelope, ok := value.(map[string]any)
+		if !ok {
+			invalid = append(invalid, fmt.Sprintf("%s=%T", key, value))
+			continue
+		}
+		foundInner := false
+		for _, innerKey := range innerKeys {
+			innerValue, innerExists := envelope[innerKey]
+			if !innerExists {
+				continue
+			}
+			foundInner = true
+			if list, ok := innerValue.([]any); ok {
+				return list, nil
+			}
+			invalid = append(invalid, fmt.Sprintf("%s.%s=%T", key, innerKey, innerValue))
+		}
+		if !foundInner {
+			invalid = append(invalid, fmt.Sprintf("%s has no recognized collection", key))
+		}
+	}
+	if len(invalid) > 0 {
+		return nil, fmt.Errorf("%s response has no valid list collection (%v)", operation, invalid)
+	}
+	return nil, fmt.Errorf("%s response is missing its list collection", operation)
+}
+
 // ─────────────────────────────────────────────────────────────
 // base: Base 管理（server: aitable）
 // ─────────────────────────────────────────────────────────────
@@ -116,7 +161,10 @@ var BaseList = shortcut.Shortcut{
 		if err != nil {
 			return err
 		}
-		bases := baseListProject(data)
+		bases, err := baseListProject("list_bases", data)
+		if err != nil {
+			return err
+		}
 		return rt.Output(map[string]any{"count": len(bases), "bases": bases})
 	},
 }
@@ -124,15 +172,19 @@ var BaseList = shortcut.Shortcut{
 // baseListProject reshapes list_bases / search_bases responses into a clean
 // {baseId, baseName} list — clean output projection. Both the list
 // container and the per-item field names are probed defensively across
-// candidate keys, so an empty/unknown shape yields an empty list rather than a
-// crash or fabricated data.
-func baseListProject(data map[string]any) []map[string]any {
-	raw := baseListResolveList(data)
+// candidate keys. An explicitly returned empty array is a successful empty
+// result; a missing/malformed collection is a protocol error and must not be
+// reported as count=0.
+func baseListProject(operation string, data map[string]any) ([]map[string]any, error) {
+	raw, err := baseListResolveList(operation, data)
+	if err != nil {
+		return nil, err
+	}
 	out := make([]map[string]any, 0, len(raw))
-	for _, item := range raw {
+	for index, item := range raw {
 		m, ok := item.(map[string]any)
 		if !ok {
-			continue
+			return nil, fmt.Errorf("%s response item %d must be an object, got %T", operation, index, item)
 		}
 		row := map[string]any{}
 		if v, ok := baseListFirst(m, "baseId", "base_id", "id"); ok {
@@ -141,36 +193,20 @@ func baseListProject(data map[string]any) []map[string]any {
 		if v, ok := baseListFirst(m, "baseName", "base_name", "name", "title"); ok {
 			row["baseName"] = v
 		}
-		if len(row) > 0 {
-			out = append(out, row)
+		if _, ok := row["baseId"]; !ok {
+			return nil, fmt.Errorf("%s response item %d is missing baseId", operation, index)
 		}
+		out = append(out, row)
 	}
-	return out
+	return out, nil
 }
 
 // baseListResolveList locates the list payload inside the response, tolerating a
 // bare top-level array container or nesting one level under a common envelope.
-func baseListResolveList(data map[string]any) []any {
-	if data == nil {
-		return []any{}
-	}
-	for _, key := range []string{"bases", "result", "data", "list", "items", "records"} {
-		v, ok := data[key]
-		if !ok {
-			continue
-		}
-		if arr, ok := v.([]any); ok {
-			return arr
-		}
-		if inner, ok := v.(map[string]any); ok {
-			for _, ik := range []string{"bases", "list", "items", "result", "data", "records"} {
-				if arr, ok := inner[ik].([]any); ok {
-					return arr
-				}
-			}
-		}
-	}
-	return []any{}
+func baseListResolveList(operation string, data map[string]any) ([]any, error) {
+	return resolveNamedList(operation, data,
+		[]string{"bases", "result", "data", "list", "items", "records"},
+		[]string{"bases", "list", "items", "result", "data", "records"})
 }
 
 // baseListFirst returns the first present candidate key's value.
@@ -230,7 +266,10 @@ var BaseSearch = shortcut.Shortcut{
 		if err != nil {
 			return err
 		}
-		bases := baseListProject(data)
+		bases, err := baseListProject("search_bases", data)
+		if err != nil {
+			return err
+		}
 		return rt.Output(map[string]any{"count": len(bases), "bases": bases})
 	},
 }
@@ -339,7 +378,7 @@ var BaseDelete = shortcut.Shortcut{
 		{Name: "base-id", Type: shortcut.FlagString, Desc: "待删除 Base ID", Required: true},
 		{Name: "reason", Type: shortcut.FlagString, Desc: "删除原因（可选）"},
 	},
-	Tips: []string{`dws aitable +base-delete --base-id BASE_ID --yes`},
+	Tips: []string{`dws aitable +base-delete --base-id BASE_ID`},
 	Execute: func(rt *shortcut.RuntimeContext) error {
 		params := map[string]any{"baseId": rt.Str("base-id")}
 		if rt.Changed("reason") {
@@ -474,7 +513,7 @@ var TableDelete = shortcut.Shortcut{
 		{Name: "table-id", Type: shortcut.FlagString, Desc: "待删除 Table ID", Required: true},
 		{Name: "reason", Type: shortcut.FlagString, Desc: "删除原因（可选）"},
 	},
-	Tips: []string{`dws aitable +table-delete --base-id B --table-id T --yes`},
+	Tips: []string{`dws aitable +table-delete --base-id B --table-id T`},
 	Execute: func(rt *shortcut.RuntimeContext) error {
 		params := map[string]any{
 			"baseId":  rt.Str("base-id"),
@@ -601,7 +640,7 @@ var FieldDelete = shortcut.Shortcut{
 		{Name: "table-id", Type: shortcut.FlagString, Desc: "Table ID", Required: true},
 		{Name: "field-id", Type: shortcut.FlagString, Desc: "待删除字段 ID", Required: true},
 	},
-	Tips: []string{`dws aitable +field-delete --base-id B --table-id T --field-id F --yes`},
+	Tips: []string{`dws aitable +field-delete --base-id B --table-id T --field-id F`},
 	Execute: func(rt *shortcut.RuntimeContext) error {
 		return rt.CallMCP("delete_field", map[string]any{
 			"baseId":  rt.Str("base-id"),
@@ -704,26 +743,16 @@ var RecordUpdate = shortcut.Shortcut{
 	Service:     "aitable",
 	Command:     "+record-update",
 	Product:     serverMain,
-	Description: "批量更新记录（每条含 recordId 与 cells），单次最多 100 条",
-	Intent:      "当你要批量修改已有行的字段值（如把一批任务标为已完成，每条须带 recordId）时使用；会实际覆盖对应单元格，单次最多 100 条。",
+	Description: "批量更新记录，自动按 100 条分片并逐批读回验证",
+	Intent:      "当你要批量修改已有行的字段值（每条须带 recordId）时使用；自动按 100 条分片，任一批失败或读回不一致即非零退出并给出 nextOffset。",
 	Risk:        shortcut.RiskWrite,
 	Flags: []shortcut.Flag{
 		{Name: "base-id", Type: shortcut.FlagString, Desc: "Base ID", Required: true},
 		{Name: "table-id", Type: shortcut.FlagString, Desc: "Table ID", Required: true},
 		{Name: "records", Type: shortcut.FlagString, Desc: "记录 JSON 数组，如 '[{\"recordId\":\"rec\",\"cells\":{...}}]'", Required: true},
 	},
-	Tips: []string{`dws aitable +record-update --base-id B --table-id T --records '[{"recordId":"rec","cells":{"fldStatusId":"已完成"}}]'`},
-	Execute: func(rt *shortcut.RuntimeContext) error {
-		recs, err := parseJSONAny("records", rt.Str("records"))
-		if err != nil {
-			return err
-		}
-		return rt.CallMCP("update_records", map[string]any{
-			"baseId":  rt.Str("base-id"),
-			"tableId": rt.Str("table-id"),
-			"records": recs,
-		})
-	},
+	Tips:    []string{`dws aitable +record-update --base-id B --table-id T --records '[{"recordId":"rec","cells":{"fldStatusId":"已完成"}}]'`},
+	Execute: executeRecordUpdateBatches,
 }
 
 // RecordDelete 删除行记录（delete_records）。
@@ -731,22 +760,16 @@ var RecordDelete = shortcut.Shortcut{
 	Service:     "aitable",
 	Command:     "+record-delete",
 	Product:     serverMain,
-	Description: "批量删除记录（不可逆），单次最多 100 条",
-	Intent:      "当你确认要批量删除若干行记录时使用；不可逆，按 recordId 列表删除，单次最多 100 条。",
+	Description: "批量删除记录（不可逆），自动按 100 条分片并逐批确认记录已不存在",
+	Intent:      "当你确认要批量删除若干行记录时使用；自动按 100 条分片，只有每批全部 recordId 读回均不存在才判定成功，否则非零退出并给出 nextOffset。",
 	Risk:        shortcut.RiskHighWrite,
 	Flags: []shortcut.Flag{
 		{Name: "base-id", Type: shortcut.FlagString, Desc: "Base ID", Required: true},
 		{Name: "table-id", Type: shortcut.FlagString, Desc: "Table ID", Required: true},
-		{Name: "record-ids", Type: shortcut.FlagStringSlice, Desc: "待删除记录 ID 列表，逗号分隔", Required: true},
-	},
-	Tips: []string{`dws aitable +record-delete --base-id B --table-id T --record-ids rec1,rec2 --yes`},
-	Execute: func(rt *shortcut.RuntimeContext) error {
-		return rt.CallMCP("delete_records", map[string]any{
-			"baseId":    rt.Str("base-id"),
-			"tableId":   rt.Str("table-id"),
-			"recordIds": rt.StrSlice("record-ids"),
-		})
+		{Name: "record-ids", Type: shortcut.FlagStringSlice, Desc: "待删除记录 ID 列表，逗号分隔；最多 10000 个唯一 ID", Required: true},
 	},
+	Tips:    []string{`dws aitable +record-delete --base-id B --table-id T --record-ids rec1,rec2`},
+	Execute: executeRecordDeleteBatches,
 }
 
 // RecordQueryEmpty 查询空行（query_empty_records，server: aitable-helper）。
@@ -919,26 +942,16 @@ var RecordUpsert = shortcut.Shortcut{
 	Service:     "aitable",
 	Command:     "+record-upsert",
 	Product:     serverHelper,
-	Description: "按 records 是否带 recordId 自动拆分 create / update",
-	Intent:      "当你有一批数据、其中部分是新增部分是更新、不想自己区分时使用；按记录是否带 recordId 自动拆分为创建或更新并实际写入，单次最多 100 条。",
+	Description: "按 recordId 自动拆分 create/update，按 100 条分片并读回验证",
+	Intent:      "当一批数据中部分新增、部分更新时使用；自动按 recordId 拆分并以 100 条为批次写入，创建必须返回新 recordId，全部批次都需读回验证。",
 	Risk:        shortcut.RiskWrite,
 	Flags: []shortcut.Flag{
 		{Name: "base-id", Type: shortcut.FlagString, Desc: "Base ID", Required: true},
 		{Name: "table-id", Type: shortcut.FlagString, Desc: "Table ID", Required: true},
 		{Name: "records", Type: shortcut.FlagString, Desc: "记录 JSON 数组，单次最多 100 条", Required: true},
 	},
-	Tips: []string{`dws aitable +record-upsert --base-id B --table-id T --records '[{"cells":{...}}]'`},
-	Execute: func(rt *shortcut.RuntimeContext) error {
-		recs, err := parseJSONAny("records", rt.Str("records"))
-		if err != nil {
-			return err
-		}
-		return rt.CallMCP("record_upsert", map[string]any{
-			"baseId":  rt.Str("base-id"),
-			"tableId": rt.Str("table-id"),
-			"records": recs,
-		})
-	},
+	Tips:    []string{`dws aitable +record-upsert --base-id B --table-id T --records '[{"cells":{...}}]'`},
+	Execute: executeRecordUpsertBatches,
 }
 
 // RecordPrimaryDocGet 查询记录主键文档（get_primary_doc，server: aitable-helper）。
@@ -1047,7 +1060,10 @@ var TemplateSearch = shortcut.Shortcut{
 		if err != nil {
 			return err
 		}
-		templates := templateSearchProject(data)
+		templates, err := templateSearchProject(data)
+		if err != nil {
+			return err
+		}
 		return rt.Output(map[string]any{"count": len(templates), "templates": templates})
 	},
 }
@@ -1055,15 +1071,17 @@ var TemplateSearch = shortcut.Shortcut{
 // templateSearchProject reshapes the raw search_templates response into a clean
 // {templateId, templateName} list — clean output projection. Both
 // the list container and the per-item field names are probed defensively across
-// candidate keys, so an empty/unknown shape yields an empty list rather than a
-// crash or fabricated data.
-func templateSearchProject(data map[string]any) []map[string]any {
-	raw := templateSearchResolveList(data)
+// candidate keys. Only an explicit array can establish a successful result.
+func templateSearchProject(data map[string]any) ([]map[string]any, error) {
+	raw, err := templateSearchResolveList(data)
+	if err != nil {
+		return nil, err
+	}
 	out := make([]map[string]any, 0, len(raw))
-	for _, item := range raw {
+	for index, item := range raw {
 		m, ok := item.(map[string]any)
 		if !ok {
-			continue
+			return nil, fmt.Errorf("search_templates response item %d must be an object, got %T", index, item)
 		}
 		row := map[string]any{}
 		if v, ok := templateSearchFirst(m, "templateId", "template_id", "id"); ok {
@@ -1072,37 +1090,21 @@ func templateSearchProject(data map[string]any) []map[string]any {
 		if v, ok := templateSearchFirst(m, "templateName", "template_name", "name", "title"); ok {
 			row["templateName"] = v
 		}
-		if len(row) > 0 {
-			out = append(out, row)
+		if _, ok := row["templateId"]; !ok {
+			return nil, fmt.Errorf("search_templates response item %d is missing templateId", index)
 		}
+		out = append(out, row)
 	}
-	return out
+	return out, nil
 }
 
 // templateSearchResolveList locates the list payload inside the response,
 // tolerating a bare top-level array container or nesting one level under a
 // common envelope key.
-func templateSearchResolveList(data map[string]any) []any {
-	if data == nil {
-		return []any{}
-	}
-	for _, key := range []string{"templates", "result", "data", "list", "items"} {
-		v, ok := data[key]
-		if !ok {
-			continue
-		}
-		if arr, ok := v.([]any); ok {
-			return arr
-		}
-		if inner, ok := v.(map[string]any); ok {
-			for _, ik := range []string{"templates", "list", "items", "result", "data"} {
-				if arr, ok := inner[ik].([]any); ok {
-					return arr
-				}
-			}
-		}
-	}
-	return []any{}
+func templateSearchResolveList(data map[string]any) ([]any, error) {
+	return resolveNamedList("search_templates", data,
+		[]string{"templates", "result", "data", "list", "items"},
+		[]string{"templates", "list", "items", "result", "data"})
 }
 
 // templateSearchFirst returns the first present candidate key's value.
@@ -1198,7 +1200,10 @@ var ViewGet = shortcut.Shortcut{
 		if err != nil {
 			return err
 		}
-		views := viewGetProject(data)
+		views, err := viewGetProject(data)
+		if err != nil {
+			return err
+		}
 		return rt.Output(map[string]any{"count": len(views), "views": views})
 	},
 }
@@ -1206,15 +1211,17 @@ var ViewGet = shortcut.Shortcut{
 // viewGetProject reshapes the raw get_views response into a clean
 // {viewId, viewName, viewType} list — clean output projection. Both
 // the list container and the per-item field names are probed defensively across
-// candidate keys, so an empty/unknown shape yields an empty list rather than a
-// crash or fabricated data.
-func viewGetProject(data map[string]any) []map[string]any {
-	raw := viewGetResolveList(data)
+// candidate keys. Only an explicit array can establish a successful result.
+func viewGetProject(data map[string]any) ([]map[string]any, error) {
+	raw, err := viewGetResolveList(data)
+	if err != nil {
+		return nil, err
+	}
 	out := make([]map[string]any, 0, len(raw))
-	for _, item := range raw {
+	for index, item := range raw {
 		m, ok := item.(map[string]any)
 		if !ok {
-			continue
+			return nil, fmt.Errorf("get_views response item %d must be an object, got %T", index, item)
 		}
 		row := map[string]any{}
 		if v, ok := viewGetFirst(m, "viewId", "view_id", "id"); ok {
@@ -1226,37 +1233,21 @@ func viewGetProject(data map[string]any) []map[string]any {
 		if v, ok := viewGetFirst(m, "viewType", "view_type", "type"); ok {
 			row["viewType"] = v
 		}
-		if len(row) > 0 {
-			out = append(out, row)
+		if _, ok := row["viewId"]; !ok {
+			return nil, fmt.Errorf("get_views response item %d is missing viewId", index)
 		}
+		out = append(out, row)
 	}
-	return out
+	return out, nil
 }
 
 // viewGetResolveList locates the list payload inside the response, tolerating a
 // bare top-level array container or nesting one level under a common envelope
 // key.
-func viewGetResolveList(data map[string]any) []any {
-	if data == nil {
-		return []any{}
-	}
-	for _, key := range []string{"views", "result", "data", "list", "items"} {
-		v, ok := data[key]
-		if !ok {
-			continue
-		}
-		if arr, ok := v.([]any); ok {
-			return arr
-		}
-		if inner, ok := v.(map[string]any); ok {
-			for _, ik := range []string{"views", "list", "items", "result", "data"} {
-				if arr, ok := inner[ik].([]any); ok {
-					return arr
-				}
-			}
-		}
-	}
-	return []any{}
+func viewGetResolveList(data map[string]any) ([]any, error) {
+	return resolveNamedList("get_views", data,
+		[]string{"views", "result", "data", "list", "items"},
+		[]string{"views", "list", "items", "result", "data"})
 }
 
 // viewGetFirst returns the first present candidate key's value.
@@ -1355,7 +1346,7 @@ var ViewDelete = shortcut.Shortcut{
 		{Name: "table-id", Type: shortcut.FlagString, Desc: "Table ID", Required: true},
 		{Name: "view-id", Type: shortcut.FlagString, Desc: "待删除 View ID", Required: true},
 	},
-	Tips: []string{`dws aitable +view-delete --base-id B --table-id T --view-id V --yes`},
+	Tips: []string{`dws aitable +view-delete --base-id B --table-id T --view-id V`},
 	Execute: func(rt *shortcut.RuntimeContext) error {
 		return rt.CallMCP("delete_view", map[string]any{
 			"baseId":  rt.Str("base-id"),
@@ -1668,23 +1659,28 @@ var FormList = shortcut.Shortcut{
 		if err != nil {
 			return err
 		}
-		forms := formListProject(data)
+		forms, err := formListProject(data)
+		if err != nil {
+			return err
+		}
 		return rt.Output(map[string]any{"count": len(forms), "forms": forms})
 	},
 }
 
 // formListProject reshapes the raw list_form_views response into a clean form
 // view list ({viewId, viewName, viewType}) — clean output projection.
-// Both the list container and the per-item field names are probed
-// defensively across candidate keys, so an empty/unknown shape yields an empty
-// list rather than a crash or fabricated data.
-func formListProject(data map[string]any) []map[string]any {
-	raw := formListResolveList(data)
+// Both the list container and the per-item field names are probed defensively
+// across candidate keys. Only an explicit array can establish success.
+func formListProject(data map[string]any) ([]map[string]any, error) {
+	raw, err := formListResolveList(data)
+	if err != nil {
+		return nil, err
+	}
 	out := make([]map[string]any, 0, len(raw))
-	for _, item := range raw {
+	for index, item := range raw {
 		m, ok := item.(map[string]any)
 		if !ok {
-			continue
+			return nil, fmt.Errorf("list_form_views response item %d must be an object, got %T", index, item)
 		}
 		row := map[string]any{}
 		if v, ok := formListFirst(m, "viewId", "view_id", "id"); ok {
@@ -1696,36 +1692,20 @@ func formListProject(data map[string]any) []map[string]any {
 		if v, ok := formListFirst(m, "viewType", "view_type", "type"); ok {
 			row["viewType"] = v
 		}
-		if len(row) > 0 {
-			out = append(out, row)
+		if _, ok := row["viewId"]; !ok {
+			return nil, fmt.Errorf("list_form_views response item %d is missing viewId", index)
 		}
+		out = append(out, row)
 	}
-	return out
+	return out, nil
 }
 
 // formListResolveList locates the list payload inside the response, tolerating a
 // bare top-level array container or nesting one level under a common envelope.
-func formListResolveList(data map[string]any) []any {
-	if data == nil {
-		return []any{}
-	}
-	for _, key := range []string{"forms", "views", "formViews", "result", "data", "list", "items"} {
-		v, ok := data[key]
-		if !ok {
-			continue
-		}
-		if arr, ok := v.([]any); ok {
-			return arr
-		}
-		if inner, ok := v.(map[string]any); ok {
-			for _, ik := range []string{"forms", "views", "formViews", "list", "items", "result", "data"} {
-				if arr, ok := inner[ik].([]any); ok {
-					return arr
-				}
-			}
-		}
-	}
-	return []any{}
+func formListResolveList(data map[string]any) ([]any, error) {
+	return resolveNamedList("list_form_views", data,
+		[]string{"forms", "views", "formViews", "result", "data", "list", "items"},
+		[]string{"forms", "views", "formViews", "list", "items", "result", "data"})
 }
 
 // formListFirst returns the first present candidate key's value.
@@ -1751,7 +1731,7 @@ var FormDelete = shortcut.Shortcut{
 		{Name: "table-id", Type: shortcut.FlagString, Desc: "Table ID", Required: true},
 		{Name: "view-id", Type: shortcut.FlagString, Desc: "表单 View ID", Required: true},
 	},
-	Tips: []string{`dws aitable +form-delete --base-id B --table-id T --view-id V --yes`},
+	Tips: []string{`dws aitable +form-delete --base-id B --table-id T --view-id V`},
 	Execute: func(rt *shortcut.RuntimeContext) error {
 		return rt.CallMCP("delete_form_view", map[string]any{
 			"baseId":  rt.Str("base-id"),
@@ -2013,7 +1993,7 @@ var WorkflowDisable = shortcut.Shortcut{
 		{Name: "base-id", Type: shortcut.FlagString, Desc: "Base ID", Required: true},
 		{Name: "workflow-id", Type: shortcut.FlagString, Desc: "Workflow ID", Required: true},
 	},
-	Tips: []string{`dws aitable +workflow-disable --base-id B --workflow-id W --yes`},
+	Tips: []string{`dws aitable +workflow-disable --base-id B --workflow-id W`},
 	Execute: func(rt *shortcut.RuntimeContext) error {
 		return rt.CallMCP("disable_workflow", map[string]any{
 			"baseId":     rt.Str("base-id"),
@@ -2069,23 +2049,27 @@ var WorkflowList = shortcut.Shortcut{
 		if err != nil {
 			return err
 		}
-		workflows := workflowListProject(data)
+		workflows, err := workflowListProject(data)
+		if err != nil {
+			return err
+		}
 		return rt.Output(map[string]any{"count": len(workflows), "workflows": workflows})
 	},
 }
 
 // workflowListProject reshapes the raw list_workflows response into a clean
 // workflow list ({workflowId, name, status/enabled}) — output-projection
-// clean output projection. Both the list container and the per-item field names are
-// probed defensively across candidate keys, so an empty/unknown shape yields an
-// empty list rather than a crash or fabricated data.
-func workflowListProject(data map[string]any) []map[string]any {
-	raw := workflowListResolveList(data)
+// clean output projection. Only an explicit array can establish success.
+func workflowListProject(data map[string]any) ([]map[string]any, error) {
+	raw, err := workflowListResolveList(data)
+	if err != nil {
+		return nil, err
+	}
 	out := make([]map[string]any, 0, len(raw))
-	for _, item := range raw {
+	for index, item := range raw {
 		m, ok := item.(map[string]any)
 		if !ok {
-			continue
+			return nil, fmt.Errorf("list_workflows response item %d must be an object, got %T", index, item)
 		}
 		row := map[string]any{}
 		if v, ok := workflowListFirst(m, "workflowId", "workflow_id", "id"); ok {
@@ -2097,37 +2081,21 @@ func workflowListProject(data map[string]any) []map[string]any {
 		if v, ok := workflowListFirst(m, "status", "enabled", "isEnabled", "state"); ok {
 			row["status"] = v
 		}
-		if len(row) > 0 {
-			out = append(out, row)
+		if _, ok := row["workflowId"]; !ok {
+			return nil, fmt.Errorf("list_workflows response item %d is missing workflowId", index)
 		}
+		out = append(out, row)
 	}
-	return out
+	return out, nil
 }
 
 // workflowListResolveList locates the list payload inside the response,
 // tolerating a bare top-level array container or nesting one level under a
 // common envelope.
-func workflowListResolveList(data map[string]any) []any {
-	if data == nil {
-		return []any{}
-	}
-	for _, key := range []string{"workflows", "result", "data", "list", "items"} {
-		v, ok := data[key]
-		if !ok {
-			continue
-		}
-		if arr, ok := v.([]any); ok {
-			return arr
-		}
-		if inner, ok := v.(map[string]any); ok {
-			for _, ik := range []string{"workflows", "list", "items", "result", "data"} {
-				if arr, ok := inner[ik].([]any); ok {
-					return arr
-				}
-			}
-		}
-	}
-	return []any{}
+func workflowListResolveList(data map[string]any) ([]any, error) {
+	return resolveNamedList("list_workflows", data,
+		[]string{"workflows", "result", "data", "list", "items"},
+		[]string{"workflows", "list", "items", "result", "data"})
 }
 
 // workflowListFirst returns the first present candidate key's value.
@@ -2281,7 +2249,7 @@ var DashboardDelete = shortcut.Shortcut{
 		{Name: "dashboard-id", Type: shortcut.FlagString, Desc: "Dashboard ID", Required: true},
 		{Name: "reason", Type: shortcut.FlagString, Desc: "删除原因（可选）"},
 	},
-	Tips: []string{`dws aitable +dashboard-delete --base-id B --dashboard-id D --yes`},
+	Tips: []string{`dws aitable +dashboard-delete --base-id B --dashboard-id D`},
 	Execute: func(rt *shortcut.RuntimeContext) error {
 		params := map[string]any{
 			"baseId":      rt.Str("base-id"),
@@ -2512,7 +2480,7 @@ var ChartDelete = shortcut.Shortcut{
 		{Name: "chart-id", Type: shortcut.FlagString, Desc: "Chart ID", Required: true},
 		{Name: "reason", Type: shortcut.FlagString, Desc: "删除原因（可选）"},
 	},
-	Tips: []string{`dws aitable +chart-delete --base-id B --dashboard-id D --chart-id C --yes`},
+	Tips: []string{`dws aitable +chart-delete --base-id B --dashboard-id D --chart-id C`},
 	Execute: func(rt *shortcut.RuntimeContext) error {
 		params := map[string]any{
 			"baseId":      rt.Str("base-id"),
@@ -2744,7 +2712,7 @@ var AdvpermDisable = shortcut.Shortcut{
 	Flags: []shortcut.Flag{
 		{Name: "base-id", Type: shortcut.FlagString, Desc: "Base ID", Required: true},
 	},
-	Tips: []string{`dws aitable +advperm-disable --base-id B --yes`},
+	Tips: []string{`dws aitable +advperm-disable --base-id B`},
 	Execute: func(rt *shortcut.RuntimeContext) error {
 		return rt.CallMCP("set_advanced_permission", map[string]any{
 			"baseId":  rt.Str("base-id"),
@@ -2908,7 +2876,7 @@ var RoleDelete = shortcut.Shortcut{
 		{Name: "base-id", Type: shortcut.FlagString, Desc: "Base ID", Required: true},
 		{Name: "role-id", Type: shortcut.FlagString, Desc: "Role ID（数字 long 字符串）", Required: true},
 	},
-	Tips: []string{`dws aitable +role-delete --base-id B --role-id R --yes`},
+	Tips: []string{`dws aitable +role-delete --base-id B --role-id R`},
 	Execute: func(rt *shortcut.RuntimeContext) error {
 		return rt.CallMCP("delete_role", map[string]any{
 			"baseId": rt.Str("base-id"),
@@ -2986,7 +2954,7 @@ var SectionDelete = shortcut.Shortcut{
 		{Name: "base-id", Type: shortcut.FlagString, Desc: "Base ID", Required: true},
 		{Name: "section-id", Type: shortcut.FlagString, Desc: "Section ID", Required: true},
 	},
-	Tips: []string{`dws aitable +section-delete --base-id B --section-id S --yes`},
+	Tips: []string{`dws aitable +section-delete --base-id B --section-id S`},
 	Execute: func(rt *shortcut.RuntimeContext) error {
 		return rt.CallMCP("delete_section", map[string]any{
 			"baseId":    rt.Str("base-id"),
@@ -3131,7 +3099,7 @@ var SectionMoveNode = shortcut.Shortcut{
 }
 
 func init() {
-	shortcut.Register(
+	shortcut.Register(withReviewedAITableShortcutContracts(
 		BaseList,
 		BaseSearch,
 		BaseGet,
@@ -3139,7 +3107,10 @@ func init() {
 		BaseUpdate,
 		BaseDelete,
 		BaseCopy,
+		BaseSchemaSnapshot,
+		BaseBootstrap,
 		TableGet,
+		TableCopy,
 		TableUpdate,
 		TableDelete,
 		FieldGet,
@@ -3152,12 +3123,17 @@ func init() {
 		RecordHistoryList,
 		RecordShareURL,
 		RecordUpsert,
+		RecordUpsertByKey,
+		RecordBulkPatch,
 		RecordPrimaryDocGet,
 		RecordPrimaryDocCreate,
 		TemplateSearch,
 		AttachmentUpload,
+		AttachmentPut,
+		AttachmentRemove,
 		ViewGet,
 		ViewUpdate,
+		ViewPresetApply,
 		ViewDuplicate,
 		ViewDelete,
 		ViewGetLock,
@@ -3179,6 +3155,7 @@ func init() {
 		WorkflowDisable,
 		WorkflowGet,
 		WorkflowList,
+		WorkflowDeploy,
 		DashboardConfigExample,
 		DashboardGet,
 		DashboardUpdate,
@@ -3209,5 +3186,6 @@ func init() {
 		SectionListEmpty,
 		SectionListNodes,
 		SectionMoveNode,
-	)
+		URLResolve,
+	)...)
 }

internal/shortcut/aitable/attachment_composite.go

@@ -0,0 +1,372 @@
+// Copyright 2026 Alibaba Group
+// SPDX-License-Identifier: Apache-2.0
+
+package aitable
+
+import (
+	"fmt"
+	"io"
+	"mime"
+	"net"
+	"net/http"
+	"net/url"
+	"os"
+	"path/filepath"
+	"strings"
+
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/corecmd/contract"
+	apperrors "github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/errors"
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/shortcut"
+)
+
+const maxAttachmentFileSize = int64(100 * 1024 * 1024)
+
+func rejectAttachmentUploadRedirect(*http.Request, []*http.Request) error {
+	// Upload credentials are issued for one exact endpoint. Do not let a 3xx
+	// response redirect local file bytes or the credential to another host or
+	// to a URL outside validateAttachmentUploadURL's transport policy.
+	return http.ErrUseLastResponse
+}
+
+var attachmentHTTPDo = (&http.Client{CheckRedirect: rejectAttachmentUploadRedirect}).Do
+
+var AttachmentPut = shortcut.Shortcut{
+	Service:     "aitable",
+	Command:     "+attachment-put",
+	Product:     serverMain,
+	Description: "准备凭证、实际 PUT 本地文件、写入 attachment 单元格并读回验证",
+	Intent:      "当你要把本地文件真正上传并落到指定记录附件字段时使用；不是只返回 uploadUrl，支持 replace 或在现有项可安全重写时 append。",
+	Risk:        shortcut.RiskWrite,
+	Safety: contract.SafetySpec{
+		Effect: "write", Risk: "medium", Confirmation: "user_required", Idempotency: "non_idempotent",
+	},
+	Contract: aitableCompositeContract(
+		"+attachment-put",
+		"准备凭证、实际 PUT 本地文件、写入 attachment 单元格并读回验证",
+		"当你要把本地文件真正上传并落到指定记录附件字段时使用；不是只返回 uploadUrl，支持 replace 或在现有项可安全重写时 append。",
+		"只申请上传凭证用 attachment upload；不要用 drive fileId；现有附件读回缺 fileToken 时无法安全 append，只能 replace 或停止",
+		`dws aitable +attachment-put --base-id B --table-id T --record-id R --field-id F --file ./report.pdf --mode replace`,
+	),
+	Flags: []shortcut.Flag{
+		{Name: "base-id", Type: shortcut.FlagString, Desc: "Base ID", Required: true},
+		{Name: "table-id", Type: shortcut.FlagString, Desc: "Table ID", Required: true},
+		{Name: "record-id", Type: shortcut.FlagString, Desc: "Record ID", Required: true},
+		{Name: "field-id", Type: shortcut.FlagString, Desc: "attachment Field ID", Required: true},
+		{Name: "file", Type: shortcut.FlagString, Desc: "本地文件路径", Required: true},
+		{Name: "mode", Type: shortcut.FlagString, Default: "append", Desc: "append 保留现有附件；replace 整体替换", Enum: []string{"append", "replace"}},
+		{Name: "mime-type", Type: shortcut.FlagString, Desc: "覆盖自动推断的 MIME type（可选）"},
+	},
+	Tips: []string{`dws aitable +attachment-put --base-id B --table-id T --record-id R --field-id F --file ./report.pdf --mode replace`},
+	Execute: func(rt *shortcut.RuntimeContext) error {
+		return executeAttachmentPut(rt)
+	},
+}
+
+var AttachmentRemove = shortcut.Shortcut{
+	Service:     "aitable",
+	Command:     "+attachment-remove",
+	Product:     serverMain,
+	Description: "从 attachment 字段清空全部或按文件名移除，写前确保剩余项具有可重写 fileToken，并读回验证",
+	Intent:      "当你要清空附件字段，或按精确文件名移除且安全保留其他附件时使用；读回若不给剩余项 fileToken 会在写前停止并说明服务边界。",
+	Risk:        shortcut.RiskHighWrite,
+	Safety: contract.SafetySpec{
+		Effect: "write", Risk: "high", Confirmation: "user_required", Idempotency: "idempotent",
+	},
+	Contract: aitableCompositeContract(
+		"+attachment-remove",
+		"从 attachment 字段清空全部或按文件名移除，写前确保剩余项具有可重写 fileToken，并读回验证",
+		"当你要清空附件字段，或按精确文件名移除且安全保留其他附件时使用；读回若不给剩余项 fileToken 会在写前停止并说明服务边界。",
+		"读回仅有下载 URL 而没有剩余附件 fileToken 时，DWS 无法在覆盖式写接口上安全对齐逐项删除；此时只能 --clear-all 或停止",
+		`dws aitable +attachment-remove --base-id B --table-id T --record-id R --field-id F --clear-all`,
+	),
+	Flags: []shortcut.Flag{
+		{Name: "base-id", Type: shortcut.FlagString, Desc: "Base ID", Required: true},
+		{Name: "table-id", Type: shortcut.FlagString, Desc: "Table ID", Required: true},
+		{Name: "record-id", Type: shortcut.FlagString, Desc: "Record ID", Required: true},
+		{Name: "field-id", Type: shortcut.FlagString, Desc: "attachment Field ID", Required: true},
+		{Name: "remove-name", Type: shortcut.FlagString, Desc: "移除精确文件名的所有匹配项；与 --clear-all 二选一"},
+		{Name: "clear-all", Type: shortcut.FlagBool, Desc: "清空该字段全部附件；与 --remove-name 二选一"},
+	},
+	Tips: []string{`dws aitable +attachment-remove --base-id B --table-id T --record-id R --field-id F --clear-all`},
+	Execute: func(rt *shortcut.RuntimeContext) error {
+		return executeAttachmentRemove(rt)
+	},
+}
+
+func executeAttachmentPut(rt *shortcut.RuntimeContext) error {
+	file, info, mimeType, err := openAttachmentFile(rt.Str("file"), rt.Str("mime-type"))
+	if err != nil {
+		return err
+	}
+	defer file.Close()
+	baseID, tableID, recordIDValue, fieldID := rt.Str("base-id"), rt.Str("table-id"), rt.Str("record-id"), rt.Str("field-id")
+	record, existing, err := readAttachmentCell(rt, baseID, tableID, recordIDValue, fieldID)
+	if err != nil {
+		return err
+	}
+	_ = record
+	desired := make([]any, 0, len(existing)+1)
+	if rt.Str("mode") == "append" {
+		for index, item := range existing {
+			token := attachmentToken(item)
+			if token == "" {
+				return apperrors.NewValidation(fmt.Sprintf("现有附件[%d]读回缺少 fileToken，覆盖式写接口无法安全 append；请用 --mode replace 或保留原数据后人工处理", index),
+					apperrors.WithReason("attachment_tokens_unavailable"), apperrors.WithExecutionStarted(false))
+			}
+			desired = append(desired, map[string]any{"fileToken": token})
+		}
+	}
+	result := newCompositeResult("attachment_put")
+	result.Resolved = map[string]any{"baseId": baseID, "tableId": tableID, "recordId": recordIDValue, "fieldId": fieldID, "fileName": info.Name(), "size": info.Size(), "mode": rt.Str("mode")}
+	result.Plan = []compositeStep{
+		{Index: 1, Name: "prepare attachment upload", Tool: "prepare_attachment_upload", Status: "planned"},
+		{Index: 2, Name: "PUT file bytes", Tool: "HTTP PUT", Status: "planned"},
+		{Index: 3, Name: "write and verify attachment cell", Tool: "update_records", Status: "planned"},
+	}
+	if rt.DryRun() {
+		result.Status = "planned"
+		result.Executed = false
+		return rt.Output(result)
+	}
+	prepareData, err := rt.CallMCPWriteDataStrict(serverMain, "prepare_attachment_upload", map[string]any{"baseId": baseID, "fileName": info.Name(), "size": info.Size(), "mimeType": mimeType})
+	if err != nil {
+		result.Status = "unknown"
+		return compositeError(result, err, false)
+	}
+	uploadURL := findStringByKeys(prepareData, "uploadUrl")
+	fileToken := findStringByKeys(prepareData, "fileToken")
+	if uploadURL == "" || fileToken == "" {
+		result.Status = "unknown"
+		return compositeError(result, fmt.Errorf("prepare_attachment_upload response is missing uploadUrl or fileToken"), false)
+	}
+	if err := validateAttachmentUploadURL(uploadURL); err != nil {
+		result.Status = "unknown"
+		return compositeError(result, err, false)
+	}
+	// validateAttachmentUploadURL has already rejected every URL shape for
+	// which NewRequestWithContext can fail; PUT is a fixed valid method.
+	request, _ := http.NewRequestWithContext(rt.Command().Context(), http.MethodPut, uploadURL, file)
+	request.Header.Set("Content-Type", mimeType)
+	request.ContentLength = info.Size()
+	response, err := attachmentHTTPDo(request)
+	if err != nil {
+		result.Status = "partial_success"
+		result.KnownEffects = append(result.KnownEffects, map[string]any{"tool": "prepare_attachment_upload", "fileToken": fileToken})
+		return compositeError(result, fmt.Errorf("attachment HTTP PUT failed: %w", err), false)
+	}
+	_, _ = io.Copy(io.Discard, io.LimitReader(response.Body, 1<<20))
+	closeErr := response.Body.Close()
+	if response.StatusCode < 200 || response.StatusCode >= 300 || closeErr != nil {
+		result.Status = "partial_success"
+		result.KnownEffects = append(result.KnownEffects, map[string]any{"tool": "prepare_attachment_upload", "fileToken": fileToken, "httpStatus": response.StatusCode})
+		return compositeError(result, fmt.Errorf("attachment HTTP PUT returned status %d (close error: %v)", response.StatusCode, closeErr), false)
+	}
+	desired = append(desired, map[string]any{"fileToken": fileToken})
+	result.KnownEffects = append(result.KnownEffects, map[string]any{"tool": "HTTP PUT", "fileToken": fileToken, "fileName": info.Name(), "size": info.Size()})
+	_, writeErr := rt.CallMCPWriteDataStrict(serverMain, "update_records", map[string]any{
+		"baseId": baseID, "tableId": tableID,
+		"records": []any{map[string]any{"recordId": recordIDValue, "cells": map[string]any{fieldID: desired}}},
+	})
+	_, readBack, verifyErr := readAttachmentCell(rt, baseID, tableID, recordIDValue, fieldID)
+	if verifyErr == nil {
+		verifyErr = verifyAttachmentPut(readBack, len(desired), fileToken, info.Name(), info.Size())
+	}
+	if verifyErr != nil {
+		result.Status = "partial_success"
+		result.KnownEffects = append(result.KnownEffects, map[string]any{"tool": "update_records", "recordId": recordIDValue, "fieldId": fieldID})
+		if writeErr != nil {
+			result.Warnings = append(result.Warnings, "record write response error: "+writeErr.Error())
+		}
+		return compositeError(result, verifyErr, false)
+	}
+	result.CompletedCount = 1
+	result.Verification = map[string]any{"status": "verified", "attachmentCount": len(readBack), "fileName": info.Name(), "size": info.Size()}
+	result.Result = map[string]any{"fileToken": fileToken, "fileName": info.Name(), "size": info.Size(), "mimeType": mimeType, "attachmentCount": len(readBack)}
+	if writeErr != nil {
+		result.Status = "recovered"
+		result.Warnings = append(result.Warnings, "record write response was an error, but the attachment cell was proven by read-back")
+	}
+	return rt.Output(result)
+}
+
+func executeAttachmentRemove(rt *shortcut.RuntimeContext) error {
+	removeName := strings.TrimSpace(rt.Str("remove-name"))
+	clearAll := rt.Bool("clear-all")
+	if (removeName == "") == !clearAll {
+		return apperrors.NewValidation("必须且只能提供 --remove-name 或 --clear-all")
+	}
+	baseID, tableID, recordIDValue, fieldID := rt.Str("base-id"), rt.Str("table-id"), rt.Str("record-id"), rt.Str("field-id")
+	_, existing, err := readAttachmentCell(rt, baseID, tableID, recordIDValue, fieldID)
+	if err != nil {
+		return err
+	}
+	desired := make([]any, 0, len(existing))
+	removed := 0
+	if clearAll {
+		removed = len(existing)
+	} else {
+		for index, item := range existing {
+			if attachmentName(item) == removeName {
+				removed++
+				continue
+			}
+			token := attachmentToken(item)
+			if token == "" {
+				return apperrors.NewValidation(fmt.Sprintf("剩余附件[%d]读回缺少 fileToken，覆盖式写接口无法安全保留后删除；服务端当前不支持按附件增量删除", index),
+					apperrors.WithReason("attachment_tokens_unavailable"), apperrors.WithExecutionStarted(false))
+			}
+			desired = append(desired, map[string]any{"fileToken": token})
+		}
+	}
+	result := newCompositeResult("attachment_remove")
+	result.RequestedCount = removed
+	result.Resolved = map[string]any{"baseId": baseID, "tableId": tableID, "recordId": recordIDValue, "fieldId": fieldID, "removeName": removeName, "clearAll": clearAll}
+	if removed == 0 {
+		result.Status = "unchanged"
+		result.Executed = false
+		result.Verification = map[string]any{"status": "verified", "attachmentCount": len(existing), "removedCount": 0}
+		return rt.Output(result)
+	}
+	result.Plan = []compositeStep{{Index: 1, Name: "replace attachment cell without selected items", Tool: "update_records", Status: "planned", Count: removed}}
+	if rt.DryRun() {
+		result.Status = "planned"
+		result.Executed = false
+		return rt.Output(result)
+	}
+	_, writeErr := rt.CallMCPWriteDataStrict(serverMain, "update_records", map[string]any{
+		"baseId": baseID, "tableId": tableID,
+		"records": []any{map[string]any{"recordId": recordIDValue, "cells": map[string]any{fieldID: desired}}},
+	})
+	_, readBack, verifyErr := readAttachmentCell(rt, baseID, tableID, recordIDValue, fieldID)
+	if verifyErr == nil && len(readBack) != len(desired) {
+		verifyErr = fmt.Errorf("attachment read-back count is %d, want %d", len(readBack), len(desired))
+	}
+	if verifyErr == nil && removeName != "" {
+		for _, item := range readBack {
+			if attachmentName(item) == removeName {
+				verifyErr = fmt.Errorf("attachment %q is still present after removal", removeName)
+				break
+			}
+		}
+	}
+	if verifyErr != nil {
+		result.Status = "unknown"
+		if writeErr != nil {
+			result.Warnings = append(result.Warnings, "record write response error: "+writeErr.Error())
+		}
+		return compositeError(result, verifyErr, true)
+	}
+	result.CompletedCount = removed
+	result.Verification = map[string]any{"status": "verified", "removedCount": removed, "attachmentCount": len(readBack)}
+	result.Result = map[string]any{"removedCount": removed, "attachmentCount": len(readBack)}
+	if writeErr != nil {
+		result.Status = "recovered"
+		result.Warnings = append(result.Warnings, "record write response was an error, but removal was proven by read-back")
+	}
+	return rt.Output(result)
+}
+
+func openAttachmentFile(path, overrideMIME string) (*os.File, os.FileInfo, string, error) {
+	file, err := os.Open(path)
+	if err != nil {
+		return nil, nil, "", apperrors.NewValidation("无法打开 --file: " + err.Error())
+	}
+	info, err := file.Stat()
+	if err != nil || !info.Mode().IsRegular() || info.Size() <= 0 || info.Size() > maxAttachmentFileSize {
+		file.Close()
+		return nil, nil, "", apperrors.NewValidation(fmt.Sprintf("--file 必须是 1..%d 字节的普通文件", maxAttachmentFileSize))
+	}
+	mimeType := strings.TrimSpace(overrideMIME)
+	if mimeType == "" {
+		mimeType = mime.TypeByExtension(strings.ToLower(filepath.Ext(info.Name())))
+	}
+	if mimeType == "" {
+		header := make([]byte, 512)
+		read, _ := file.Read(header)
+		mimeType = http.DetectContentType(header[:read])
+		_, _ = file.Seek(0, io.SeekStart)
+	}
+	return file, info, mimeType, nil
+}
+
+func validateAttachmentUploadURL(raw string) error {
+	parsed, err := url.Parse(raw)
+	if err != nil || parsed.Host == "" || parsed.User != nil || (parsed.Scheme != "https" && parsed.Scheme != "http") {
+		return fmt.Errorf("prepare_attachment_upload returned an invalid uploadUrl")
+	}
+	if parsed.Scheme == "http" {
+		host := parsed.Hostname()
+		ip := net.ParseIP(host)
+		if host != "localhost" && (ip == nil || !ip.IsLoopback()) {
+			return fmt.Errorf("plaintext attachment uploadUrl is allowed only for loopback test servers")
+		}
+	}
+	return nil
+}
+
+func readAttachmentCell(rt *shortcut.RuntimeContext, baseID, tableID, recordIDValue, fieldID string) (map[string]any, []map[string]any, error) {
+	records, err := queryRecordsByIDs(rt, baseID, tableID, []string{recordIDValue})
+	if err != nil {
+		return nil, nil, err
+	}
+	if len(records) != 1 || recordID(records[0]) != recordIDValue {
+		return nil, nil, fmt.Errorf("attachment read-back returned %d exact records for %s, want 1", len(records), recordIDValue)
+	}
+	cells, ok := records[0]["cells"].(map[string]any)
+	if !ok {
+		return nil, nil, fmt.Errorf("record %s is missing cells", recordIDValue)
+	}
+	raw, exists := cells[fieldID]
+	if !exists || raw == nil {
+		return records[0], []map[string]any{}, nil
+	}
+	list, ok := raw.([]any)
+	if !ok {
+		return nil, nil, fmt.Errorf("attachment field %s is not an array", fieldID)
+	}
+	out := make([]map[string]any, 0, len(list))
+	for index, item := range list {
+		object, ok := item.(map[string]any)
+		if !ok {
+			return nil, nil, fmt.Errorf("attachment field %s item %d is not an object", fieldID, index)
+		}
+		out = append(out, object)
+	}
+	return records[0], out, nil
+}
+
+func attachmentToken(item map[string]any) string {
+	return stringValue(item, "fileToken", "file_token")
+}
+
+func attachmentName(item map[string]any) string {
+	return stringValue(item, "fileName", "filename", "name")
+}
+
+func verifyAttachmentPut(actual []map[string]any, expectedCount int, token, name string, size int64) error {
+	if len(actual) != expectedCount {
+		return fmt.Errorf("attachment read-back count is %d, want %d", len(actual), expectedCount)
+	}
+	for _, item := range actual {
+		if attachmentToken(item) == token {
+			return nil
+		}
+		itemSize, sizeOK := numericInt64(item["size"])
+		if attachmentName(item) == name && sizeOK && itemSize == size {
+			return nil
+		}
+	}
+	return fmt.Errorf("attachment read-back does not identify uploaded token or file %s/%d", name, size)
+}
+
+func numericInt64(value any) (int64, bool) {
+	switch typed := value.(type) {
+	case float64:
+		return int64(typed), typed == float64(int64(typed))
+	case int:
+		return int64(typed), true
+	case int64:
+		return typed, true
+	}
+	return 0, false
+}

internal/shortcut/aitable/attachment_composite_e2e_test.go

@@ -0,0 +1,461 @@
+// Copyright 2026 Alibaba Group
+// SPDX-License-Identifier: Apache-2.0
+
+package aitable
+
+import (
+	"errors"
+	"io"
+	"net/http"
+	"net/http/httptest"
+	"os"
+	"path/filepath"
+	"strings"
+	"testing"
+
+	apperrors "github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/errors"
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/testseam"
+)
+
+func writeAttachmentFixture(t *testing.T, content string) string {
+	t.Helper()
+	path := filepath.Join(t.TempDir(), "report.txt")
+	if err := os.WriteFile(path, []byte(content), 0o600); err != nil {
+		t.Fatal(err)
+	}
+	return path
+}
+
+func attachmentRecordJSON(t *testing.T, fieldID string, attachments []any) string {
+	t.Helper()
+	return mustJSONText(t, map[string]any{"records": []any{map[string]any{
+		"recordId": "record", "cells": map[string]any{fieldID: attachments},
+	}}})
+}
+
+func TestCrossPlatformCoverageAttachmentPutPerformsHTTPAndReadBackE2E(t *testing.T) {
+	filePath := writeAttachmentFixture(t, "actual attachment bytes")
+	fileName := filepath.Base(filePath)
+	uploaded := false
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, request *http.Request) {
+		body, _ := io.ReadAll(request.Body)
+		if request.Method != http.MethodPut || request.Header.Get("Content-Type") != "text/plain" || string(body) != "actual attachment bytes" {
+			t.Errorf("upload request = method:%s type:%s body:%q", request.Method, request.Header.Get("Content-Type"), body)
+			http.Error(w, "bad upload", http.StatusBadRequest)
+			return
+		}
+		uploaded = true
+		w.WriteHeader(http.StatusOK)
+	}))
+	defer server.Close()
+	caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+		{text: attachmentRecordJSON(t, "field", []any{})},
+		{text: mustJSONText(t, map[string]any{"uploadUrl": server.URL + "/put", "fileToken": "ft1"})},
+		{text: ""},
+		{text: attachmentRecordJSON(t, "field", []any{map[string]any{"filename": fileName, "size": len("actual attachment bytes")}})},
+	}}
+	out, err := runAITableCompositeCLI(t, caller, "+attachment-put",
+		"--base-id", "base", "--table-id", "table", "--record-id", "record", "--field-id", "field", "--file", filePath, "--mode", "replace", "--mime-type", "text/plain", "--yes")
+	if err != nil || !uploaded {
+		t.Fatalf("attachment put = output:%q err:%v uploaded:%v", out, err, uploaded)
+	}
+	for _, want := range []string{`"status": "recovered"`, `"fileToken": "ft1"`, `"attachmentCount": 1`, `"status": "verified"`} {
+		if !strings.Contains(out, want) {
+			t.Fatalf("attachment output missing %s: %s", want, out)
+		}
+	}
+	if len(caller.calls) != 4 || caller.calls[1].tool != "prepare_attachment_upload" || caller.calls[2].tool != "update_records" {
+		t.Fatalf("attachment calls = %#v", caller.calls)
+	}
+}
+
+func TestCrossPlatformCoverageAttachmentPutFailuresDoNotBecomeSuccessE2E(t *testing.T) {
+	filePath := writeAttachmentFixture(t, "bytes")
+	t.Run("append stops before upload when existing tokens unavailable", func(t *testing.T) {
+		caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+			{text: attachmentRecordJSON(t, "field", []any{map[string]any{"filename": "old.pdf", "url": "https://download"}})},
+		}}
+		out, err := runAITableCompositeCLI(t, caller, "+attachment-put",
+			"--base-id", "base", "--table-id", "table", "--record-id", "record", "--field-id", "field", "--file", filePath, "--yes")
+		if err == nil || out != "" || len(caller.calls) != 1 || !strings.Contains(err.Error(), "fileToken") {
+			t.Fatalf("unsafe append = output:%q err:%v calls:%#v", out, err, caller.calls)
+		}
+	})
+
+	t.Run("HTTP failure leaves no record write", func(t *testing.T) {
+		server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, _ *http.Request) { http.Error(w, "no", http.StatusBadGateway) }))
+		defer server.Close()
+		caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+			{text: attachmentRecordJSON(t, "field", []any{})},
+			{text: mustJSONText(t, map[string]any{"uploadUrl": server.URL, "fileToken": "ft1"})},
+		}}
+		out, err := runAITableCompositeCLI(t, caller, "+attachment-put",
+			"--base-id", "base", "--table-id", "table", "--record-id", "record", "--field-id", "field", "--file", filePath, "--mode", "replace", "--yes")
+		if err == nil || out != "" || len(caller.calls) != 2 || !strings.Contains(err.Error(), "partial_success") {
+			t.Fatalf("HTTP failure = output:%q err:%v calls:%#v", out, err, caller.calls)
+		}
+		var typed *apperrors.Error
+		if !errors.As(err, &typed) || typed.Retryable {
+			t.Fatalf("partially uploaded attachment must not be blindly retryable: %#v", err)
+		}
+	})
+
+	t.Run("prepare response must carry both values", func(t *testing.T) {
+		caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+			{text: attachmentRecordJSON(t, "field", []any{})}, {text: `{"uploadUrl":"https://example.invalid"}`},
+		}}
+		out, err := runAITableCompositeCLI(t, caller, "+attachment-put",
+			"--base-id", "base", "--table-id", "table", "--record-id", "record", "--field-id", "field", "--file", filePath, "--mode", "replace", "--yes")
+		if err == nil || out != "" || len(caller.calls) != 2 {
+			t.Fatalf("bad prepare = output:%q err:%v calls:%#v", out, err, caller.calls)
+		}
+	})
+}
+
+func TestCrossPlatformCoverageAttachmentPutRejectsHTTPSRedirectToPlaintextNonLoopbackE2E(t *testing.T) {
+	filePath := writeAttachmentFixture(t, "redirect-sensitive bytes")
+	sourceHits := 0
+	server := httptest.NewTLSServer(http.HandlerFunc(func(w http.ResponseWriter, _ *http.Request) {
+		sourceHits++
+		w.Header().Set("Location", "http://example.com/upload")
+		w.WriteHeader(http.StatusFound)
+	}))
+	defer server.Close()
+	client := server.Client()
+	client.CheckRedirect = rejectAttachmentUploadRedirect
+	testseam.Swap(t, &attachmentHTTPDo, client.Do)
+
+	caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+		{text: attachmentRecordJSON(t, "field", []any{})},
+		{text: mustJSONText(t, map[string]any{"uploadUrl": server.URL + "/put", "fileToken": "ft-redirect"})},
+	}}
+	out, err := runAITableCompositeCLI(t, caller, "+attachment-put",
+		"--base-id", "base", "--table-id", "table", "--record-id", "record", "--field-id", "field", "--file", filePath, "--mode", "replace", "--yes")
+	var typed *apperrors.Error
+	if err == nil || out != "" || !errors.As(err, &typed) || typed.Reason != "aitable_composite_partial_success" ||
+		typed.Cause == nil || !strings.Contains(typed.Cause.Error(), "status 302") {
+		t.Fatalf("redirect upload = output:%q err:%v", out, err)
+	}
+	if sourceHits != 1 || len(caller.calls) != 2 {
+		t.Fatalf("redirect upload reached an unexpected target: sourceHits=%d calls=%#v", sourceHits, caller.calls)
+	}
+}
+
+func TestCrossPlatformCoverageAttachmentRemoveClearAndSelectiveE2E(t *testing.T) {
+	t.Run("clear empty write response recovered by empty read-back", func(t *testing.T) {
+		caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+			{text: attachmentRecordJSON(t, "field", []any{map[string]any{"filename": "a.pdf", "url": "https://download"}})},
+			{text: ""},
+			{text: attachmentRecordJSON(t, "field", []any{})},
+		}}
+		out, err := runAITableCompositeCLI(t, caller, "+attachment-remove",
+			"--base-id", "base", "--table-id", "table", "--record-id", "record", "--field-id", "field", "--clear-all", "--yes")
+		if err != nil || !strings.Contains(out, `"status": "recovered"`) || !strings.Contains(out, `"removedCount": 1`) {
+			t.Fatalf("clear attachments = output:%q err:%v", out, err)
+		}
+	})
+
+	t.Run("selective removal preserves tokenized remainder", func(t *testing.T) {
+		caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+			{text: attachmentRecordJSON(t, "field", []any{
+				map[string]any{"fileToken": "remove-token", "filename": "remove.pdf"},
+				map[string]any{"fileToken": "keep-token", "filename": "keep.pdf"},
+			})},
+			{text: `{"updatedCount":1}`},
+			{text: attachmentRecordJSON(t, "field", []any{map[string]any{"filename": "keep.pdf", "size": 10}})},
+		}}
+		out, err := runAITableCompositeCLI(t, caller, "+attachment-remove",
+			"--base-id", "base", "--table-id", "table", "--record-id", "record", "--field-id", "field", "--remove-name", "remove.pdf", "--yes")
+		if err != nil || !strings.Contains(out, `"removedCount": 1`) {
+			t.Fatalf("selective remove = output:%q err:%v", out, err)
+		}
+		written := caller.calls[1].args["records"].([]any)[0].(map[string]any)["cells"].(map[string]any)["field"].([]any)
+		if len(written) != 1 || written[0].(map[string]any)["fileToken"] != "keep-token" {
+			t.Fatalf("selective desired attachments = %#v", written)
+		}
+	})
+}
+
+func TestCrossPlatformCoverageAttachmentRemoveExplainsTokenBoundaryE2E(t *testing.T) {
+	caller := &upsertByKeyCaller{steps: []upsertByKeyStep{{text: attachmentRecordJSON(t, "field", []any{
+		map[string]any{"filename": "remove.pdf", "url": "https://one"},
+		map[string]any{"filename": "keep.pdf", "url": "https://two"},
+	})}}}
+	out, err := runAITableCompositeCLI(t, caller, "+attachment-remove",
+		"--base-id", "base", "--table-id", "table", "--record-id", "record", "--field-id", "field", "--remove-name", "remove.pdf", "--yes")
+	if err == nil || out != "" || len(caller.calls) != 1 || !strings.Contains(err.Error(), "fileToken") || !strings.Contains(err.Error(), "增量删除") {
+		t.Fatalf("token boundary = output:%q err:%v calls:%#v", out, err, caller.calls)
+	}
+}
+
+func TestCrossPlatformCoverageAttachmentPutRemainingOutcomesE2E(t *testing.T) {
+	filePath := writeAttachmentFixture(t, "bytes")
+
+	t.Run("file open failure", func(t *testing.T) {
+		out, err := runAITableCompositeCLI(t, &upsertByKeyCaller{}, "+attachment-put",
+			"--base-id", "base", "--table-id", "table", "--record-id", "record", "--field-id", "field", "--file", filepath.Join(t.TempDir(), "missing"), "--yes")
+		if err == nil || out != "" {
+			t.Fatalf("file open = output:%q err:%v", out, err)
+		}
+	})
+
+	t.Run("initial read failure", func(t *testing.T) {
+		caller := &upsertByKeyCaller{steps: []upsertByKeyStep{{err: errors.New("read failed")}}}
+		out, err := runAITableCompositeCLI(t, caller, "+attachment-put",
+			"--base-id", "base", "--table-id", "table", "--record-id", "record", "--field-id", "field", "--file", filePath, "--yes")
+		if err == nil || out != "" || !strings.Contains(err.Error(), "read failed") {
+			t.Fatalf("initial read = output:%q err:%v", out, err)
+		}
+	})
+
+	t.Run("append dry run preserves token", func(t *testing.T) {
+		caller := &upsertByKeyCaller{dryRun: true, steps: []upsertByKeyStep{{text: attachmentRecordJSON(t, "field", []any{
+			map[string]any{"file_token": "old-token", "filename": "old.txt"},
+		})}}}
+		out, err := runAITableCompositeCLI(t, caller, "+attachment-put",
+			"--base-id", "base", "--table-id", "table", "--record-id", "record", "--field-id", "field", "--file", filePath, "--dry-run", "--yes")
+		if err != nil || !strings.Contains(out, `"status": "planned"`) || len(caller.calls) != 1 {
+			t.Fatalf("append dry run = output:%q err:%v calls:%#v", out, err, caller.calls)
+		}
+	})
+
+	t.Run("prepare call failure", func(t *testing.T) {
+		caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+			{text: attachmentRecordJSON(t, "field", []any{})},
+			{err: errors.New("prepare failed")},
+		}}
+		out, err := runAITableCompositeCLI(t, caller, "+attachment-put",
+			"--base-id", "base", "--table-id", "table", "--record-id", "record", "--field-id", "field", "--file", filePath, "--mode", "replace", "--yes")
+		if err == nil || out != "" {
+			t.Fatalf("prepare failure = output:%q err:%v", out, err)
+		}
+	})
+
+	t.Run("invalid upload URL", func(t *testing.T) {
+		caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+			{text: attachmentRecordJSON(t, "field", []any{})},
+			{text: `{"uploadUrl":"http://example.com/upload","fileToken":"ft1"}`},
+		}}
+		out, err := runAITableCompositeCLI(t, caller, "+attachment-put",
+			"--base-id", "base", "--table-id", "table", "--record-id", "record", "--field-id", "field", "--file", filePath, "--mode", "replace", "--yes")
+		if err == nil || out != "" {
+			t.Fatalf("invalid upload URL = output:%q err:%v", out, err)
+		}
+	})
+
+	t.Run("HTTP transport failure", func(t *testing.T) {
+		testseam.Swap(t, &attachmentHTTPDo, func(*http.Request) (*http.Response, error) {
+			return nil, errors.New("transport failed")
+		})
+		caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+			{text: attachmentRecordJSON(t, "field", []any{})},
+			{text: `{"uploadUrl":"https://uploads.example.test/put","fileToken":"ft1"}`},
+		}}
+		out, err := runAITableCompositeCLI(t, caller, "+attachment-put",
+			"--base-id", "base", "--table-id", "table", "--record-id", "record", "--field-id", "field", "--file", filePath, "--mode", "replace", "--yes")
+		if err == nil || out != "" {
+			t.Fatalf("transport failure = output:%q err:%v", out, err)
+		}
+	})
+
+	t.Run("write and read-back both fail", func(t *testing.T) {
+		server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, _ *http.Request) { w.WriteHeader(http.StatusNoContent) }))
+		defer server.Close()
+		caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+			{text: attachmentRecordJSON(t, "field", []any{})},
+			{text: mustJSONText(t, map[string]any{"uploadUrl": server.URL, "fileToken": "ft1"})},
+			{err: errors.New("write reply failed")},
+			{err: errors.New("verification read failed")},
+		}}
+		out, err := runAITableCompositeCLI(t, caller, "+attachment-put",
+			"--base-id", "base", "--table-id", "table", "--record-id", "record", "--field-id", "field", "--file", filePath, "--mode", "replace", "--yes")
+		if err == nil || out != "" {
+			t.Fatalf("verification failure = output:%q err:%v", out, err)
+		}
+	})
+}
+
+func TestCrossPlatformCoverageAttachmentRemoveRemainingOutcomesE2E(t *testing.T) {
+	run := func(t *testing.T, caller *upsertByKeyCaller, extra ...string) (string, error) {
+		args := []string{"--base-id", "base", "--table-id", "table", "--record-id", "record", "--field-id", "field"}
+		args = append(args, extra...)
+		args = append(args, "--yes")
+		return runAITableCompositeCLI(t, caller, "+attachment-remove", args...)
+	}
+
+	t.Run("initial read failure", func(t *testing.T) {
+		out, err := run(t, &upsertByKeyCaller{steps: []upsertByKeyStep{{err: errors.New("read failed")}}}, "--clear-all")
+		if err == nil || out != "" || !strings.Contains(err.Error(), "read failed") {
+			t.Fatalf("initial read = output:%q err:%v", out, err)
+		}
+	})
+
+	t.Run("unchanged selection", func(t *testing.T) {
+		caller := &upsertByKeyCaller{steps: []upsertByKeyStep{{text: attachmentRecordJSON(t, "field", []any{
+			map[string]any{"fileToken": "keep", "filename": "keep.pdf"},
+		})}}}
+		out, err := run(t, caller, "--remove-name", "missing.pdf")
+		if err != nil || !strings.Contains(out, `"status": "unchanged"`) || len(caller.calls) != 1 {
+			t.Fatalf("unchanged = output:%q err:%v", out, err)
+		}
+	})
+
+	t.Run("dry run", func(t *testing.T) {
+		caller := &upsertByKeyCaller{dryRun: true, steps: []upsertByKeyStep{{text: attachmentRecordJSON(t, "field", []any{
+			map[string]any{"fileToken": "remove", "filename": "remove.pdf"},
+		})}}}
+		out, err := run(t, caller, "--remove-name", "remove.pdf", "--dry-run")
+		if err != nil || !strings.Contains(out, `"status": "planned"`) || len(caller.calls) != 1 {
+			t.Fatalf("dry run = output:%q err:%v calls:%#v", out, err, caller.calls)
+		}
+	})
+
+	t.Run("read-back count mismatch", func(t *testing.T) {
+		caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+			{text: attachmentRecordJSON(t, "field", []any{map[string]any{"fileToken": "remove", "filename": "remove.pdf"}})},
+			{text: `{"updatedCount":1}`},
+			{text: attachmentRecordJSON(t, "field", []any{map[string]any{"filename": "other.pdf"}})},
+		}}
+		out, err := run(t, caller, "--clear-all")
+		if err == nil || out != "" {
+			t.Fatalf("count mismatch = output:%q err:%v", out, err)
+		}
+	})
+
+	t.Run("removed name remains", func(t *testing.T) {
+		caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+			{text: attachmentRecordJSON(t, "field", []any{
+				map[string]any{"fileToken": "remove", "filename": "remove.pdf"},
+				map[string]any{"fileToken": "keep", "filename": "keep.pdf"},
+			})},
+			{text: `{"updatedCount":1}`},
+			{text: attachmentRecordJSON(t, "field", []any{map[string]any{"filename": "remove.pdf"}})},
+		}}
+		out, err := run(t, caller, "--remove-name", "remove.pdf")
+		if err == nil || out != "" {
+			t.Fatalf("removed item remains = output:%q err:%v", out, err)
+		}
+	})
+
+	t.Run("write and verification fail", func(t *testing.T) {
+		caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+			{text: attachmentRecordJSON(t, "field", []any{map[string]any{"fileToken": "remove", "filename": "remove.pdf"}})},
+			{err: errors.New("write reply failed")},
+			{err: errors.New("verification failed")},
+		}}
+		out, err := run(t, caller, "--clear-all")
+		if err == nil || out != "" {
+			t.Fatalf("verification failure = output:%q err:%v", out, err)
+		}
+	})
+}
+
+func TestCrossPlatformCoverageAttachmentParsingAndValidation(t *testing.T) {
+	t.Run("file validation and MIME sniffing", func(t *testing.T) {
+		if file, _, _, err := openAttachmentFile(filepath.Join(t.TempDir(), "missing"), ""); err == nil || file != nil {
+			t.Fatalf("missing file must be rejected: file:%v err:%v", file, err)
+		}
+		if file, _, _, err := openAttachmentFile(t.TempDir(), ""); err == nil || file != nil {
+			t.Fatalf("directory must be rejected: file:%v err:%v", file, err)
+		}
+		empty := filepath.Join(t.TempDir(), "empty")
+		if err := os.WriteFile(empty, nil, 0o600); err != nil {
+			t.Fatal(err)
+		}
+		if file, _, _, err := openAttachmentFile(empty, ""); err == nil || file != nil {
+			t.Fatalf("empty file must be rejected: file:%v err:%v", file, err)
+		}
+		unknown := filepath.Join(t.TempDir(), "payload.unknown-extension")
+		if err := os.WriteFile(unknown, []byte("plain text"), 0o600); err != nil {
+			t.Fatal(err)
+		}
+		file, _, mimeType, err := openAttachmentFile(unknown, "")
+		if err != nil {
+			t.Fatal(err)
+		}
+		defer file.Close()
+		if mimeType == "" {
+			t.Fatal("sniffed MIME type is empty")
+		}
+	})
+
+	for name, raw := range map[string]string{
+		"missing host": "https:///upload",
+		"credentials":  "https://user:pass@example.com/upload",
+		"wrong scheme": "ftp://example.com/upload",
+	} {
+		t.Run(name, func(t *testing.T) {
+			if err := validateAttachmentUploadURL(raw); err == nil {
+				t.Fatalf("%q must be rejected", raw)
+			}
+		})
+	}
+	for _, raw := range []string{"http://localhost/upload", "http://127.0.0.1/upload", "https://example.com/upload"} {
+		if err := validateAttachmentUploadURL(raw); err != nil {
+			t.Fatalf("%q = %v", raw, err)
+		}
+	}
+
+	if err := verifyAttachmentPut(nil, 1, "token", "name", 1); err == nil {
+		t.Fatal("count mismatch must fail")
+	}
+	if err := verifyAttachmentPut([]map[string]any{{"fileToken": "token"}}, 1, "token", "name", 1); err != nil {
+		t.Fatal(err)
+	}
+	if err := verifyAttachmentPut([]map[string]any{{"filename": "other", "size": 1}}, 1, "token", "name", 1); err == nil {
+		t.Fatal("identity mismatch must fail")
+	}
+	for value, want := range map[any]bool{int(1): true, int64(1): true, "1": false} {
+		_, ok := numericInt64(value)
+		if ok != want {
+			t.Fatalf("numericInt64(%T) = %v, want %v", value, ok, want)
+		}
+	}
+}
+
+func TestCrossPlatformCoverageAttachmentRemoveSelectionValidationE2E(t *testing.T) {
+	for name, extra := range map[string][]string{
+		"neither": nil,
+		"both":    {"--remove-name", "a.pdf", "--clear-all"},
+	} {
+		t.Run(name, func(t *testing.T) {
+			args := []string{"--base-id", "base", "--table-id", "table", "--record-id", "record", "--field-id", "field"}
+			args = append(args, extra...)
+			args = append(args, "--yes")
+			out, err := runAITableCompositeCLI(t, &upsertByKeyCaller{}, "+attachment-remove", args...)
+			if err == nil || out != "" {
+				t.Fatalf("selection validation = output:%q err:%v", out, err)
+			}
+		})
+	}
+}
+
+func TestCrossPlatformCoverageAttachmentReadBackShapesE2E(t *testing.T) {
+	cases := []struct {
+		name string
+		step upsertByKeyStep
+		want string
+	}{
+		{name: "query error", step: upsertByKeyStep{err: errors.New("query failed")}, want: "query failed"},
+		{name: "wrong record", step: upsertByKeyStep{text: `{"records":[{"recordId":"other","cells":{}}]}`}, want: "exact records"},
+		{name: "missing cells", step: upsertByKeyStep{text: `{"records":[{"recordId":"record"}]}`}, want: "missing cells"},
+		{name: "non-array field", step: upsertByKeyStep{text: `{"records":[{"recordId":"record","cells":{"field":"bad"}}]}`}, want: "not an array"},
+		{name: "non-object item", step: upsertByKeyStep{text: `{"records":[{"recordId":"record","cells":{"field":["bad"]}}]}`}, want: "not an object"},
+	}
+	for _, tc := range cases {
+		t.Run(tc.name, func(t *testing.T) {
+			caller := &upsertByKeyCaller{steps: []upsertByKeyStep{tc.step}}
+			out, err := runAITableCompositeCLI(t, caller, "+attachment-remove",
+				"--base-id", "base", "--table-id", "table", "--record-id", "record", "--field-id", "field", "--clear-all", "--yes")
+			if err == nil || out != "" || !strings.Contains(err.Error(), tc.want) {
+				t.Fatalf("shape = output:%q err:%v", out, err)
+			}
+		})
+	}
+	caller := &upsertByKeyCaller{steps: []upsertByKeyStep{{text: `{"records":[{"recordId":"record","cells":{}}]}`}}}
+	out, err := runAITableCompositeCLI(t, caller, "+attachment-remove",
+		"--base-id", "base", "--table-id", "table", "--record-id", "record", "--field-id", "field", "--clear-all", "--yes")
+	if err != nil || !strings.Contains(out, `"status": "unchanged"`) {
+		t.Fatalf("missing field = output:%q err:%v", out, err)
+	}
+}

internal/shortcut/aitable/base_composite.go

@@ -0,0 +1,393 @@
+// Copyright 2026 Alibaba Group
+// SPDX-License-Identifier: Apache-2.0
+
+package aitable
+
+import (
+	"fmt"
+	"strings"
+
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/corecmd/contract"
+	apperrors "github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/errors"
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/shortcut"
+)
+
+var BaseSchemaSnapshot = shortcut.Shortcut{
+	Service:     "aitable",
+	Command:     "+base-schema-snapshot",
+	Product:     serverMain,
+	Description: "读取 Base、全部数据表、字段和视图的可复用结构快照，并严格校验每层响应",
+	Intent:      "当你要审计、迁移或复制一个 Base 的完整结构但不需要记录数据时使用；明确空 tables/fields/views 合法，缺失容器或缺少请求 ID 则失败。",
+	Risk:        shortcut.RiskRead,
+	Safety: contract.SafetySpec{
+		Effect: "read", Risk: "low", Confirmation: "not_required", Idempotency: "idempotent",
+	},
+	Contract: aitableCompositeContract(
+		"+base-schema-snapshot",
+		"读取 Base、全部数据表、字段和视图的可复用结构快照，并严格校验每层响应",
+		"当你要审计、迁移或复制一个 Base 的完整结构但不需要记录数据时使用；明确空 tables/fields/views 合法，缺失容器或缺少请求 ID 则失败。",
+		"需要记录数据用 record query；只看 Base 目录用 base get；快照不会创建或修改资源",
+		"dws aitable +base-schema-snapshot --base-id BASE_ID",
+	),
+	Flags: []shortcut.Flag{{Name: "base-id", Type: shortcut.FlagString, Desc: "Base ID", Required: true}},
+	Tips:  []string{`dws aitable +base-schema-snapshot --base-id BASE_ID`},
+	Execute: func(rt *shortcut.RuntimeContext) error {
+		return executeBaseSchemaSnapshot(rt)
+	},
+}
+
+var BaseBootstrap = shortcut.Shortcut{
+	Service:     "aitable",
+	Command:     "+base-bootstrap",
+	Product:     serverMain,
+	Description: "一次创建 Base、数据表和字段，逐层读回验证并在中断时报告已知副作用",
+	Intent:      "当你已有声明式 tables JSON、想一次搭好一套 AI 表格结构时使用；表内字段自动按 15 个拆批，每次创建都读回验证。",
+	Risk:        shortcut.RiskWrite,
+	Safety: contract.SafetySpec{
+		Effect: "write", Risk: "medium", Confirmation: "user_required", Idempotency: "non_idempotent",
+	},
+	Contract: aitableCompositeContract(
+		"+base-bootstrap",
+		"一次创建 Base、数据表和字段，逐层读回验证并在中断时报告已知副作用",
+		"当你已有声明式 tables JSON、想一次搭好一套 AI 表格结构时使用；表内字段自动按 15 个拆批，每次创建都读回验证。",
+		"已有 Base 只需新增一张表时用 table create；复制现有 Base 用 base copy；不要对失败请求盲目重试",
+		`dws aitable +base-bootstrap --name "项目管理" --tables '[{"name":"任务","fields":[]}]'`,
+	),
+	Flags: []shortcut.Flag{
+		{Name: "name", Type: shortcut.FlagString, Desc: "新 Base 名称", Required: true},
+		{Name: "folder-id", Type: shortcut.FlagString, Desc: "目标知识库文件夹 ID（可选）"},
+		{Name: "template-id", Type: shortcut.FlagString, Desc: "模板 ID（可选）"},
+		{Name: "tables", Type: shortcut.FlagString, Desc: `表结构 JSON 数组：[{'name':'任务','fields':[...]}]`, Required: true},
+	},
+	Tips: []string{`dws aitable +base-bootstrap --name "项目管理" --tables '[{"name":"任务","fields":[]}]'`},
+	Execute: func(rt *shortcut.RuntimeContext) error {
+		return executeBaseBootstrap(rt)
+	},
+}
+
+type bootstrapTable struct {
+	Name   string
+	Fields []any
+}
+
+func parseBootstrapTables(raw string) ([]bootstrapTable, error) {
+	value, err := parseJSONAny("tables", raw)
+	if err != nil {
+		return nil, err
+	}
+	items, ok := value.([]any)
+	if !ok || len(items) == 0 {
+		return nil, apperrors.NewValidation("--tables 必须是非空 JSON 数组")
+	}
+	if len(items) > 100 {
+		return nil, apperrors.NewValidation("--tables 最多接受 100 张表")
+	}
+	seen := map[string]bool{}
+	out := make([]bootstrapTable, 0, len(items))
+	for index, item := range items {
+		object, ok := item.(map[string]any)
+		if !ok {
+			return nil, apperrors.NewValidation(fmt.Sprintf("--tables[%d] 必须是 JSON 对象", index))
+		}
+		name := strings.TrimSpace(stringValue(object, "name"))
+		if name == "" || seen[name] {
+			return nil, apperrors.NewValidation(fmt.Sprintf("--tables[%d].name 必须非空且不能重复", index))
+		}
+		seen[name] = true
+		fields := []any{}
+		if rawFields, exists := object["fields"]; exists {
+			var fieldsOK bool
+			fields, fieldsOK = rawFields.([]any)
+			if !fieldsOK {
+				return nil, apperrors.NewValidation(fmt.Sprintf("--tables[%d].fields 必须是 JSON 数组", index))
+			}
+		}
+		for fieldIndex, field := range fields {
+			object, ok := field.(map[string]any)
+			if !ok || strings.TrimSpace(stringValue(object, "fieldName", "name")) == "" || strings.TrimSpace(stringValue(object, "type")) == "" {
+				return nil, apperrors.NewValidation(fmt.Sprintf("--tables[%d].fields[%d] 必须包含 fieldName 和 type", index, fieldIndex))
+			}
+		}
+		out = append(out, bootstrapTable{Name: name, Fields: fields})
+	}
+	return out, nil
+}
+
+func executeBaseSchemaSnapshot(rt *shortcut.RuntimeContext) error {
+	baseID := rt.Str("base-id")
+	base, err := rt.CallMCPData(serverMain, "get_base", map[string]any{"baseId": baseID})
+	if err != nil {
+		return err
+	}
+	if !deepContainsString(base, baseID) {
+		return fmt.Errorf("get_base response does not identify requested baseId %s", baseID)
+	}
+	tables, found := findNamedObjectList(base, "tables", "tableList", "sheets")
+	if !found {
+		return fmt.Errorf("get_base response is missing the tables collection")
+	}
+	result := newCompositeResult("base_schema_snapshot")
+	result.RequestedCount = len(tables)
+	result.Resolved = map[string]any{"baseId": baseID}
+	tableSnapshots := make([]any, 0, len(tables))
+	for index, table := range tables {
+		tableID := stringValue(table, "tableId", "sheetId", "id")
+		if tableID == "" {
+			return fmt.Errorf("get_base tables[%d] is missing tableId", index)
+		}
+		detail, err := rt.CallMCPData(serverMain, "get_tables", map[string]any{"baseId": baseID, "tableIds": []string{tableID}})
+		if err != nil || !deepContainsString(detail, tableID) {
+			if err == nil {
+				err = fmt.Errorf("get_tables response does not identify tableId %s", tableID)
+			}
+			return err
+		}
+		fieldsData, err := rt.CallMCPData(serverMain, "get_fields", map[string]any{"baseId": baseID, "tableId": tableID})
+		if err != nil {
+			return err
+		}
+		fields, found := findNamedObjectList(fieldsData, "fields", "fieldList")
+		if !found {
+			return fmt.Errorf("get_fields response for %s is missing the fields collection", tableID)
+		}
+		viewsData, err := rt.CallMCPData(serverMain, "get_views", map[string]any{"baseId": baseID, "tableId": tableID})
+		if err != nil {
+			return err
+		}
+		views, found := findNamedObjectList(viewsData, "views", "viewList")
+		if !found {
+			return fmt.Errorf("get_views response for %s is missing the views collection", tableID)
+		}
+		tableSnapshots = append(tableSnapshots, map[string]any{
+			"tableId": tableID, "summary": table, "detail": detail, "fields": fields, "views": views,
+		})
+		result.CompletedCount = index + 1
+	}
+	result.Result = map[string]any{"base": base, "tables": tableSnapshots}
+	result.Verification = map[string]any{"status": "verified", "tableCount": len(tables)}
+	return rt.Output(result)
+}
+
+func executeBaseBootstrap(rt *shortcut.RuntimeContext) error {
+	tables, err := parseBootstrapTables(rt.Str("tables"))
+	if err != nil {
+		return err
+	}
+	result := newCompositeResult("base_bootstrap")
+	result.RequestedCount = len(tables)
+	result.Plan = append(result.Plan, compositeStep{Index: 1, Name: "create base", Tool: "create_base", Status: "planned"})
+	for index, table := range tables {
+		result.Plan = append(result.Plan, compositeStep{Index: index + 2, Name: "create and verify table", Tool: "create_table", Status: "planned", Count: len(table.Fields), Arguments: map[string]any{"name": table.Name}})
+	}
+	if rt.DryRun() {
+		result.Status = "planned"
+		result.Executed = false
+		return rt.Output(result)
+	}
+	baseArgs := map[string]any{"baseName": rt.Str("name")}
+	if rt.Changed("folder-id") {
+		baseArgs["folderId"] = rt.Str("folder-id")
+	}
+	if rt.Changed("template-id") {
+		baseArgs["templateId"] = rt.Str("template-id")
+	}
+	baseData, err := rt.CallMCPWriteDataStrict(serverMain, "create_base", baseArgs)
+	if err != nil {
+		result.Status = "unknown"
+		result.FailedCount = len(tables)
+		result.Checkpoint = map[string]any{"nextStep": "resolve whether base was created before retrying"}
+		return compositeError(result, err, false)
+	}
+	baseID := findStringByKeys(baseData, "baseId")
+	if baseID == "" {
+		result.Status = "unknown"
+		result.Result = baseData
+		result.Checkpoint = map[string]any{"nextStep": "locate the created Base by exact name before retrying"}
+		return compositeError(result, fmt.Errorf("create_base response is missing baseId"), false)
+	}
+	result.Resolved = map[string]any{"baseId": baseID}
+	result.KnownEffects = append(result.KnownEffects, map[string]any{"tool": "create_base", "baseId": baseID})
+	result.CompletedSteps = append(result.CompletedSteps, compositeStep{Index: 1, Name: "create base", Tool: "create_base", Status: "completed", Result: baseData})
+	baseRead, err := rt.CallMCPData(serverMain, "get_base", map[string]any{"baseId": baseID})
+	if err != nil || !deepContainsString(baseRead, baseID) {
+		if err == nil {
+			err = fmt.Errorf("get_base does not identify created baseId %s", baseID)
+		}
+		result.Status = "partial_success"
+		result.Checkpoint = map[string]any{"nextTableIndex": 0, "baseId": baseID}
+		return compositeError(result, err, false)
+	}
+
+	createdTables := make([]any, 0, len(tables))
+	for index, spec := range tables {
+		initialEnd := minInt(15, len(spec.Fields))
+		initialFields := spec.Fields[:initialEnd]
+		createData, createErr := rt.CallMCPWriteDataStrict(serverMain, "create_table", map[string]any{
+			"baseId": baseID, "tableName": spec.Name, "fields": initialFields,
+		})
+		tableID := findStringByKeys(createData, "tableId", "sheetId")
+		if createErr != nil || tableID == "" {
+			if createErr == nil {
+				createErr = fmt.Errorf("create_table response for %q is missing tableId", spec.Name)
+			}
+			result.Status = "partial_success"
+			result.CompletedCount = index
+			result.FailedCount = len(tables) - index
+			result.Checkpoint = map[string]any{"nextTableIndex": index, "baseId": baseID}
+			return compositeError(result, createErr, false)
+		}
+		result.KnownEffects = append(result.KnownEffects, map[string]any{"tool": "create_table", "baseId": baseID, "tableId": tableID, "name": spec.Name})
+		for offset := initialEnd; offset < len(spec.Fields); offset += 15 {
+			end := minInt(offset+15, len(spec.Fields))
+			_, fieldErr := rt.CallMCPWriteDataStrict(serverMain, "create_fields", map[string]any{
+				"baseId": baseID, "tableId": tableID, "fields": spec.Fields[offset:end],
+			})
+			if fieldErr != nil {
+				result.Warnings = append(result.Warnings, fmt.Sprintf("create_fields offset %d returned an error; checking final field state: %v", offset, fieldErr))
+			}
+		}
+		detail, verifyErr := rt.CallMCPData(serverMain, "get_tables", map[string]any{"baseId": baseID, "tableIds": []string{tableID}})
+		if verifyErr != nil || !deepContainsString(detail, tableID) {
+			if verifyErr == nil {
+				verifyErr = fmt.Errorf("get_tables does not identify created tableId %s", tableID)
+			}
+			result.Status = "partial_success"
+			result.CompletedCount = index
+			result.FailedCount = len(tables) - index
+			result.Checkpoint = map[string]any{"nextTableIndex": index, "baseId": baseID, "tableId": tableID, "step": "verify table and fields"}
+			return compositeError(result, verifyErr, false)
+		}
+		fieldsData, verifyErr := rt.CallMCPData(serverMain, "get_fields", map[string]any{"baseId": baseID, "tableId": tableID})
+		fields, found := findNamedObjectList(fieldsData, "fields", "fieldList")
+		if verifyErr != nil || !found || !containsAllFieldNames(fields, spec.Fields) {
+			if verifyErr == nil {
+				verifyErr = fmt.Errorf("field read-back for table %s does not contain the declared field set", tableID)
+			}
+			result.Status = "partial_success"
+			result.CompletedCount = index
+			result.FailedCount = len(tables) - index
+			result.Checkpoint = map[string]any{"nextTableIndex": index, "baseId": baseID, "tableId": tableID, "step": "repair fields"}
+			return compositeError(result, verifyErr, false)
+		}
+		result.CompletedCount = index + 1
+		result.CompletedSteps = append(result.CompletedSteps, compositeStep{Index: index + 2, Name: "create and verify table", Tool: "create_table", Status: "completed", Count: len(spec.Fields), Result: map[string]any{"tableId": tableID, "fieldCount": len(fields)}})
+		createdTables = append(createdTables, map[string]any{"tableId": tableID, "name": spec.Name, "fieldCount": len(fields)})
+	}
+	result.Verification = map[string]any{"status": "verified", "baseId": baseID, "tableCount": len(createdTables)}
+	result.Result = map[string]any{"baseId": baseID, "tables": createdTables}
+	return rt.Output(result)
+}
+
+func findNamedObjectList(value any, names ...string) ([]map[string]any, bool) {
+	nameSet := map[string]bool{}
+	for _, name := range names {
+		nameSet[strings.ToLower(name)] = true
+	}
+	var walk func(any) ([]map[string]any, bool)
+	walk = func(current any) ([]map[string]any, bool) {
+		object, ok := current.(map[string]any)
+		if !ok {
+			return nil, false
+		}
+		for key, child := range object {
+			if nameSet[strings.ToLower(key)] {
+				items, ok := child.([]any)
+				if !ok {
+					return nil, false
+				}
+				out := make([]map[string]any, 0, len(items))
+				for _, item := range items {
+					entry, ok := item.(map[string]any)
+					if !ok {
+						return nil, false
+					}
+					out = append(out, entry)
+				}
+				return out, true
+			}
+		}
+		for _, child := range object {
+			if found, ok := walk(child); ok {
+				return found, true
+			}
+		}
+		return nil, false
+	}
+	return walk(value)
+}
+
+func stringValue(object map[string]any, keys ...string) string {
+	for _, key := range keys {
+		if value, ok := object[key].(string); ok {
+			return value
+		}
+	}
+	return ""
+}
+
+func findStringByKeys(value any, keys ...string) string {
+	keySet := map[string]bool{}
+	for _, key := range keys {
+		keySet[strings.ToLower(key)] = true
+	}
+	var walk func(any) string
+	walk = func(current any) string {
+		switch typed := current.(type) {
+		case map[string]any:
+			for key, child := range typed {
+				if keySet[strings.ToLower(key)] {
+					if text, ok := child.(string); ok && strings.TrimSpace(text) != "" {
+						return strings.TrimSpace(text)
+					}
+				}
+			}
+			for _, child := range typed {
+				if found := walk(child); found != "" {
+					return found
+				}
+			}
+		case []any:
+			for _, child := range typed {
+				if found := walk(child); found != "" {
+					return found
+				}
+			}
+		}
+		return ""
+	}
+	return walk(value)
+}
+
+func deepContainsString(value any, expected string) bool {
+	switch typed := value.(type) {
+	case map[string]any:
+		for _, child := range typed {
+			if deepContainsString(child, expected) {
+				return true
+			}
+		}
+	case []any:
+		for _, child := range typed {
+			if deepContainsString(child, expected) {
+				return true
+			}
+		}
+	case string:
+		return typed == expected
+	}
+	return false
+}
+
+func containsAllFieldNames(actual []map[string]any, expected []any) bool {
+	names := map[string]bool{}
+	for _, field := range actual {
+		names[stringValue(field, "fieldName", "name")] = true
+	}
+	for _, raw := range expected {
+		field, _ := raw.(map[string]any)
+		if !names[stringValue(field, "fieldName", "name")] {
+			return false
+		}
+	}
+	return true
+}

internal/shortcut/aitable/base_composite_e2e_test.go

@@ -0,0 +1,311 @@
+// Copyright 2026 Alibaba Group
+// SPDX-License-Identifier: Apache-2.0
+
+package aitable
+
+import (
+	"bytes"
+	"encoding/json"
+	"errors"
+	"fmt"
+	"strings"
+	"testing"
+
+	apperrors "github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/errors"
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/helpers"
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/shortcut"
+	"github.com/spf13/cobra"
+)
+
+func runAITableCompositeCLI(t *testing.T, caller *upsertByKeyCaller, command string, args ...string) (string, error) {
+	t.Helper()
+	helpers.InitDepsForTest(t, caller)
+	root := &cobra.Command{Use: "dws", SilenceErrors: true, SilenceUsage: true}
+	root.PersistentFlags().Bool("yes", false, "")
+	root.PersistentFlags().Bool("dry-run", false, "")
+	root.PersistentFlags().String("format", "json", "")
+	root.AddCommand(shortcut.Commands()...)
+	stdout := &bytes.Buffer{}
+	root.SetOut(stdout)
+	root.SetErr(&bytes.Buffer{})
+	root.SetArgs(append([]string{"aitable", command}, args...))
+	err := root.Execute()
+	return stdout.String(), err
+}
+
+func TestCrossPlatformCoverageBaseSchemaSnapshotE2E(t *testing.T) {
+	t.Run("one table with legal empty fields and views", func(t *testing.T) {
+		caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+			{text: `{"data":{"baseId":"base","tables":[{"tableId":"t1","name":"任务"}]}}`},
+			{text: `{"data":{"tables":[{"tableId":"t1","name":"任务"}]}}`},
+			{text: `{"data":{"fields":[]}}`},
+			{text: `{"data":{"views":[]}}`},
+		}}
+		out, err := runAITableCompositeCLI(t, caller, "+base-schema-snapshot", "--base-id", "base")
+		if err != nil {
+			t.Fatalf("snapshot error = %v", err)
+		}
+		for _, want := range []string{`"status": "success"`, `"tableCount": 1`, `"fields": []`, `"views": []`} {
+			if !strings.Contains(out, want) {
+				t.Fatalf("snapshot output missing %s: %s", want, out)
+			}
+		}
+	})
+
+	t.Run("empty tables is explicit success", func(t *testing.T) {
+		caller := &upsertByKeyCaller{steps: []upsertByKeyStep{{text: `{"baseId":"base","tables":[]}`}}}
+		out, err := runAITableCompositeCLI(t, caller, "+base-schema-snapshot", "--base-id", "base")
+		if err != nil || !strings.Contains(out, `"tableCount": 0`) || len(caller.calls) != 1 {
+			t.Fatalf("empty snapshot = output:%q err:%v calls:%#v", out, err, caller.calls)
+		}
+	})
+
+	t.Run("missing tables contract fails", func(t *testing.T) {
+		caller := &upsertByKeyCaller{steps: []upsertByKeyStep{{text: `{"baseId":"base","data":{}}`}}}
+		out, err := runAITableCompositeCLI(t, caller, "+base-schema-snapshot", "--base-id", "base")
+		if err == nil || out != "" {
+			t.Fatalf("missing tables = output:%q err:%v", out, err)
+		}
+	})
+}
+
+func bootstrapFields(count int) []any {
+	fields := make([]any, 0, count)
+	for index := 0; index < count; index++ {
+		fields = append(fields, map[string]any{"fieldName": fmt.Sprintf("F%02d", index), "type": "text"})
+	}
+	return fields
+}
+
+func marshalBootstrapTables(t *testing.T, fields []any) string {
+	t.Helper()
+	if fields == nil {
+		fields = []any{}
+	}
+	raw, err := json.Marshal([]any{map[string]any{"name": "任务", "fields": fields}})
+	if err != nil {
+		t.Fatal(err)
+	}
+	return string(raw)
+}
+
+func fieldReadBackJSON(t *testing.T, fields []any) string {
+	t.Helper()
+	raw, err := json.Marshal(map[string]any{"fields": fields})
+	if err != nil {
+		t.Fatal(err)
+	}
+	return string(raw)
+}
+
+func TestCrossPlatformCoverageBaseBootstrapCreatesChunksAndVerifiesE2E(t *testing.T) {
+	fields := bootstrapFields(16)
+	caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+		{text: `{"data":{"baseId":"b-new"}}`},
+		{text: `{"baseId":"b-new","tables":[]}`},
+		{text: `{"data":{"tableId":"t-new"}}`},
+		{text: `{"createdFields":[{"fieldId":"f16"}]}`},
+		{text: `{"tables":[{"tableId":"t-new","name":"任务"}]}`},
+		{text: fieldReadBackJSON(t, fields)},
+	}}
+	out, err := runAITableCompositeCLI(t, caller, "+base-bootstrap", "--name", "项目", "--tables", marshalBootstrapTables(t, fields), "--yes")
+	if err != nil {
+		t.Fatalf("bootstrap error = %v", err)
+	}
+	for _, want := range []string{`"baseId": "b-new"`, `"tableId": "t-new"`, `"status": "verified"`} {
+		if !strings.Contains(out, want) {
+			t.Fatalf("bootstrap output missing %s: %s", want, out)
+		}
+	}
+	if len(caller.calls) != 6 || caller.calls[0].tool != "create_base" || caller.calls[2].tool != "create_table" || caller.calls[3].tool != "create_fields" {
+		t.Fatalf("bootstrap calls = %#v", caller.calls)
+	}
+	if caller.calls[0].args["baseName"] != "项目" {
+		t.Fatalf("create_base args = %#v", caller.calls[0].args)
+	}
+	createBaseCalls := 0
+	for _, call := range caller.calls {
+		if call.tool == "create_base" {
+			createBaseCalls++
+		}
+	}
+	if createBaseCalls != 1 {
+		t.Fatalf("create_base calls = %d, want exactly 1; calls = %#v", createBaseCalls, caller.calls)
+	}
+	if got := len(caller.calls[2].args["fields"].([]any)); got != 15 {
+		t.Fatalf("create_table fields = %d, want 15", got)
+	}
+	if got := len(caller.calls[3].args["fields"].([]any)); got != 1 {
+		t.Fatalf("create_fields fields = %d, want 1", got)
+	}
+}
+
+func TestCrossPlatformCoverageBaseBootstrapRequiresConfirmationBeforeMCP(t *testing.T) {
+	caller := &upsertByKeyCaller{}
+	out, err := runAITableCompositeCLI(t, caller, "+base-bootstrap", "--name", "项目", "--tables", marshalBootstrapTables(t, nil))
+	if out != "" {
+		t.Fatalf("unconfirmed bootstrap output = %q, want empty", out)
+	}
+	var typed *apperrors.Error
+	if !errors.As(err, &typed) || typed.Reason != "confirmation_required" {
+		t.Fatalf("unconfirmed bootstrap error = %#v, want confirmation_required", err)
+	}
+	if len(caller.calls) != 0 {
+		t.Fatalf("unconfirmed bootstrap reached MCP: %#v", caller.calls)
+	}
+}
+
+func TestCrossPlatformCoverageBaseBootstrapUnknownAndDryRunE2E(t *testing.T) {
+	t.Run("empty create response is unknown and not retry-safe", func(t *testing.T) {
+		caller := &upsertByKeyCaller{steps: []upsertByKeyStep{{text: ""}}}
+		out, err := runAITableCompositeCLI(t, caller, "+base-bootstrap", "--name", "项目", "--tables", marshalBootstrapTables(t, nil), "--yes")
+		if err == nil || out != "" {
+			t.Fatalf("unknown base create = output:%q err:%v", out, err)
+		}
+		var typed *apperrors.Error
+		if !errors.As(err, &typed) || typed.Reason != "aitable_composite_unknown" || typed.Retryable {
+			t.Fatalf("unknown base error = %#v", err)
+		}
+	})
+
+	t.Run("dry run validates and makes no calls", func(t *testing.T) {
+		caller := &upsertByKeyCaller{dryRun: true}
+		out, err := runAITableCompositeCLI(t, caller, "+base-bootstrap", "--name", "项目", "--tables", marshalBootstrapTables(t, nil), "--dry-run")
+		if err != nil || len(caller.calls) != 0 || !strings.Contains(out, `"status": "planned"`) {
+			t.Fatalf("bootstrap dry run = output:%q err:%v calls:%#v", out, err, caller.calls)
+		}
+	})
+}
+
+func TestCrossPlatformCoverageBaseBootstrapInputValidation(t *testing.T) {
+	tooMany := make([]any, 101)
+	for index := range tooMany {
+		tooMany[index] = map[string]any{"name": fmt.Sprintf("T%d", index)}
+	}
+	tooManyJSON, err := json.Marshal(tooMany)
+	if err != nil {
+		t.Fatal(err)
+	}
+	cases := map[string]string{
+		"invalid JSON":       `{`,
+		"not an array":       `{}`,
+		"empty array":        `[]`,
+		"too many tables":    string(tooManyJSON),
+		"non-object table":   `[1]`,
+		"empty name":         `[{"name":" "}]`,
+		"duplicate name":     `[{"name":"T"},{"name":"T"}]`,
+		"fields not array":   `[{"name":"T","fields":{}}]`,
+		"field not object":   `[{"name":"T","fields":[1]}]`,
+		"field missing type": `[{"name":"T","fields":[{"fieldName":"F"}]}]`,
+	}
+	for name, raw := range cases {
+		t.Run(name, func(t *testing.T) {
+			if tables, err := parseBootstrapTables(raw); err == nil || tables != nil {
+				t.Fatalf("parseBootstrapTables(%s) = %#v, %v", raw, tables, err)
+			}
+		})
+	}
+}
+
+func TestCrossPlatformCoverageBaseSnapshotFailureStagesE2E(t *testing.T) {
+	base := `{"baseId":"base","tables":[{"tableId":"t1","name":"T"}]}`
+	cases := []struct {
+		name  string
+		steps []upsertByKeyStep
+	}{
+		{name: "get base error", steps: []upsertByKeyStep{{err: errors.New("get base failed")}}},
+		{name: "get base wrong id", steps: []upsertByKeyStep{{text: `{"baseId":"other","tables":[]}`}}},
+		{name: "table missing id", steps: []upsertByKeyStep{{text: `{"baseId":"base","tables":[{"name":"T"}]}`}}},
+		{name: "get tables error", steps: []upsertByKeyStep{{text: base}, {err: errors.New("get tables failed")}}},
+		{name: "get tables wrong id", steps: []upsertByKeyStep{{text: base}, {text: `{"tables":[{"tableId":"other"}]}`}}},
+		{name: "get fields error", steps: []upsertByKeyStep{{text: base}, {text: `{"tables":[{"tableId":"t1"}]}`}, {err: errors.New("get fields failed")}}},
+		{name: "get fields missing collection", steps: []upsertByKeyStep{{text: base}, {text: `{"tables":[{"tableId":"t1"}]}`}, {text: `{}`}}},
+		{name: "get views error", steps: []upsertByKeyStep{{text: base}, {text: `{"tables":[{"tableId":"t1"}]}`}, {text: `{"fields":[]}`}, {err: errors.New("get views failed")}}},
+		{name: "get views missing collection", steps: []upsertByKeyStep{{text: base}, {text: `{"tables":[{"tableId":"t1"}]}`}, {text: `{"fields":[]}`}, {text: `{}`}}},
+	}
+	for _, tc := range cases {
+		t.Run(tc.name, func(t *testing.T) {
+			out, err := runAITableCompositeCLI(t, &upsertByKeyCaller{steps: tc.steps}, "+base-schema-snapshot", "--base-id", "base")
+			if err == nil || out != "" {
+				t.Fatalf("snapshot failure = output:%q err:%v", out, err)
+			}
+		})
+	}
+}
+
+func TestCrossPlatformCoverageBaseBootstrapExecuteRejectsInvalidTablesE2E(t *testing.T) {
+	out, err := runAITableCompositeCLI(t, &upsertByKeyCaller{}, "+base-bootstrap", "--name", "Project", "--tables", `{`, "--yes")
+	if err == nil || out != "" {
+		t.Fatalf("invalid bootstrap tables = output:%q err:%v", out, err)
+	}
+}
+
+func TestCrossPlatformCoverageBaseBootstrapFailureStagesE2E(t *testing.T) {
+	tables := marshalBootstrapTables(t, nil)
+	cases := []struct {
+		name      string
+		steps     []upsertByKeyStep
+		extra     []string
+		withField bool
+	}{
+		{name: "missing base id", steps: []upsertByKeyStep{{text: `{}`}}},
+		{name: "base verification error", steps: []upsertByKeyStep{{text: `{"baseId":"b"}`}, {err: errors.New("verify base failed")}}},
+		{name: "base verification wrong id", steps: []upsertByKeyStep{{text: `{"baseId":"b"}`}, {text: `{"baseId":"other"}`}}},
+		{name: "create table error", steps: []upsertByKeyStep{{text: `{"baseId":"b"}`}, {text: `{"baseId":"b"}`}, {err: errors.New("create table failed")}}},
+		{name: "create table missing id", steps: []upsertByKeyStep{{text: `{"baseId":"b"}`}, {text: `{"baseId":"b"}`}, {text: `{}`}}},
+		{name: "verify table error", steps: []upsertByKeyStep{{text: `{"baseId":"b"}`}, {text: `{"baseId":"b"}`}, {text: `{"tableId":"t"}`}, {err: errors.New("verify table failed")}}},
+		{name: "verify table wrong id", steps: []upsertByKeyStep{{text: `{"baseId":"b"}`}, {text: `{"baseId":"b"}`}, {text: `{"tableId":"t"}`}, {text: `{"tables":[]}`}}},
+		{name: "verify fields error", steps: []upsertByKeyStep{{text: `{"baseId":"b"}`}, {text: `{"baseId":"b"}`}, {text: `{"tableId":"t"}`}, {text: `{"tables":[{"tableId":"t"}]}`}, {err: errors.New("verify fields failed")}}},
+		{name: "verify fields mismatch", steps: []upsertByKeyStep{{text: `{"baseId":"b"}`}, {text: `{"baseId":"b"}`}, {text: `{"tableId":"t"}`}, {text: `{"tables":[{"tableId":"t"}]}`}, {text: `{"fields":[]}`}}, extra: []string{"--folder-id", "folder", "--template-id", "template"}, withField: true},
+	}
+	for _, tc := range cases {
+		t.Run(tc.name, func(t *testing.T) {
+			inputTables := tables
+			if tc.withField {
+				inputTables = marshalBootstrapTables(t, bootstrapFields(1))
+			}
+			args := []string{"--name", "Project", "--tables", inputTables}
+			args = append(args, tc.extra...)
+			args = append(args, "--yes")
+			out, err := runAITableCompositeCLI(t, &upsertByKeyCaller{steps: tc.steps}, "+base-bootstrap", args...)
+			if err == nil || out != "" {
+				t.Fatalf("bootstrap failure = output:%q err:%v", out, err)
+			}
+		})
+	}
+}
+
+func TestCrossPlatformCoverageBaseBootstrapRecoversFieldCallErrorE2E(t *testing.T) {
+	fields := bootstrapFields(16)
+	caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+		{text: `{"baseId":"b"}`},
+		{text: `{"baseId":"b"}`},
+		{text: `{"tableId":"t"}`},
+		{err: errors.New("create fields reply failed")},
+		{text: `{"tables":[{"tableId":"t"}]}`},
+		{text: fieldReadBackJSON(t, fields)},
+	}}
+	out, err := runAITableCompositeCLI(t, caller, "+base-bootstrap",
+		"--name", "Project", "--tables", marshalBootstrapTables(t, fields), "--folder-id", "folder", "--template-id", "template", "--yes")
+	if err != nil || !strings.Contains(out, `"status": "success"`) || !strings.Contains(out, "create_fields offset") {
+		t.Fatalf("field recovery = output:%q err:%v", out, err)
+	}
+	if caller.calls[0].args["folderId"] != "folder" || caller.calls[0].args["templateId"] != "template" {
+		t.Fatalf("create base args = %#v", caller.calls[0].args)
+	}
+}
+
+func TestCrossPlatformCoverageBaseCompositeShapeHelpers(t *testing.T) {
+	if _, ok := findNamedObjectList(map[string]any{"tables": "bad"}, "tables"); ok {
+		t.Fatal("non-array named child must fail")
+	}
+	if _, ok := findNamedObjectList(map[string]any{"tables": []any{"bad"}}, "tables"); ok {
+		t.Fatal("non-object list item must fail")
+	}
+	if containsAllFieldNames([]map[string]any{{"fieldName": "A"}}, []any{map[string]any{"fieldName": "B"}}) {
+		t.Fatal("missing field name must fail")
+	}
+	if got := findStringByKeys(map[string]any{"items": []any{map[string]any{"nested": " value "}}}, "nested"); got != "value" {
+		t.Fatalf("nested array string = %q", got)
+	}
+}

internal/shortcut/aitable/composite_contract.go

@@ -0,0 +1,38 @@
+// Copyright 2026 Alibaba Group
+// SPDX-License-Identifier: Apache-2.0
+
+package aitable
+
+import (
+	"strings"
+
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/corecmd"
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/corecmd/contract"
+)
+
+func aitableCompositeContract(command, description, useWhen, avoidWhen, example string) corecmd.ContractDecl {
+	name := strings.TrimPrefix(command, "+")
+	name = strings.ReplaceAll(name, "-", "_")
+	cliPath := "aitable " + command
+	return corecmd.ContractDecl{
+		Identity: contract.ToolIdentitySpec{
+			ProductID:      "aitable",
+			Name:           "shortcut_" + name,
+			CanonicalPath:  "aitable.shortcut_" + name,
+			CLIPath:        cliPath,
+			PrimaryCLIPath: cliPath,
+		},
+		Description: description,
+		Interface: &contract.InterfaceSpec{
+			Mode:         "composite",
+			Availability: "available",
+			Reason:       "DWS reviewed composite orchestration with explicit read-back success predicates, partial-effect reporting, and resumable checkpoints; no single MCP operation represents the full command contract.",
+		},
+		Selection: contract.SelectionSpec{
+			AgentSummary: description,
+			UseWhen:      []string{useWhen},
+			AvoidWhen:    []string{avoidWhen},
+			Examples:     []string{example},
+		},
+	}
+}

internal/shortcut/aitable/composite_output.go

@@ -0,0 +1,71 @@
+// Copyright 2026 Alibaba Group
+// SPDX-License-Identifier: Apache-2.0
+
+package aitable
+
+import (
+	"fmt"
+
+	apperrors "github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/errors"
+)
+
+const aitableCompositeContractVersion = "aitable.composite.v1"
+
+type compositeResult struct {
+	ContractVersion string           `json:"contractVersion"`
+	Operation       string           `json:"operation"`
+	Status          string           `json:"status"`
+	Executed        bool             `json:"executed"`
+	Retryable       bool             `json:"retryable"`
+	RequestedCount  int              `json:"requestedCount,omitempty"`
+	CompletedCount  int              `json:"completedCount,omitempty"`
+	FailedCount     int              `json:"failedCount,omitempty"`
+	Resolved        map[string]any   `json:"resolved,omitempty"`
+	Plan            []compositeStep  `json:"plan,omitempty"`
+	CompletedSteps  []compositeStep  `json:"completedSteps,omitempty"`
+	Verification    map[string]any   `json:"verification,omitempty"`
+	Checkpoint      map[string]any   `json:"checkpoint,omitempty"`
+	NextCommand     string           `json:"nextCommand,omitempty"`
+	KnownEffects    []map[string]any `json:"knownSideEffects,omitempty"`
+	Warnings        []string         `json:"warnings,omitempty"`
+	Result          map[string]any   `json:"result,omitempty"`
+}
+
+type compositeStep struct {
+	Index     int            `json:"index"`
+	Name      string         `json:"name"`
+	Tool      string         `json:"tool,omitempty"`
+	Status    string         `json:"status"`
+	Offset    int            `json:"offset,omitempty"`
+	Count     int            `json:"count,omitempty"`
+	Arguments map[string]any `json:"arguments,omitempty"`
+	Result    map[string]any `json:"result,omitempty"`
+	Error     string         `json:"error,omitempty"`
+}
+
+func newCompositeResult(operation string) compositeResult {
+	return compositeResult{
+		ContractVersion: aitableCompositeContractVersion,
+		Operation:       operation,
+		Status:          "success",
+		Executed:        true,
+	}
+}
+
+func compositeError(result compositeResult, cause error, retryable bool) error {
+	result.Retryable = retryable
+	if result.Status == "success" {
+		result.Status = "unknown"
+	}
+	return apperrors.NewAPI(fmt.Sprintf("AI Table composite %s ended with status %s", result.Operation, result.Status),
+		apperrors.WithOperation("aitable."+result.Operation),
+		apperrors.WithOrigin("mcp"),
+		apperrors.WithFailureStage("composite_execution"),
+		apperrors.WithExecutionStarted(result.Executed),
+		apperrors.WithRetryable(retryable),
+		apperrors.WithReason("aitable_composite_"+result.Status),
+		apperrors.WithHint("inspect error.details.result before retrying; unknown means the remote effect could not be proven"),
+		apperrors.WithDetails(map[string]any{"result": result}),
+		apperrors.WithCause(cause),
+	)
+}

internal/shortcut/aitable/list_projection_e2e_test.go

@@ -0,0 +1,92 @@
+// Copyright 2026 Alibaba Group
+// SPDX-License-Identifier: Apache-2.0
+
+package aitable
+
+import (
+	"strings"
+	"testing"
+)
+
+func TestCrossPlatformCoverageAITableListProjectionExplicitEmptyIsSuccessE2E(t *testing.T) {
+	tests := []struct {
+		name    string
+		command string
+		args    []string
+		payload string
+	}{
+		{name: "base list", command: "+base-list", payload: `{"bases":[]}`},
+		{name: "base search", command: "+base-search", args: []string{"--query", "none"}, payload: `{"data":{"bases":[]}}`},
+		{name: "template search", command: "+template-search", payload: `{"data":{"templates":[]}}`},
+		{name: "view get", command: "+view-get", args: []string{"--base-id", "base", "--table-id", "table"}, payload: `{"views":[]}`},
+		{name: "form list", command: "+form-list", args: []string{"--base-id", "base", "--table-id", "table"}, payload: `{"data":{"formViews":[]}}`},
+		{name: "workflow list", command: "+workflow-list", args: []string{"--base-id", "base"}, payload: `{"data":{"list":[]}}`},
+	}
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			caller := &upsertByKeyCaller{steps: []upsertByKeyStep{{text: test.payload}}}
+			out, err := runAITableCompositeCLI(t, caller, test.command, test.args...)
+			if err != nil || !strings.Contains(out, `"count": 0`) || len(caller.calls) != 1 {
+				t.Fatalf("explicit empty list = output:%q err:%v calls:%#v", out, err, caller.calls)
+			}
+		})
+	}
+}
+
+func TestCrossPlatformCoverageAITableListProjectionUnknownIsNotEmptySuccessE2E(t *testing.T) {
+	commands := []struct {
+		name    string
+		command string
+		args    []string
+	}{
+		{name: "base list", command: "+base-list"},
+		{name: "base search", command: "+base-search", args: []string{"--query", "none"}},
+		{name: "template search", command: "+template-search"},
+		{name: "view get", command: "+view-get", args: []string{"--base-id", "base", "--table-id", "table"}},
+		{name: "form list", command: "+form-list", args: []string{"--base-id", "base", "--table-id", "table"}},
+		{name: "workflow list", command: "+workflow-list", args: []string{"--base-id", "base"}},
+	}
+	payloads := []struct {
+		name string
+		text string
+	}{
+		{name: "empty response", text: ""},
+		{name: "missing collection", text: `{"success":true}`},
+		{name: "wrong collection type", text: `{"data":{"list":{}}}`},
+	}
+	for _, command := range commands {
+		for _, payload := range payloads {
+			t.Run(command.name+"/"+payload.name, func(t *testing.T) {
+				caller := &upsertByKeyCaller{steps: []upsertByKeyStep{{text: payload.text}}}
+				out, err := runAITableCompositeCLI(t, caller, command.command, command.args...)
+				if err == nil || out != "" || len(caller.calls) != 1 {
+					t.Fatalf("unknown list result was accepted = output:%q err:%v calls:%#v", out, err, caller.calls)
+				}
+			})
+		}
+	}
+}
+
+func TestCrossPlatformCoverageAITableListProjectionMalformedItemIsNotSilentlyDroppedE2E(t *testing.T) {
+	tests := []struct {
+		name    string
+		command string
+		args    []string
+		payload string
+	}{
+		{name: "base without id", command: "+base-list", payload: `{"bases":[{"baseName":"orphan"}]}`},
+		{name: "template scalar", command: "+template-search", payload: `{"templates":["bad"]}`},
+		{name: "view without id", command: "+view-get", args: []string{"--base-id", "base", "--table-id", "table"}, payload: `{"views":[{"viewName":"orphan"}]}`},
+		{name: "form without id", command: "+form-list", args: []string{"--base-id", "base", "--table-id", "table"}, payload: `{"forms":[{"viewName":"orphan"}]}`},
+		{name: "workflow without id", command: "+workflow-list", args: []string{"--base-id", "base"}, payload: `{"workflows":[{"name":"orphan"}]}`},
+	}
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			caller := &upsertByKeyCaller{steps: []upsertByKeyStep{{text: test.payload}}}
+			out, err := runAITableCompositeCLI(t, caller, test.command, test.args...)
+			if err == nil || out != "" || len(caller.calls) != 1 {
+				t.Fatalf("malformed list item was accepted = output:%q err:%v calls:%#v", out, err, caller.calls)
+			}
+		})
+	}
+}

internal/shortcut/aitable/record_batch.go

@@ -0,0 +1,406 @@
+// Copyright 2026 Alibaba Group
+// SPDX-License-Identifier: Apache-2.0
+
+package aitable
+
+import (
+	"fmt"
+	"sort"
+	"strings"
+
+	apperrors "github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/errors"
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/shortcut"
+)
+
+const (
+	recordBatchSize       = 100
+	maxCompositeRecordRun = 10000
+)
+
+func parseRecordObjects(raw string, requireRecordID bool) ([]map[string]any, error) {
+	value, err := parseJSONAny("records", raw)
+	if err != nil {
+		return nil, err
+	}
+	list, ok := value.([]any)
+	if !ok || len(list) == 0 {
+		return nil, apperrors.NewValidation("--records 必须是非空 JSON 数组")
+	}
+	if len(list) > maxCompositeRecordRun {
+		return nil, apperrors.NewValidation(fmt.Sprintf("--records 最多接受 %d 条，got %d", maxCompositeRecordRun, len(list)))
+	}
+	out := make([]map[string]any, 0, len(list))
+	for index, item := range list {
+		record, ok := item.(map[string]any)
+		if !ok {
+			return nil, apperrors.NewValidation(fmt.Sprintf("--records[%d] 必须是 JSON 对象", index))
+		}
+		cells, ok := record["cells"].(map[string]any)
+		if !ok || len(cells) == 0 {
+			return nil, apperrors.NewValidation(fmt.Sprintf("--records[%d].cells 必须是非空 JSON 对象", index))
+		}
+		if requireRecordID && recordID(record) == "" {
+			return nil, apperrors.NewValidation(fmt.Sprintf("--records[%d] 缺少 recordId", index))
+		}
+		out = append(out, record)
+	}
+	return out, nil
+}
+
+func executeRecordUpdateBatches(rt *shortcut.RuntimeContext) error {
+	records, err := parseRecordObjects(rt.Str("records"), true)
+	if err != nil {
+		return err
+	}
+	return executeRecordBatches(rt, "record_update", "update_records", serverMain, records, verifyUpdateBatch)
+}
+
+func executeRecordUpsertBatches(rt *shortcut.RuntimeContext) error {
+	records, err := parseRecordObjects(rt.Str("records"), false)
+	if err != nil {
+		return err
+	}
+	return executeRecordBatches(rt, "record_upsert", "record_upsert", serverHelper, records, verifyUpsertBatch)
+}
+
+func executeRecordDeleteBatches(rt *shortcut.RuntimeContext) error {
+	ids, err := parseRecordIDs(rt.StrSlice("record-ids"))
+	if err != nil {
+		return err
+	}
+	baseID, tableID := rt.Str("base-id"), rt.Str("table-id")
+	result := newCompositeResult("record_delete")
+	result.RequestedCount = len(ids)
+	for offset := 0; offset < len(ids); offset += recordBatchSize {
+		end := minInt(offset+recordBatchSize, len(ids))
+		result.Plan = append(result.Plan, compositeStep{
+			Index: len(result.Plan) + 1, Name: "delete record batch", Tool: "delete_records",
+			Status: "planned", Offset: offset, Count: end - offset,
+		})
+	}
+	if rt.DryRun() {
+		result.Status = "planned"
+		result.Executed = false
+		return rt.Output(result)
+	}
+
+	for offset := 0; offset < len(ids); offset += recordBatchSize {
+		end := minInt(offset+recordBatchSize, len(ids))
+		batch := ids[offset:end]
+		writeData, writeErr := rt.CallMCPWriteDataStrict(serverMain, "delete_records", map[string]any{
+			"baseId": baseID, "tableId": tableID, "recordIds": batch,
+		})
+		step := compositeStep{
+			Index: len(result.CompletedSteps) + 1, Name: "delete record batch", Tool: "delete_records",
+			Status: "completed", Offset: offset, Count: len(batch), Result: writeData,
+		}
+		if writeErr != nil {
+			step.Status = "unknown"
+			step.Error = writeErr.Error()
+		}
+		remaining, verifyErr := queryRecordsByIDs(rt, baseID, tableID, batch)
+		if verifyErr == nil && len(remaining) > 0 {
+			verifyErr = fmt.Errorf("read-back still contains deleted record IDs: %s", strings.Join(recordIDs(remaining), ","))
+		}
+		if verifyErr != nil {
+			if step.Error == "" {
+				step.Error = verifyErr.Error()
+			}
+			step.Status = "unknown"
+			result.CompletedSteps = append(result.CompletedSteps, step)
+			result.CompletedCount = offset
+			result.FailedCount = len(ids) - offset
+			result.Status = "unknown"
+			if offset > 0 {
+				result.Status = "partial_success"
+			}
+			result.Verification = map[string]any{"status": "failed", "error": verifyErr.Error(), "batchOffset": offset}
+			result.Checkpoint = map[string]any{"nextOffset": offset, "batchSize": recordBatchSize}
+			if writeErr != nil {
+				result.Warnings = append(result.Warnings, "write response error: "+writeErr.Error())
+			}
+			return compositeError(result, verifyErr, true)
+		}
+		if writeErr != nil {
+			step.Status = "recovered"
+			result.Warnings = append(result.Warnings, fmt.Sprintf("batch at offset %d had a write-response error but deletion was proven by read-back", offset))
+		}
+		step.Result = map[string]any{"status": "verified_absent", "recordIds": batch}
+		result.CompletedSteps = append(result.CompletedSteps, step)
+		result.CompletedCount = end
+		result.KnownEffects = append(result.KnownEffects, map[string]any{"tool": "delete_records", "offset": offset, "recordIds": batch})
+	}
+	result.Verification = map[string]any{"status": "verified_absent", "verifiedCount": len(ids)}
+	result.Result = map[string]any{"deletedCount": len(ids), "batchCount": len(result.CompletedSteps)}
+	return rt.Output(result)
+}
+
+func parseRecordIDs(values []string) ([]string, error) {
+	seen := map[string]bool{}
+	ids := make([]string, 0, len(values))
+	for _, value := range values {
+		id := strings.TrimSpace(value)
+		if id == "" {
+			continue
+		}
+		if !seen[id] {
+			seen[id] = true
+			ids = append(ids, id)
+		}
+	}
+	if len(ids) == 0 {
+		return nil, apperrors.NewValidation("--record-ids 必须至少包含一个非空 recordId")
+	}
+	if len(ids) > maxCompositeRecordRun {
+		return nil, apperrors.NewValidation(fmt.Sprintf("--record-ids 最多接受 %d 个唯一 ID，got %d", maxCompositeRecordRun, len(ids)))
+	}
+	// Stable ordering makes checkpoints and retries deterministic even when the
+	// flag parser or a caller supplied IDs through more than one occurrence.
+	sort.Strings(ids)
+	return ids, nil
+}
+
+type recordBatchVerifier func(*shortcut.RuntimeContext, string, string, []map[string]any, map[string]any) (map[string]any, error)
+
+func executeRecordBatches(
+	rt *shortcut.RuntimeContext,
+	operation, tool, product string,
+	records []map[string]any,
+	verify recordBatchVerifier,
+) error {
+	baseID, tableID := rt.Str("base-id"), rt.Str("table-id")
+	result := newCompositeResult(operation)
+	result.RequestedCount = len(records)
+	for offset := 0; offset < len(records); offset += recordBatchSize {
+		end := minInt(offset+recordBatchSize, len(records))
+		result.Plan = append(result.Plan, compositeStep{
+			Index: len(result.Plan) + 1, Name: "write record batch", Tool: tool,
+			Status: "planned", Offset: offset, Count: end - offset,
+		})
+	}
+	if rt.DryRun() {
+		result.Status = "planned"
+		result.Executed = false
+		return rt.Output(result)
+	}
+
+	for offset := 0; offset < len(records); offset += recordBatchSize {
+		end := minInt(offset+recordBatchSize, len(records))
+		batch := records[offset:end]
+		wireRecords := make([]any, 0, len(batch))
+		for _, record := range batch {
+			wireRecords = append(wireRecords, record)
+		}
+		params := map[string]any{"baseId": baseID, "tableId": tableID, "records": wireRecords}
+		writeData, writeErr := rt.CallMCPWriteDataStrict(product, tool, params)
+		step := compositeStep{
+			Index: len(result.CompletedSteps) + 1, Name: "write record batch", Tool: tool,
+			Status: "completed", Offset: offset, Count: len(batch), Result: writeData,
+		}
+		if writeErr != nil {
+			step.Status = "unknown"
+			step.Error = writeErr.Error()
+		}
+		verification, verifyErr := verify(rt, baseID, tableID, batch, writeData)
+		if verifyErr != nil {
+			step.Status = "unknown"
+			if step.Error == "" {
+				step.Error = verifyErr.Error()
+			}
+			result.CompletedSteps = append(result.CompletedSteps, step)
+			result.CompletedCount = offset
+			result.FailedCount = len(records) - offset
+			result.Status = "unknown"
+			if offset > 0 {
+				result.Status = "partial_success"
+			}
+			result.Verification = map[string]any{"status": "failed", "error": verifyErr.Error(), "batchOffset": offset}
+			result.Checkpoint = map[string]any{"nextOffset": offset, "batchSize": recordBatchSize}
+			if writeErr != nil {
+				result.Warnings = append(result.Warnings, "write response error: "+writeErr.Error())
+			}
+			retryable := true
+			if tool == "record_upsert" {
+				for _, record := range batch {
+					if recordID(record) == "" {
+						retryable = false
+						break
+					}
+				}
+			}
+			return compositeError(result, verifyErr, retryable)
+		}
+		if writeErr != nil {
+			step.Status = "recovered"
+			result.Warnings = append(result.Warnings, fmt.Sprintf("batch at offset %d had a write-response error but its final state was proven by read-back", offset))
+		}
+		step.Result = verification
+		result.CompletedSteps = append(result.CompletedSteps, step)
+		result.CompletedCount = end
+		result.KnownEffects = append(result.KnownEffects, map[string]any{"tool": tool, "offset": offset, "count": len(batch)})
+	}
+	result.Verification = map[string]any{"status": "verified", "verifiedCount": len(records)}
+	result.Result = map[string]any{"processedCount": len(records), "batchCount": len(result.CompletedSteps)}
+	return rt.Output(result)
+}
+
+func verifyUpdateBatch(rt *shortcut.RuntimeContext, baseID, tableID string, batch []map[string]any, _ map[string]any) (map[string]any, error) {
+	ids := recordIDs(batch)
+	actual, err := queryRecordsByIDs(rt, baseID, tableID, ids)
+	if err != nil {
+		return nil, err
+	}
+	byID := make(map[string]map[string]any, len(actual))
+	for _, record := range actual {
+		byID[recordID(record)] = record
+	}
+	for _, expected := range batch {
+		id := recordID(expected)
+		got := byID[id]
+		if got == nil {
+			return nil, fmt.Errorf("read-back is missing updated record %s", id)
+		}
+		if err := verifyRecordCells(got, expected["cells"].(map[string]any)); err != nil {
+			return nil, err
+		}
+	}
+	return map[string]any{"status": "verified", "recordIds": ids}, nil
+}
+
+func verifyUpsertBatch(rt *shortcut.RuntimeContext, baseID, tableID string, batch []map[string]any, writeData map[string]any) (map[string]any, error) {
+	updates := make([]map[string]any, 0)
+	creates := make([]map[string]any, 0)
+	for _, record := range batch {
+		if recordID(record) == "" {
+			creates = append(creates, record)
+		} else {
+			updates = append(updates, record)
+		}
+	}
+	verifiedIDs := make([]string, 0, len(batch))
+	if len(updates) > 0 {
+		verified, err := verifyUpdateBatch(rt, baseID, tableID, updates, writeData)
+		if err != nil {
+			return nil, err
+		}
+		verifiedIDs = append(verifiedIDs, stringSlice(verified["recordIds"])...)
+	}
+	if len(creates) > 0 {
+		createdIDs := createdRecordIDs(writeData)
+		if len(createdIDs) != len(creates) {
+			return nil, fmt.Errorf("record_upsert returned %d created record IDs for %d creates", len(createdIDs), len(creates))
+		}
+		actual, err := queryRecordsByIDs(rt, baseID, tableID, createdIDs)
+		if err != nil {
+			return nil, err
+		}
+		if err := matchCreatedCells(creates, actual); err != nil {
+			return nil, err
+		}
+		verifiedIDs = append(verifiedIDs, createdIDs...)
+	}
+	return map[string]any{"status": "verified", "recordIds": verifiedIDs}, nil
+}
+
+func queryRecordsByIDs(rt *shortcut.RuntimeContext, baseID, tableID string, ids []string) ([]map[string]any, error) {
+	data, err := rt.CallMCPData(serverMain, "query_records", map[string]any{
+		"baseId": baseID, "tableId": tableID, "recordIds": ids, "limit": len(ids),
+	})
+	if err != nil {
+		return nil, err
+	}
+	records, found := findRecords(data)
+	if !found {
+		return nil, fmt.Errorf("query_records read-back is missing records")
+	}
+	if responseHasMore(data) {
+		return nil, fmt.Errorf("query_records read-back is incomplete")
+	}
+	return records, nil
+}
+
+func recordIDs(records []map[string]any) []string {
+	out := make([]string, 0, len(records))
+	seen := map[string]bool{}
+	for _, record := range records {
+		id := recordID(record)
+		if id != "" && !seen[id] {
+			seen[id] = true
+			out = append(out, id)
+		}
+	}
+	return out
+}
+
+func createdRecordIDs(data map[string]any) []string {
+	seen := map[string]bool{}
+	out := make([]string, 0)
+	var walk func(any, bool)
+	walk = func(value any, createdContext bool) {
+		switch typed := value.(type) {
+		case map[string]any:
+			if createdContext {
+				if id := recordID(typed); id != "" && !seen[id] {
+					seen[id] = true
+					out = append(out, id)
+				}
+			}
+			for key, child := range typed {
+				lower := strings.ToLower(key)
+				nextContext := createdContext || strings.Contains(lower, "created")
+				walk(child, nextContext)
+			}
+		case []any:
+			for _, child := range typed {
+				if createdContext {
+					if id, ok := child.(string); ok {
+						id = strings.TrimSpace(id)
+						if id != "" && !seen[id] {
+							seen[id] = true
+							out = append(out, id)
+							continue
+						}
+					}
+				}
+				walk(child, createdContext)
+			}
+		}
+	}
+	walk(data, false)
+	return out
+}
+
+func matchCreatedCells(expected, actual []map[string]any) error {
+	if len(actual) != len(expected) {
+		return fmt.Errorf("read-back returned %d created records, want %d", len(actual), len(expected))
+	}
+	used := make([]bool, len(actual))
+	for _, wantRecord := range expected {
+		wantCells := wantRecord["cells"].(map[string]any)
+		matched := false
+		for index, gotRecord := range actual {
+			if used[index] || verifyRecordCells(gotRecord, wantCells) != nil {
+				continue
+			}
+			used[index] = true
+			matched = true
+			break
+		}
+		if !matched {
+			return fmt.Errorf("no created read-back record matches cells %#v", wantCells)
+		}
+	}
+	return nil
+}
+
+func stringSlice(value any) []string {
+	values, _ := value.([]string)
+	return values
+}
+
+func minInt(a, b int) int {
+	if a < b {
+		return a
+	}
+	return b
+}

internal/shortcut/aitable/record_bulk_patch.go

@@ -0,0 +1,176 @@
+// Copyright 2026 Alibaba Group
+// SPDX-License-Identifier: Apache-2.0
+
+package aitable
+
+import (
+	"fmt"
+	"strings"
+
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/corecmd/contract"
+	apperrors "github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/errors"
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/shortcut"
+)
+
+var RecordBulkPatch = shortcut.Shortcut{
+	Service:     "aitable",
+	Command:     "+record-bulk-patch",
+	Product:     serverMain,
+	Description: "完整查询目标记录后批量合并同一组 cells，自动分片并逐条读回验证",
+	Intent:      "当你要按 filters/关键词/recordIds 对一批已有记录应用相同字段补丁时使用；必须显式选择范围或 --all，并受 max-matches 写前上限保护。",
+	Risk:        shortcut.RiskHighWrite,
+	Safety: contract.SafetySpec{
+		Effect: "write", Risk: "high", Confirmation: "user_required", Idempotency: "idempotent",
+	},
+	Contract: aitableCompositeContract(
+		"+record-bulk-patch",
+		"完整查询目标记录后批量合并同一组 cells，自动分片并逐条读回验证",
+		"当你要按 filters/关键词/recordIds 对一批已有记录应用相同字段补丁时使用；必须显式选择范围或 --all，并受 max-matches 写前上限保护。",
+		"单条已知 recordId 用 record update；每条记录写不同值用 record update/upsert；未明确范围时不要执行",
+		`dws aitable +record-bulk-patch --base-id B --table-id T --query "待处理" --patch '{"fldStatus":"完成"}'`,
+	),
+	Flags: []shortcut.Flag{
+		{Name: "base-id", Type: shortcut.FlagString, Desc: "Base ID", Required: true},
+		{Name: "table-id", Type: shortcut.FlagString, Desc: "Table ID", Required: true},
+		{Name: "patch", Type: shortcut.FlagString, Desc: "要合并到每条记录的非空 cells JSON 对象", Required: true},
+		{Name: "filters", Type: shortcut.FlagString, Desc: "query_records filters JSON（选择条件之一）"},
+		{Name: "query", Type: shortcut.FlagString, Desc: "全文关键词（选择条件之一）"},
+		{Name: "record-ids", Type: shortcut.FlagStringSlice, Desc: "明确的 recordId 列表（选择条件之一）"},
+		{Name: "view-id", Type: shortcut.FlagString, Desc: "可选视图上下文"},
+		{Name: "all", Type: shortcut.FlagBool, Desc: "明确允许匹配整张表"},
+		{Name: "max-matches", Type: shortcut.FlagInt, Default: "1000", Desc: "写入前允许匹配的最大记录数，1-10000"},
+	},
+	Tips: []string{`dws aitable +record-bulk-patch --base-id B --table-id T --query "待处理" --patch '{"fldStatus":"完成"}'`},
+	Execute: func(rt *shortcut.RuntimeContext) error {
+		return executeRecordBulkPatch(rt)
+	},
+}
+
+func executeRecordBulkPatch(rt *shortcut.RuntimeContext) error {
+	patch, err := parseJSONObject("patch", rt.Str("patch"))
+	if err != nil {
+		return err
+	}
+	if len(patch) == 0 {
+		return apperrors.NewValidation("--patch 必须是非空 JSON 对象")
+	}
+	params := map[string]any{"baseId": rt.Str("base-id"), "tableId": rt.Str("table-id")}
+	selectorCount := 0
+	if rt.Changed("filters") {
+		filters, err := parseBulkPatchFilters(rt.Str("filters"))
+		if err != nil {
+			return err
+		}
+		params["filters"] = filters
+		selectorCount++
+	}
+	if rt.Changed("query") {
+		query := strings.TrimSpace(rt.Str("query"))
+		if query == "" {
+			return apperrors.NewValidation("--query 必须包含非空白关键词")
+		}
+		params["keyword"] = query
+		selectorCount++
+	}
+	if rt.Changed("record-ids") {
+		ids, err := parseRecordIDs(rt.StrSlice("record-ids"))
+		if err != nil {
+			return err
+		}
+		params["recordIds"] = ids
+		selectorCount++
+	}
+	if rt.Changed("all") {
+		if !rt.Bool("all") {
+			return apperrors.NewValidation("--all=false 不构成全表授权")
+		}
+		selectorCount++
+	}
+	if selectorCount == 0 {
+		return apperrors.NewValidation("必须显式提供 --filters、--query、--record-ids 或 --all")
+	}
+	maxMatches := rt.Int("max-matches")
+	if maxMatches < 1 || maxMatches > maxCompositeRecordRun {
+		return apperrors.NewValidation(fmt.Sprintf("--max-matches 必须在 1..%d", maxCompositeRecordRun))
+	}
+	if rt.Changed("view-id") {
+		params["viewId"] = rt.Str("view-id")
+	}
+	records, err := queryAllRecords(rt, params, maxMatches)
+	if err != nil {
+		return err
+	}
+	updates := make([]map[string]any, 0, len(records))
+	for _, record := range records {
+		id := recordID(record)
+		if id == "" {
+			return fmt.Errorf("query_records returned a matched record without recordId")
+		}
+		updates = append(updates, map[string]any{"recordId": id, "cells": cloneAnyMap(patch)})
+	}
+	if len(updates) == 0 {
+		result := newCompositeResult("record_bulk_patch")
+		result.RequestedCount = 0
+		result.Verification = map[string]any{"status": "verified_no_matches"}
+		result.Result = map[string]any{"matchedCount": 0, "updatedCount": 0}
+		if rt.DryRun() {
+			result.Status = "planned"
+			result.Executed = false
+		}
+		return rt.Output(result)
+	}
+	return executeRecordBatches(rt, "record_bulk_patch", "update_records", serverMain, updates, verifyUpdateBatch)
+}
+
+func parseBulkPatchFilters(raw string) (map[string]any, error) {
+	filters, err := parseJSONObject("filters", raw)
+	if err != nil {
+		return nil, err
+	}
+	if !bulkPatchFilterConstrictsScope(filters, true) {
+		return nil, apperrors.NewValidation("--filters 必须是包含非空筛选条件的 JSON 对象，根 operator 必须为 and/or")
+	}
+	return filters, nil
+}
+
+func bulkPatchFilterConstrictsScope(value any, root bool) bool {
+	filter, ok := value.(map[string]any)
+	if !ok {
+		return false
+	}
+	operator, ok := filter["operator"].(string)
+	if !ok {
+		return false
+	}
+	operator = strings.ToLower(strings.TrimSpace(operator))
+	if operator == "" {
+		return false
+	}
+	operands, ok := filter["operands"].([]any)
+	if !ok || len(operands) == 0 {
+		return false
+	}
+	if root && operator != "and" && operator != "or" {
+		return false
+	}
+	if operator == "and" || operator == "or" {
+		for _, operand := range operands {
+			if !bulkPatchFilterConstrictsScope(operand, false) {
+				return false
+			}
+		}
+		return true
+	}
+	switch operator {
+	case "eq", "ne", "gt", "lt", "gte", "lte", "contain", "exclusive",
+		"exist", "un_exist", "any_of", "all_of", "none_of", "date_eq",
+		"before", "after", "not_before", "not_after":
+	default:
+		return false
+	}
+	fieldID, ok := operands[0].(string)
+	if !ok || strings.TrimSpace(fieldID) == "" {
+		return false
+	}
+	return (operator == "exist" || operator == "un_exist") || len(operands) >= 2
+}

internal/shortcut/aitable/record_coverage_gap_test.go

@@ -0,0 +1,368 @@
+// Copyright 2026 Alibaba Group
+// SPDX-License-Identifier: Apache-2.0
+
+package aitable
+
+import (
+	"errors"
+	"fmt"
+	"strconv"
+	"strings"
+	"testing"
+)
+
+func TestCrossPlatformCoverageRecordObjectAndIDValidation(t *testing.T) {
+	tooMany := make([]any, maxCompositeRecordRun+1)
+	for index := range tooMany {
+		tooMany[index] = map[string]any{"cells": map[string]any{"f": index}}
+	}
+	cases := []struct {
+		name      string
+		raw       string
+		requireID bool
+	}{
+		{name: "invalid JSON", raw: `{`},
+		{name: "not array", raw: `{}`},
+		{name: "empty array", raw: `[]`},
+		{name: "too many", raw: mustJSONText(t, tooMany)},
+		{name: "non-object", raw: `[1]`},
+		{name: "missing cells", raw: `[{"recordId":"r"}]`},
+		{name: "empty cells", raw: `[{"recordId":"r","cells":{}}]`},
+		{name: "missing record id", raw: `[{"cells":{"f":1}}]`, requireID: true},
+	}
+	for _, tc := range cases {
+		t.Run(tc.name, func(t *testing.T) {
+			if records, err := parseRecordObjects(tc.raw, tc.requireID); err == nil || records != nil {
+				t.Fatalf("parseRecordObjects = %#v, %v", records, err)
+			}
+		})
+	}
+	if ids, err := parseRecordIDs([]string{" ", "b", "a", "a"}); err != nil || strings.Join(ids, ",") != "a,b" {
+		t.Fatalf("parseRecordIDs = %#v, %v", ids, err)
+	}
+	tooManyIDs := make([]string, maxCompositeRecordRun+1)
+	for index := range tooManyIDs {
+		tooManyIDs[index] = strconv.Itoa(index)
+	}
+	if ids, err := parseRecordIDs(tooManyIDs); err == nil || ids != nil {
+		t.Fatalf("too many IDs = %#v, %v", ids, err)
+	}
+}
+
+func TestCrossPlatformCoverageRecordBatchEntryValidationE2E(t *testing.T) {
+	for _, command := range []string{"+record-update", "+record-upsert"} {
+		t.Run(command, func(t *testing.T) {
+			out, err := runAITableCompositeCLI(t, &upsertByKeyCaller{}, command,
+				"--base-id", "base", "--table-id", "table", "--records", `{`, "--yes")
+			if err == nil || out != "" {
+				t.Fatalf("invalid records = output:%q err:%v", out, err)
+			}
+		})
+	}
+	out, err := runAITableCompositeCLI(t, &upsertByKeyCaller{}, "+record-delete",
+		"--base-id", "base", "--table-id", "table", "--record-ids", ` `, "--yes")
+	if err == nil || out != "" {
+		t.Fatalf("invalid record IDs = output:%q err:%v", out, err)
+	}
+}
+
+func TestCrossPlatformCoverageRecordDeleteDryRunAndDualFailureE2E(t *testing.T) {
+	caller := &upsertByKeyCaller{dryRun: true}
+	out, err := runRecordDeleteCLI(t, caller, []string{"r1"}, "--dry-run")
+	if err != nil || len(caller.calls) != 0 || !strings.Contains(out, `"status": "planned"`) {
+		t.Fatalf("delete dry run = output:%q err:%v calls:%#v", out, err, caller.calls)
+	}
+
+	caller = &upsertByKeyCaller{steps: []upsertByKeyStep{{err: errors.New("write failed")}, {err: errors.New("verify failed")}}}
+	out, err = runRecordDeleteCLI(t, caller, []string{"r1"})
+	if err == nil || out != "" {
+		t.Fatalf("delete dual failure = output:%q err:%v", out, err)
+	}
+
+	out, err = runRecordDeleteCLI(t, &upsertByKeyCaller{}, []string{"", " "})
+	if err == nil || out != "" {
+		t.Fatalf("delete parsed empty IDs = output:%q err:%v", out, err)
+	}
+
+	out, err = runRecordDeleteCLI(t, &upsertByKeyCaller{}, recordIDFixtures(maxCompositeRecordRun+1))
+	if err == nil || out != "" {
+		t.Fatalf("delete excessive IDs = output:%q err:%v", out, err)
+	}
+}
+
+func TestCrossPlatformCoverageRecordBatchVerificationEdgesE2E(t *testing.T) {
+	record := []map[string]any{{"recordId": "r1", "cells": map[string]any{"f": "new"}}}
+
+	t.Run("read-back query error", func(t *testing.T) {
+		caller := &upsertByKeyCaller{steps: []upsertByKeyStep{{text: `{"updatedCount":1}`}, {err: errors.New("read failed")}}}
+		out, err := runRecordBatchCLI(t, caller, "+record-update", record)
+		if err == nil || out != "" {
+			t.Fatalf("query error = output:%q err:%v", out, err)
+		}
+	})
+
+	t.Run("read-back cells mismatch", func(t *testing.T) {
+		caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+			{text: `{"updatedCount":1}`},
+			{text: `{"records":[{"recordId":"r1","cells":{"f":"old"}}]}`},
+		}}
+		out, err := runRecordBatchCLI(t, caller, "+record-update", record)
+		if err == nil || out != "" {
+			t.Fatalf("cell mismatch = output:%q err:%v", out, err)
+		}
+	})
+
+	t.Run("upsert create read error", func(t *testing.T) {
+		caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+			{text: `{"createdRecords":[{"recordId":"new"}]}`},
+			{err: errors.New("created read failed")},
+		}}
+		out, err := runRecordBatchCLI(t, caller, "+record-upsert", []map[string]any{{"cells": map[string]any{"f": "new"}}})
+		if err == nil || out != "" {
+			t.Fatalf("create read error = output:%q err:%v", out, err)
+		}
+	})
+
+	t.Run("upsert create cells mismatch", func(t *testing.T) {
+		caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+			{text: `{"createdRecords":[{"recordId":"new"}]}`},
+			{text: `{"records":[{"recordId":"new","cells":{"f":"old"}}]}`},
+		}}
+		out, err := runRecordBatchCLI(t, caller, "+record-upsert", []map[string]any{{"cells": map[string]any{"f": "new"}}})
+		if err == nil || out != "" {
+			t.Fatalf("create mismatch = output:%q err:%v", out, err)
+		}
+	})
+
+	t.Run("upsert update read error", func(t *testing.T) {
+		caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+			{text: `{"createdRecords":[],"updatedCount":1}`},
+			{err: errors.New("updated read failed")},
+		}}
+		out, err := runRecordBatchCLI(t, caller, "+record-upsert", record)
+		if err == nil || out != "" {
+			t.Fatalf("update read error = output:%q err:%v", out, err)
+		}
+	})
+
+	t.Run("incomplete update read", func(t *testing.T) {
+		caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+			{text: `{"updatedCount":1}`},
+			{text: `{"records":[{"recordId":"r1","cells":{"f":"new"}}],"hasMore":true}`},
+		}}
+		out, err := runRecordBatchCLI(t, caller, "+record-update", record)
+		if err == nil || out != "" {
+			t.Fatalf("incomplete read = output:%q err:%v", out, err)
+		}
+	})
+}
+
+func TestCrossPlatformCoverageRecordBatchPureHelpers(t *testing.T) {
+	if err := matchCreatedCells([]map[string]any{{"cells": map[string]any{"f": 1}}}, nil); err == nil {
+		t.Fatal("created result count mismatch must fail")
+	}
+	if err := matchCreatedCells(
+		[]map[string]any{{"cells": map[string]any{"f": 1}}},
+		[]map[string]any{{"recordId": "r", "cells": map[string]any{"f": 2}}},
+	); err == nil {
+		t.Fatal("unmatched created cells must fail")
+	}
+	ids := createdRecordIDs(map[string]any{"createdIds": []any{" r1 ", "r1", map[string]any{"record_id": "r2"}}})
+	if strings.Join(ids, ",") != "r1,r2" {
+		t.Fatalf("created IDs = %#v", ids)
+	}
+	if got := stringSlice("bad"); got != nil {
+		t.Fatalf("stringSlice = %#v", got)
+	}
+	if minInt(1, 2) != 1 || minInt(2, 1) != 1 {
+		t.Fatal("minInt branch mismatch")
+	}
+}
+
+func TestCrossPlatformCoverageRecordBulkPatchValidationAndSelectorsE2E(t *testing.T) {
+	cases := []struct {
+		name string
+		args []string
+	}{
+		{name: "invalid patch", args: []string{"--all", "--patch", `{`}},
+		{name: "empty patch", args: []string{"--all", "--patch", `{}`}},
+		{name: "false all", args: []string{"--all=false", "--patch", `{"f":1}`}},
+		{name: "invalid max", args: []string{"--all", "--patch", `{"f":1}`, "--max-matches", "0"}},
+		{name: "invalid filters", args: []string{"--filters", `{`, "--patch", `{"f":1}`}},
+		{name: "null filters", args: []string{"--filters", `null`, "--patch", `{"f":1}`}},
+		{name: "empty filters object", args: []string{"--filters", `{}`, "--patch", `{"f":1}`}},
+		{name: "empty filters operands", args: []string{"--filters", `{"operator":"and","operands":[]}`, "--patch", `{"f":1}`}},
+		{name: "missing filters operator", args: []string{"--filters", `{"operands":[{"operator":"eq","operands":["fld","x"]}]}`, "--patch", `{"f":1}`}},
+		{name: "blank child operator", args: []string{"--filters", `{"operator":"and","operands":[{"operator":" ","operands":["fld","x"]}]}`, "--patch", `{"f":1}`}},
+		{name: "null filters operands", args: []string{"--filters", `{"operator":"and","operands":null}`, "--patch", `{"f":1}`}},
+		{name: "non condition operand", args: []string{"--filters", `{"operator":"and","operands":[null]}`, "--patch", `{"f":1}`}},
+		{name: "root comparison", args: []string{"--filters", `{"operator":"eq","operands":["fld","x"]}`, "--patch", `{"f":1}`}},
+		{name: "blank filter field", args: []string{"--filters", `{"operator":"and","operands":[{"operator":"eq","operands":[" ","x"]}]}`, "--patch", `{"f":1}`}},
+		{name: "unknown filter operator", args: []string{"--filters", `{"operator":"and","operands":[{"operator":"unknown","operands":["fld","x"]}]}`, "--patch", `{"f":1}`}},
+		{name: "missing comparison value", args: []string{"--filters", `{"operator":"and","operands":[{"operator":"eq","operands":["fld"]}]}`, "--patch", `{"f":1}`}},
+		{name: "blank query", args: []string{"--query", ` `, "--patch", `{"f":1}`}},
+		{name: "empty record IDs", args: []string{"--record-ids", ` `, "--patch", `{"f":1}`}},
+	}
+	for _, tc := range cases {
+		t.Run(tc.name, func(t *testing.T) {
+			args := []string{"--base-id", "base", "--table-id", "table"}
+			args = append(args, tc.args...)
+			args = append(args, "--yes")
+			caller := &upsertByKeyCaller{}
+			out, err := runAITableCompositeCLI(t, caller, "+record-bulk-patch", args...)
+			if err == nil || out != "" || len(caller.calls) != 0 {
+				t.Fatalf("validation = output:%q err:%v calls:%#v", out, err, caller.calls)
+			}
+		})
+	}
+
+	caller := &upsertByKeyCaller{steps: []upsertByKeyStep{{text: `{"records":[]}`}}}
+	out, err := runAITableCompositeCLI(t, caller, "+record-bulk-patch",
+		"--base-id", "base", "--table-id", "table", "--filters", `{"operator":"and","operands":[{"operator":"or","operands":[{"operator":"eq","operands":["fld","x"]}]}]}`, "--query", "text",
+		"--record-ids", "r2,r1", "--view-id", "view", "--patch", `{"f":1}`, "--yes")
+	if err != nil || !strings.Contains(out, `"matchedCount": 0`) {
+		t.Fatalf("selectors = output:%q err:%v", out, err)
+	}
+	for _, key := range []string{"filters", "keyword", "recordIds", "viewId"} {
+		if _, ok := caller.calls[0].args[key]; !ok {
+			t.Fatalf("query args missing %s: %#v", key, caller.calls[0].args)
+		}
+	}
+
+	caller = &upsertByKeyCaller{dryRun: true, steps: []upsertByKeyStep{{text: `{"records":[]}`}}}
+	out, err = runAITableCompositeCLI(t, caller, "+record-bulk-patch",
+		"--base-id", "base", "--table-id", "table", "--query", "none", "--patch", `{"f":1}`, "--dry-run", "--yes")
+	if err != nil || !strings.Contains(out, `"status": "planned"`) {
+		t.Fatalf("empty dry run = output:%q err:%v", out, err)
+	}
+}
+
+func TestCrossPlatformCoverageRecordBulkPatchQueryFailuresE2E(t *testing.T) {
+	cases := []struct {
+		name  string
+		steps []upsertByKeyStep
+	}{
+		{name: "query error", steps: []upsertByKeyStep{{err: errors.New("query failed")}}},
+		{name: "missing records", steps: []upsertByKeyStep{{text: `{}`}}},
+		{name: "more without cursor", steps: []upsertByKeyStep{{text: `{"records":[],"hasMore":true}`}}},
+		{name: "cursor cycle", steps: []upsertByKeyStep{
+			{text: `{"records":[],"hasMore":true,"nextCursor":"same"}`},
+			{text: `{"records":[],"hasMore":true,"nextCursor":"same"}`},
+		}},
+		{name: "matched record missing id", steps: []upsertByKeyStep{{text: `{"records":[{"cells":{"f":1}}]}`}}},
+	}
+	for _, tc := range cases {
+		t.Run(tc.name, func(t *testing.T) {
+			out, err := runAITableCompositeCLI(t, &upsertByKeyCaller{steps: tc.steps}, "+record-bulk-patch",
+				"--base-id", "base", "--table-id", "table", "--all", "--patch", `{"f":1}`, "--yes")
+			if err == nil || out != "" {
+				t.Fatalf("query failure = output:%q err:%v", out, err)
+			}
+		})
+	}
+}
+
+func TestCrossPlatformCoverageRecordQueryPageSafetyBoundE2E(t *testing.T) {
+	caller := &upsertByKeyCaller{callFn: func(index int, _, tool string, _ map[string]any) (string, error) {
+		if tool != "query_records" {
+			return "", fmt.Errorf("unexpected tool %s", tool)
+		}
+		return fmt.Sprintf(`{"records":[],"hasMore":true,"nextCursor":"cursor-%d"}`, index), nil
+	}}
+	out, err := runAITableCompositeCLI(t, caller, "+record-bulk-patch",
+		"--base-id", "base", "--table-id", "table", "--all", "--patch", `{"f":1}`, "--yes")
+	if err == nil || out != "" || len(caller.calls) != 10000 {
+		t.Fatalf("page safety = output:%q err:%v calls:%d", out, err, len(caller.calls))
+	}
+}
+
+func TestCrossPlatformCoverageRecordUpsertInputAndQueryFailuresE2E(t *testing.T) {
+	baseArgs := []string{"--base-id", "base", "--table-id", "table", "--key-field-id", "key", "--cells", `{"value":1}`, "--yes"}
+	cases := []struct {
+		name  string
+		key   []string
+		cells string
+		steps []upsertByKeyStep
+	}{
+		{name: "empty string key", key: []string{"--key-value", " "}},
+		{name: "invalid JSON key", key: []string{"--key-value-json", `{`}},
+		{name: "null JSON key", key: []string{"--key-value-json", `null`}},
+		{name: "object JSON key", key: []string{"--key-value-json", `{}`}},
+		{name: "invalid cells", key: []string{"--key-value", "v"}, cells: `{`},
+		{name: "conflicting key", key: []string{"--key-value", "v"}, cells: `{"key":"other"}`},
+		{name: "query error", key: []string{"--key-value", "v"}, steps: []upsertByKeyStep{{err: errors.New("query failed")}}},
+		{name: "missing records", key: []string{"--key-value", "v"}, steps: []upsertByKeyStep{{text: `{}`}}},
+		{name: "incomplete unique query", key: []string{"--key-value", "v"}, steps: []upsertByKeyStep{{text: `{"records":[],"hasMore":true}`}}},
+		{name: "record missing id", key: []string{"--key-value", "v"}, steps: []upsertByKeyStep{{text: `{"records":[{"cells":{"key":"v"}}]}`}}},
+	}
+	for _, tc := range cases {
+		t.Run(tc.name, func(t *testing.T) {
+			args := append([]string{}, baseArgs...)
+			args = append(args, tc.key...)
+			if tc.cells != "" {
+				for index := range args {
+					if args[index] == "--cells" {
+						args[index+1] = tc.cells
+					}
+				}
+			}
+			out, err := runAITableCompositeCLI(t, &upsertByKeyCaller{steps: tc.steps}, "+record-upsert-by-key", args...)
+			if err == nil || out != "" {
+				t.Fatalf("upsert failure = output:%q err:%v", out, err)
+			}
+		})
+	}
+}
+
+func TestCrossPlatformCoverageRecordUpsertJSONScalarSuccessE2E(t *testing.T) {
+	caller := &upsertByKeyCaller{dryRun: true, steps: []upsertByKeyStep{{text: `{"records":[]}`}}}
+	out, err := runAITableCompositeCLI(t, caller, "+record-upsert-by-key",
+		"--base-id", "base", "--table-id", "table", "--key-field-id", "key", "--key-value-json", `true`,
+		"--cells", `{"value":1}`, "--dry-run", "--yes")
+	if err != nil || !strings.Contains(out, `"keyValue": true`) {
+		t.Fatalf("JSON scalar = output:%q err:%v", out, err)
+	}
+}
+
+func TestCrossPlatformCoverageRecordUpsertRejectsTrailingJSONBeforeMCPE2E(t *testing.T) {
+	for name, value := range map[string]string{
+		"trailing garbage":    `1 garbage`,
+		"second JSON value":   `1 2`,
+		"second string value": `true "extra"`,
+	} {
+		t.Run(name, func(t *testing.T) {
+			caller := &upsertByKeyCaller{}
+			out, err := runAITableCompositeCLI(t, caller, "+record-upsert-by-key",
+				"--base-id", "base", "--table-id", "table", "--key-field-id", "key", "--key-value-json", value,
+				"--cells", `{"value":1}`, "--yes")
+			if err == nil || out != "" || len(caller.calls) != 0 {
+				t.Fatalf("trailing JSON = output:%q err:%v calls:%#v", out, err, caller.calls)
+			}
+		})
+	}
+}
+
+func TestCrossPlatformCoverageRecordShapeHelpers(t *testing.T) {
+	if records, ok := findRecords(nil); ok || records != nil {
+		t.Fatalf("nil records = %#v, %v", records, ok)
+	}
+	if _, ok := findRecords(map[string]any{"records": "bad"}); ok {
+		t.Fatal("non-array records must fail")
+	}
+	if _, ok := findRecords(map[string]any{"records": []any{"bad"}}); ok {
+		t.Fatal("non-object record must fail")
+	}
+	if records, ok := findRecords(map[string]any{"result": map[string]any{"records": []any{map[string]any{"id": "r"}}}}); !ok || recordID(records[0]) != "r" {
+		t.Fatalf("nested records = %#v, %v", records, ok)
+	}
+	if responseHasMore(nil) || !responseHasMore(map[string]any{"cursor": "next"}) || !responseHasMore(map[string]any{"pagination": map[string]any{"hasMore": true}}) {
+		t.Fatal("responseHasMore shape mismatch")
+	}
+	if err := verifyRecordCells(map[string]any{"recordId": "r"}, map[string]any{"f": 1}); err == nil {
+		t.Fatal("missing cells must fail")
+	}
+	if responseCursor(nil) != "" || responseCursor(map[string]any{"data": map[string]any{"next_cursor": " next "}}) != "next" ||
+		responseCursor(map[string]any{"result": map[string]any{"cursor": " legacy "}}) != "legacy" {
+		t.Fatal("responseCursor shape mismatch")
+	}
+}

internal/shortcut/aitable/record_query_all.go

@@ -0,0 +1,78 @@
+// Copyright 2026 Alibaba Group
+// SPDX-License-Identifier: Apache-2.0
+
+package aitable
+
+import (
+	"fmt"
+	"strings"
+
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/shortcut"
+)
+
+func queryAllRecords(rt *shortcut.RuntimeContext, params map[string]any, maxRecords int) ([]map[string]any, error) {
+	all := make([]map[string]any, 0)
+	cursor := ""
+	seen := map[string]bool{}
+	for page := 0; ; page++ {
+		request := cloneAnyMap(params)
+		request["limit"] = recordBatchSize
+		if cursor != "" {
+			request["cursor"] = cursor
+		}
+		data, err := rt.CallMCPData(serverMain, "query_records", request)
+		if err != nil {
+			return nil, err
+		}
+		records, found := findRecords(data)
+		if !found {
+			return nil, fmt.Errorf("query_records page %d is missing the records collection", page+1)
+		}
+		all = append(all, records...)
+		if maxRecords > 0 && len(all) > maxRecords {
+			return nil, fmt.Errorf("query matched more than the allowed %d records", maxRecords)
+		}
+		if !responseHasMore(data) {
+			return all, nil
+		}
+		next := responseCursor(data)
+		if next == "" {
+			return nil, fmt.Errorf("query_records page %d reports more data but no next cursor", page+1)
+		}
+		if seen[next] || next == cursor {
+			return nil, fmt.Errorf("query_records cursor cycle detected at %q", next)
+		}
+		seen[next] = true
+		cursor = next
+		if page >= 9999 {
+			return nil, fmt.Errorf("query_records exceeded the 10000-page safety bound")
+		}
+	}
+}
+
+func responseCursor(data map[string]any) string {
+	if data == nil {
+		return ""
+	}
+	for _, key := range []string{"nextCursor", "next_cursor", "cursor"} {
+		if value, ok := data[key].(string); ok && strings.TrimSpace(value) != "" {
+			return strings.TrimSpace(value)
+		}
+	}
+	for _, key := range []string{"data", "result", "pagination", "page"} {
+		if nested, ok := data[key].(map[string]any); ok {
+			if cursor := responseCursor(nested); cursor != "" {
+				return cursor
+			}
+		}
+	}
+	return ""
+}
+
+func cloneAnyMap(input map[string]any) map[string]any {
+	out := make(map[string]any, len(input))
+	for key, value := range input {
+		out[key] = value
+	}
+	return out
+}

internal/shortcut/aitable/record_upsert_by_key.go

@@ -0,0 +1,343 @@
+// Copyright 2026 Alibaba Group
+// SPDX-License-Identifier: Apache-2.0
+
+package aitable
+
+import (
+	"encoding/json"
+	"fmt"
+	"io"
+	"reflect"
+	"sort"
+	"strings"
+
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/corecmd"
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/corecmd/contract"
+	apperrors "github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/errors"
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/shortcut"
+)
+
+// RecordUpsertByKey creates or updates exactly one record selected by a unique
+// field value. It never writes when the key is ambiguous and always verifies
+// the final record through a read-back query.
+var RecordUpsertByKey = shortcut.Shortcut{
+	Service:     "aitable",
+	Command:     "+record-upsert-by-key",
+	Product:     serverMain,
+	Description: "按唯一字段值有则更新、无则创建记录，并读回验证",
+	Intent:      "已知一个应当唯一的 fieldId/value、希望幂等同步一条记录时使用；先完整查询键值，0 条创建、1 条更新、2 条以上停止，写后再次查询并验证所有传入 cells。",
+	Risk:        shortcut.RiskWrite,
+	Safety: contract.SafetySpec{
+		Effect: "write", Risk: "medium",
+		Confirmation: "user_required", Idempotency: "idempotent",
+	},
+	Contract: corecmd.ContractDecl{
+		Identity: contract.ToolIdentitySpec{
+			ProductID:      "aitable",
+			Name:           "shortcut_record_upsert_by_key",
+			CanonicalPath:  "aitable.shortcut_record_upsert_by_key",
+			CLIPath:        "aitable +record-upsert-by-key",
+			PrimaryCLIPath: "aitable +record-upsert-by-key",
+		},
+		Description: "按唯一字段值有则更新、无则创建记录，并读回验证。",
+		Interface: &contract.InterfaceSpec{
+			Mode:         contract.InterfaceModeComposite,
+			Availability: contract.InterfaceAvailable,
+			Reason:       "The command composes query_records with create_records/update_records and a final read-back; no single RPC owns its uniqueness and verification contract.",
+		},
+		Selection: contract.SelectionSpec{
+			AgentSummary: "按唯一字段值有则更新、无则创建记录，并读回验证",
+			UseWhen:      []string{"已知一个应当唯一的 fieldId/value、希望幂等同步一条记录时使用；先完整查询键值，0 条创建、1 条更新、2 条以上停止，写后再次查询并验证所有传入 cells。"},
+			AvoidWhen:    []string{"已经知道 recordId 时用 +record-update；一次处理多条不同键值时使用批量导入或分批调用"},
+			Examples: []string{
+				"dws aitable +record-upsert-by-key --base-id B --table-id T --key-field-id fldKey --key-value TASK-001 --cells '{\"fldStatus\":\"进行中\"}'",
+			},
+			ExampleDispositions: []contract.ExampleDisposition{{
+				Index:      recordUpsertExampleIndex(),
+				Mode:       contract.ExampleDispositionModeContractOnly,
+				ReasonCode: contract.ExampleDispositionReasonStatefulPreflight,
+				Reason:     "dry-run must query the live table to prove whether the unique key matches zero or one record; the isolated Agent example runner has no remote AITable fixture",
+				Reviewed:   true,
+			}},
+		},
+		DryRun: &contract.DryRunSpec{PreviewKind: "plan", RemoteReads: true},
+	},
+	Flags: []shortcut.Flag{
+		{Name: "base-id", Type: shortcut.FlagString, Desc: "Base ID", Required: true},
+		{Name: "table-id", Type: shortcut.FlagString, Desc: "Table ID", Required: true},
+		{Name: "key-field-id", Type: shortcut.FlagString, Desc: "具有唯一语义的字段 ID", Required: true},
+		{Name: "key-value", Type: shortcut.FlagString, Desc: "字符串键值；与 --key-value-json 二选一"},
+		{Name: "key-value-json", Type: shortcut.FlagString, Desc: "JSON 类型键值；与 --key-value 二选一"},
+		{Name: "cells", Type: shortcut.FlagString, Desc: "要写入的 cells JSON 对象", Required: true},
+	},
+	Constraints: []shortcut.Constraint{
+		{Kind: shortcut.ConstraintExactlyOne, Flags: []string{"key-value", "key-value-json"}, Description: "必须且只能提供一种键值表示"},
+	},
+	Tips: []string{
+		`dws aitable +record-upsert-by-key --base-id B --table-id T --key-field-id fldKey --key-value TASK-001 --cells '{"fldStatus":"进行中"}'`,
+	},
+	Execute: executeRecordUpsertByKey,
+}
+
+func recordUpsertExampleIndex() *int {
+	index := 0
+	return &index
+}
+
+func executeRecordUpsertByKey(rt *shortcut.RuntimeContext) error {
+	keyValue, err := recordKeyValue(rt)
+	if err != nil {
+		return err
+	}
+	cells, err := parseJSONObject("cells", rt.Str("cells"))
+	if err != nil {
+		return err
+	}
+	keyFieldID := rt.Str("key-field-id")
+	if existing, present := cells[keyFieldID]; present && !reflect.DeepEqual(existing, keyValue) {
+		return apperrors.NewValidation("--cells 中的键字段值与 --key-value/--key-value-json 冲突",
+			apperrors.WithReason("key_value_conflict"),
+			apperrors.WithExecutionStarted(false),
+		)
+	}
+	cells[keyFieldID] = keyValue
+
+	baseID, tableID := rt.Str("base-id"), rt.Str("table-id")
+	preflight, err := queryUniqueRecordByKey(rt, baseID, tableID, keyFieldID, keyValue)
+	if err != nil {
+		return err
+	}
+	action, tool := "create", "create_records"
+	writeRecord := map[string]any{"cells": cells}
+	if preflight != nil {
+		action, tool = "update", "update_records"
+		writeRecord["recordId"] = recordID(preflight)
+	}
+	params := map[string]any{
+		"baseId": baseID, "tableId": tableID,
+		"records": []any{writeRecord},
+	}
+	result := newCompositeResult("record_upsert_by_key")
+	result.Resolved = map[string]any{
+		"baseId": baseID, "tableId": tableID, "keyFieldId": keyFieldID,
+		"keyValue": keyValue, "matchedCount": boolCount(preflight != nil), "action": action,
+	}
+	result.RequestedCount = 1
+	result.Plan = []compositeStep{{Index: 1, Name: action + " record", Tool: tool, Status: "planned", Count: 1, Arguments: params}}
+	if rt.DryRun() {
+		result.Status = "planned"
+		result.Executed = false
+		return rt.Output(result)
+	}
+
+	writeData, writeErr := rt.CallMCPWriteDataStrict(serverMain, tool, params)
+	writeStep := compositeStep{Index: 1, Name: action + " record", Tool: tool, Status: "completed", Count: 1, Result: writeData}
+	if writeErr != nil {
+		writeStep.Status = "unknown"
+		writeStep.Error = writeErr.Error()
+	}
+	result.CompletedSteps = append(result.CompletedSteps, writeStep)
+
+	verified, verifyErr := queryUniqueRecordByKey(rt, baseID, tableID, keyFieldID, keyValue)
+	if verifyErr == nil && verified != nil {
+		verifyErr = verifyRecordCells(verified, cells)
+	}
+	if verifyErr != nil || verified == nil {
+		result.Status = "unknown"
+		result.FailedCount = 1
+		// An update targets a known record ID and is safe to retry. A create whose
+		// effect could not be read back must be resolved by the unique-key query
+		// first; an immediate blind retry could duplicate the row under eventual
+		// consistency.
+		retryable := action == "update"
+		result.Retryable = retryable
+		if verifyErr == nil {
+			verifyErr = fmt.Errorf("read-back found no record for the unique key")
+		}
+		result.Verification = map[string]any{"status": "failed", "error": verifyErr.Error()}
+		if writeErr != nil {
+			result.Warnings = append(result.Warnings, "write call also returned an error: "+writeErr.Error())
+		}
+		result.Checkpoint = map[string]any{
+			"nextStep":   "query the unique key again and verify its cells before retrying",
+			"keyFieldId": keyFieldID,
+			"keyValue":   keyValue,
+		}
+		return compositeError(result, verifyErr, retryable)
+	}
+
+	verifiedID := recordID(verified)
+	result.CompletedCount = 1
+	result.KnownEffects = append(result.KnownEffects, map[string]any{"action": action, "recordId": verifiedID, "keyFieldId": keyFieldID, "keyValue": keyValue})
+	result.Verification = map[string]any{
+		"status": "verified", "recordId": verifiedID,
+		"checkedFields": sortedMapKeys(cells),
+	}
+	result.Result = map[string]any{"action": action, "recordId": verifiedID, "record": verified}
+	if writeErr != nil {
+		result.Warnings = append(result.Warnings, "write response was an error, but the requested final state was proven by read-back")
+		result.CompletedSteps[0].Status = "recovered"
+	}
+	return rt.Output(result)
+}
+
+func recordKeyValue(rt *shortcut.RuntimeContext) (any, error) {
+	if rt.Changed("key-value") {
+		// The command framework rejects an explicitly empty string before Execute.
+		return rt.Str("key-value"), nil
+	}
+	decoder := json.NewDecoder(strings.NewReader(rt.Str("key-value-json")))
+	decoder.UseNumber()
+	var value any
+	if err := decoder.Decode(&value); err != nil {
+		return nil, apperrors.NewValidation("--key-value-json 不是合法 JSON: " + err.Error())
+	}
+	var trailing any
+	if err := decoder.Decode(&trailing); err != io.EOF {
+		return nil, apperrors.NewValidation("--key-value-json 必须只包含一个 JSON 标量")
+	}
+	if value == nil || reflect.ValueOf(value).Kind() == reflect.Map || reflect.ValueOf(value).Kind() == reflect.Slice {
+		return nil, apperrors.NewValidation("--key-value-json 只接受 string/number/bool 标量")
+	}
+	return value, nil
+}
+
+func queryUniqueRecordByKey(rt *shortcut.RuntimeContext, baseID, tableID, fieldID string, value any) (map[string]any, error) {
+	filters := map[string]any{
+		"operator": "and",
+		"operands": []any{map[string]any{
+			"operator": "eq", "operands": []any{fieldID, value},
+		}},
+	}
+	data, err := rt.CallMCPData(serverMain, "query_records", map[string]any{
+		"baseId": baseID, "tableId": tableID, "filters": filters, "limit": 2,
+	})
+	if err != nil {
+		return nil, err
+	}
+	records, found := findRecords(data)
+	if !found {
+		return nil, apperrors.NewAPI("query_records response is missing the records collection",
+			apperrors.WithOperation("aitable/query_records"),
+			apperrors.WithReason("target_invalid_response"),
+			apperrors.WithFailureStage("response_validation"),
+			apperrors.WithExecutionStarted(false),
+		)
+	}
+	if responseHasMore(data) {
+		return nil, apperrors.NewAPI("unique-key query is incomplete and cannot prove uniqueness",
+			apperrors.WithOperation("aitable/query_records"),
+			apperrors.WithReason("target_incomplete"),
+			apperrors.WithFailureStage("target_resolution"),
+			apperrors.WithExecutionStarted(false),
+			apperrors.WithDetails(map[string]any{"records": records}),
+		)
+	}
+	switch len(records) {
+	case 0:
+		return nil, nil
+	case 1:
+		if recordID(records[0]) == "" {
+			return nil, apperrors.NewAPI("query_records returned a record without recordId",
+				apperrors.WithOperation("aitable/query_records"),
+				apperrors.WithReason("target_invalid_response"),
+				apperrors.WithExecutionStarted(false),
+			)
+		}
+		return records[0], nil
+	default:
+		return nil, apperrors.NewValidation("唯一键匹配到多条记录，已在写入前停止",
+			apperrors.WithReason("target_ambiguous"),
+			apperrors.WithExecutionStarted(false),
+			apperrors.WithDetails(map[string]any{"records": records}),
+		)
+	}
+}
+
+func findRecords(data map[string]any) ([]map[string]any, bool) {
+	if data == nil {
+		return nil, false
+	}
+	if raw, exists := data["records"]; exists {
+		list, ok := raw.([]any)
+		if !ok {
+			return nil, false
+		}
+		out := make([]map[string]any, 0, len(list))
+		for _, item := range list {
+			record, ok := item.(map[string]any)
+			if !ok {
+				return nil, false
+			}
+			out = append(out, record)
+		}
+		return out, true
+	}
+	for _, key := range []string{"data", "result"} {
+		if nested, ok := data[key].(map[string]any); ok {
+			if records, found := findRecords(nested); found {
+				return records, true
+			}
+		}
+	}
+	return nil, false
+}
+
+func responseHasMore(data map[string]any) bool {
+	if data == nil {
+		return false
+	}
+	if value, ok := data["hasMore"].(bool); ok && value {
+		return true
+	}
+	for _, key := range []string{"nextCursor", "cursor"} {
+		if value, ok := data[key].(string); ok && strings.TrimSpace(value) != "" {
+			return true
+		}
+	}
+	for _, key := range []string{"data", "result", "pagination", "page"} {
+		if nested, ok := data[key].(map[string]any); ok && responseHasMore(nested) {
+			return true
+		}
+	}
+	return false
+}
+
+func recordID(record map[string]any) string {
+	for _, key := range []string{"recordId", "record_id", "id"} {
+		if value, ok := record[key].(string); ok && strings.TrimSpace(value) != "" {
+			return strings.TrimSpace(value)
+		}
+	}
+	return ""
+}
+
+func verifyRecordCells(record map[string]any, expected map[string]any) error {
+	actual, ok := record["cells"].(map[string]any)
+	if !ok {
+		return fmt.Errorf("read-back record %q is missing cells", recordID(record))
+	}
+	for fieldID, want := range expected {
+		got, exists := actual[fieldID]
+		if !exists || !reflect.DeepEqual(got, want) {
+			return fmt.Errorf("read-back mismatch for field %s: got %#v, want %#v", fieldID, got, want)
+		}
+	}
+	return nil
+}
+
+func sortedMapKeys(values map[string]any) []string {
+	keys := make([]string, 0, len(values))
+	for key := range values {
+		keys = append(keys, key)
+	}
+	sort.Strings(keys)
+	return keys
+}
+
+func boolCount(value bool) int {
+	if value {
+		return 1
+	}
+	return 0
+}

internal/shortcut/aitable/record_upsert_by_key_e2e_test.go

@@ -0,0 +1,468 @@
+// Copyright 2026 Alibaba Group
+// SPDX-License-Identifier: Apache-2.0
+
+package aitable
+
+import (
+	"bytes"
+	"context"
+	"encoding/json"
+	"errors"
+	"fmt"
+	"strings"
+	"testing"
+
+	apperrors "github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/errors"
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/helpers"
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/shortcut"
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/pkg/edition"
+	"github.com/spf13/cobra"
+)
+
+type upsertByKeyStep struct {
+	text string
+	err  error
+}
+
+type upsertByKeyCaller struct {
+	steps  []upsertByKeyStep
+	calls  []aitableURLCall
+	dryRun bool
+	callFn func(index int, product, tool string, args map[string]any) (string, error)
+}
+
+func (c *upsertByKeyCaller) CallTool(_ context.Context, product, tool string, args map[string]any) (*edition.ToolResult, error) {
+	return c.call(product, tool, args)
+}
+
+func (c *upsertByKeyCaller) CallReadTool(_ context.Context, product, tool string, args map[string]any) (*edition.ToolResult, error) {
+	return c.call(product, tool, args)
+}
+
+func (c *upsertByKeyCaller) call(product, tool string, args map[string]any) (*edition.ToolResult, error) {
+	cloned := make(map[string]any, len(args))
+	for key, value := range args {
+		cloned[key] = value
+	}
+	c.calls = append(c.calls, aitableURLCall{product: product, tool: tool, args: cloned})
+	index := len(c.calls) - 1
+	if c.callFn != nil {
+		text, err := c.callFn(index, product, tool, args)
+		if err != nil {
+			return nil, err
+		}
+		return urlResult(text), nil
+	}
+	if index >= len(c.steps) {
+		return nil, errors.New("unexpected upsert-by-key call")
+	}
+	step := c.steps[index]
+	if step.err != nil {
+		return nil, step.err
+	}
+	return urlResult(step.text), nil
+}
+
+func (*upsertByKeyCaller) Format() string { return "json" }
+func (c *upsertByKeyCaller) DryRun() bool { return c.dryRun }
+func (*upsertByKeyCaller) Fields() string { return "" }
+func (*upsertByKeyCaller) JQ() string     { return "" }
+
+func runUpsertByKeyCLI(t *testing.T, caller *upsertByKeyCaller, extra ...string) (string, error) {
+	t.Helper()
+	helpers.InitDepsForTest(t, caller)
+	root := &cobra.Command{Use: "dws", SilenceErrors: true, SilenceUsage: true}
+	root.PersistentFlags().Bool("yes", false, "")
+	root.PersistentFlags().Bool("dry-run", false, "")
+	root.PersistentFlags().String("format", "json", "")
+	root.AddCommand(shortcut.Commands()...)
+	stdout := &bytes.Buffer{}
+	root.SetOut(stdout)
+	root.SetErr(&bytes.Buffer{})
+	args := []string{
+		"aitable", "+record-upsert-by-key",
+		"--base-id", "base", "--table-id", "table",
+		"--key-field-id", "fldKey", "--key-value", "TASK-1",
+		"--cells", `{"fldStatus":"进行中"}`,
+		"--yes",
+	}
+	args = append(args, extra...)
+	root.SetArgs(args)
+	err := root.Execute()
+	return stdout.String(), err
+}
+
+func TestCrossPlatformCoverageRecordUpsertByKeyCreateE2E(t *testing.T) {
+	caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+		{text: `{"data":{"records":[]}}`},
+		{text: `{"data":{"records":[{"recordId":"r1"}]}}`},
+		{text: `{"data":{"records":[{"recordId":"r1","cells":{"fldKey":"TASK-1","fldStatus":"进行中"}}]}}`},
+	}}
+	out, err := runUpsertByKeyCLI(t, caller)
+	if err != nil {
+		t.Fatalf("create upsert error = %v", err)
+	}
+	for _, want := range []string{`"status": "success"`, `"action": "create"`, `"recordId": "r1"`, `"status": "verified"`} {
+		if !bytes.Contains([]byte(out), []byte(want)) {
+			t.Fatalf("create output missing %s:\n%s", want, out)
+		}
+	}
+	if len(caller.calls) != 3 || caller.calls[1].tool != "create_records" || caller.calls[2].tool != "query_records" {
+		t.Fatalf("create call flow = %#v", caller.calls)
+	}
+}
+
+func TestCrossPlatformCoverageRecordUpsertByKeyUpdateE2E(t *testing.T) {
+	caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+		{text: `{"records":[{"recordId":"r1","cells":{"fldKey":"TASK-1","fldStatus":"旧"}}]}`},
+		{text: `{"updatedRecords":[{"recordId":"r1"}]}`},
+		{text: `{"records":[{"recordId":"r1","cells":{"fldKey":"TASK-1","fldStatus":"进行中"}}]}`},
+	}}
+	out, err := runUpsertByKeyCLI(t, caller)
+	if err != nil || !bytes.Contains([]byte(out), []byte(`"action": "update"`)) {
+		t.Fatalf("update upsert = output:%q err:%v", out, err)
+	}
+	if caller.calls[1].tool != "update_records" {
+		t.Fatalf("update tool = %s", caller.calls[1].tool)
+	}
+}
+
+func TestCrossPlatformCoverageRecordUpsertByKeyAmbiguousStopsBeforeWriteE2E(t *testing.T) {
+	caller := &upsertByKeyCaller{steps: []upsertByKeyStep{{text: `{"records":[
+		{"recordId":"r1","cells":{"fldKey":"TASK-1"}},
+		{"recordId":"r2","cells":{"fldKey":"TASK-1"}}
+	]}`}}}
+	out, err := runUpsertByKeyCLI(t, caller)
+	if err == nil || out != "" || len(caller.calls) != 1 {
+		t.Fatalf("ambiguous upsert = output:%q err:%v calls:%#v", out, err, caller.calls)
+	}
+	var typed *apperrors.Error
+	if !errors.As(err, &typed) || typed.Reason != "target_ambiguous" || typed.ExecutionStarted == nil || *typed.ExecutionStarted {
+		t.Fatalf("ambiguous error = %#v", err)
+	}
+}
+
+func TestCrossPlatformCoverageRecordUpsertByKeyDryRunReadsAndPlansE2E(t *testing.T) {
+	caller := &upsertByKeyCaller{dryRun: true, steps: []upsertByKeyStep{{text: `{"records":[]}`}}}
+	out, err := runUpsertByKeyCLI(t, caller, "--dry-run")
+	if err != nil {
+		t.Fatalf("dry-run upsert error = %v", err)
+	}
+	if len(caller.calls) != 1 || caller.calls[0].tool != "query_records" {
+		t.Fatalf("dry-run calls = %#v", caller.calls)
+	}
+	for _, want := range []string{`"status": "planned"`, `"executed": false`, `"tool": "create_records"`} {
+		if !bytes.Contains([]byte(out), []byte(want)) {
+			t.Fatalf("dry-run output missing %s: %s", want, out)
+		}
+	}
+}
+
+func TestCrossPlatformCoverageRecordUpsertByKeyWriteReplyNeedsReadBackE2E(t *testing.T) {
+	t.Run("empty write recovered by proven state", func(t *testing.T) {
+		caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+			{text: `{"records":[]}`},
+			{text: ""},
+			{text: `{"records":[{"recordId":"r1","cells":{"fldKey":"TASK-1","fldStatus":"进行中"}}]}`},
+		}}
+		out, err := runUpsertByKeyCLI(t, caller)
+		if err != nil || !bytes.Contains([]byte(out), []byte(`"status": "recovered"`)) || !bytes.Contains([]byte(out), []byte("proven by read-back")) {
+			t.Fatalf("recovered write = output:%q err:%v", out, err)
+		}
+	})
+
+	t.Run("empty write and absent state is unknown", func(t *testing.T) {
+		caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+			{text: `{"records":[]}`},
+			{text: ""},
+			{text: `{"records":[]}`},
+		}}
+		out, err := runUpsertByKeyCLI(t, caller)
+		if err == nil || out != "" {
+			t.Fatalf("unknown write = output:%q err:%v", out, err)
+		}
+		var typed *apperrors.Error
+		if !errors.As(err, &typed) || typed.Reason != "aitable_composite_unknown" || typed.Retryable {
+			t.Fatalf("unknown write error = %#v", err)
+		}
+		if typed.Details == nil || !strings.Contains(fmt.Sprint(typed.Details), "query the unique key again") {
+			t.Fatalf("unknown create is missing safe recovery checkpoint: %#v", typed.Details)
+		}
+	})
+
+	t.Run("unknown update remains retryable because record id is known", func(t *testing.T) {
+		caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+			{text: `{"records":[{"recordId":"r1","cells":{"fldKey":"TASK-1","fldStatus":"旧"}}]}`},
+			{text: ""},
+			{text: `{"records":[{"recordId":"r1","cells":{"fldKey":"TASK-1","fldStatus":"旧"}}]}`},
+		}}
+		out, err := runUpsertByKeyCLI(t, caller)
+		if err == nil || out != "" {
+			t.Fatalf("unknown update = output:%q err:%v", out, err)
+		}
+		var typed *apperrors.Error
+		if !errors.As(err, &typed) || typed.Reason != "aitable_composite_unknown" || !typed.Retryable {
+			t.Fatalf("unknown update error = %#v", err)
+		}
+	})
+}
+
+func TestCrossPlatformCoverageRecordUpsertByKeyVerificationMismatchFailsE2E(t *testing.T) {
+	caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+		{text: `{"records":[]}`},
+		{text: `{"createdRecords":[{"recordId":"r1"}]}`},
+		{text: `{"records":[{"recordId":"r1","cells":{"fldKey":"TASK-1","fldStatus":"错误值"}}]}`},
+	}}
+	out, err := runUpsertByKeyCLI(t, caller)
+	if err == nil || out != "" {
+		t.Fatalf("verification mismatch = output:%q err:%v", out, err)
+	}
+}
+
+func runRecordBatchCLI(t *testing.T, caller *upsertByKeyCaller, command string, records []map[string]any, extra ...string) (string, error) {
+	t.Helper()
+	helpers.InitDepsForTest(t, caller)
+	root := &cobra.Command{Use: "dws", SilenceErrors: true, SilenceUsage: true}
+	root.PersistentFlags().Bool("yes", false, "")
+	root.PersistentFlags().Bool("dry-run", false, "")
+	root.PersistentFlags().String("format", "json", "")
+	root.AddCommand(shortcut.Commands()...)
+	stdout := &bytes.Buffer{}
+	root.SetOut(stdout)
+	root.SetErr(&bytes.Buffer{})
+	raw, err := json.Marshal(records)
+	if err != nil {
+		t.Fatal(err)
+	}
+	args := []string{"aitable", command, "--base-id", "base", "--table-id", "table", "--records", string(raw), "--yes"}
+	args = append(args, extra...)
+	root.SetArgs(args)
+	err = root.Execute()
+	return stdout.String(), err
+}
+
+func updateFixtureRecords(start, count int, status string) []map[string]any {
+	records := make([]map[string]any, 0, count)
+	for index := start; index < start+count; index++ {
+		records = append(records, map[string]any{
+			"recordId": fmt.Sprintf("r%03d", index),
+			"cells":    map[string]any{"fldStatus": status},
+		})
+	}
+	return records
+}
+
+func recordListJSON(t *testing.T, records []map[string]any) string {
+	t.Helper()
+	items := make([]any, 0, len(records))
+	for _, record := range records {
+		items = append(items, record)
+	}
+	raw, err := json.Marshal(map[string]any{"data": map[string]any{"records": items}})
+	if err != nil {
+		t.Fatal(err)
+	}
+	return string(raw)
+}
+
+func TestCrossPlatformCoverageRecordUpdateAutoChunksAndVerifiesE2E(t *testing.T) {
+	records := updateFixtureRecords(0, 101, "完成")
+	caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+		{text: `{"updatedCount":100}`},
+		{text: recordListJSON(t, records[:100])},
+		{text: `{"updatedCount":1}`},
+		{text: recordListJSON(t, records[100:])},
+	}}
+	out, err := runRecordBatchCLI(t, caller, "+record-update", records)
+	if err != nil {
+		t.Fatalf("record update batches error = %v", err)
+	}
+	for _, want := range []string{`"processedCount": 101`, `"batchCount": 2`, `"verifiedCount": 101`} {
+		if !bytes.Contains([]byte(out), []byte(want)) {
+			t.Fatalf("batch output missing %s: %s", want, out)
+		}
+	}
+	if len(caller.calls) != 4 || caller.calls[0].tool != "update_records" || caller.calls[1].tool != "query_records" || caller.calls[2].tool != "update_records" {
+		t.Fatalf("batch call sequence = %#v", caller.calls)
+	}
+	firstBatch := caller.calls[0].args["records"].([]any)
+	secondBatch := caller.calls[2].args["records"].([]any)
+	if len(firstBatch) != 100 || len(secondBatch) != 1 {
+		t.Fatalf("batch sizes = %d/%d", len(firstBatch), len(secondBatch))
+	}
+}
+
+func TestCrossPlatformCoverageRecordUpdatePartialStopsWithCheckpointE2E(t *testing.T) {
+	records := updateFixtureRecords(0, 101, "完成")
+	caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+		{text: `{"updatedCount":100}`},
+		{text: recordListJSON(t, records[:100])},
+		{err: errors.New("connection reset after send")},
+		{text: `{"records":[]}`},
+	}}
+	out, err := runRecordBatchCLI(t, caller, "+record-update", records)
+	if err == nil || out != "" {
+		t.Fatalf("partial record update = output:%q err:%v", out, err)
+	}
+	var typed *apperrors.Error
+	if !errors.As(err, &typed) || typed.Reason != "aitable_composite_partial_success" {
+		t.Fatalf("partial error = %#v", err)
+	}
+	result := typed.Details["result"].(compositeResult)
+	if result.CompletedCount != 100 || result.Checkpoint["nextOffset"] != 100 {
+		t.Fatalf("partial result = %#v", result)
+	}
+}
+
+func TestCrossPlatformCoverageRecordUpdateEmptyWriteReplyRecoveredOnlyByReadBackE2E(t *testing.T) {
+	records := updateFixtureRecords(0, 1, "完成")
+	caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+		{text: ""},
+		{text: recordListJSON(t, records)},
+	}}
+	out, err := runRecordBatchCLI(t, caller, "+record-update", records)
+	if err != nil || !bytes.Contains([]byte(out), []byte(`"status": "recovered"`)) {
+		t.Fatalf("empty write reply recovery = output:%q err:%v", out, err)
+	}
+}
+
+func TestCrossPlatformCoverageRecordUpsertRequiresCreatedIDsAndReadBackE2E(t *testing.T) {
+	records := []map[string]any{
+		{"recordId": "existing", "cells": map[string]any{"fldStatus": "完成"}},
+		{"cells": map[string]any{"fldKey": "new"}},
+	}
+	caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+		{text: `{"data":{"createdRecords":[{"recordId":"created"}]}}`},
+		{text: recordListJSON(t, []map[string]any{records[0]})},
+		{text: recordListJSON(t, []map[string]any{{"recordId": "created", "cells": map[string]any{"fldKey": "new"}}})},
+	}}
+	out, err := runRecordBatchCLI(t, caller, "+record-upsert", records)
+	if err != nil || !bytes.Contains([]byte(out), []byte(`"verifiedCount": 2`)) {
+		t.Fatalf("mixed upsert = output:%q err:%v", out, err)
+	}
+
+	caller = &upsertByKeyCaller{steps: []upsertByKeyStep{{text: `{"success":true}`}}}
+	out, err = runRecordBatchCLI(t, caller, "+record-upsert", []map[string]any{{"cells": map[string]any{"fldKey": "new"}}})
+	if err == nil || out != "" {
+		t.Fatalf("upsert without created IDs = output:%q err:%v", out, err)
+	}
+	var typed *apperrors.Error
+	if !errors.As(err, &typed) || typed.Retryable {
+		t.Fatalf("upsert create with unknown effect must not be blindly retryable: %#v", err)
+	}
+}
+
+func TestCrossPlatformCoverageRecordBatchDryRunPlansWithoutWritesE2E(t *testing.T) {
+	records := updateFixtureRecords(0, 101, "完成")
+	caller := &upsertByKeyCaller{dryRun: true}
+	out, err := runRecordBatchCLI(t, caller, "+record-update", records, "--dry-run")
+	if err != nil || len(caller.calls) != 0 {
+		t.Fatalf("batch dry-run = output:%q err:%v calls:%#v", out, err, caller.calls)
+	}
+	if !bytes.Contains([]byte(out), []byte(`"status": "planned"`)) || !bytes.Contains([]byte(out), []byte(`"count": 100`)) || !bytes.Contains([]byte(out), []byte(`"count": 1`)) {
+		t.Fatalf("batch dry-run output = %s", out)
+	}
+}
+
+func runRecordDeleteCLI(t *testing.T, caller *upsertByKeyCaller, ids []string, extra ...string) (string, error) {
+	t.Helper()
+	helpers.InitDepsForTest(t, caller)
+	root := &cobra.Command{Use: "dws", SilenceErrors: true, SilenceUsage: true}
+	root.PersistentFlags().Bool("yes", false, "")
+	root.PersistentFlags().Bool("dry-run", false, "")
+	root.PersistentFlags().String("format", "json", "")
+	root.AddCommand(shortcut.Commands()...)
+	stdout := &bytes.Buffer{}
+	root.SetOut(stdout)
+	root.SetErr(&bytes.Buffer{})
+	args := []string{"aitable", "+record-delete", "--base-id", "base", "--table-id", "table", "--record-ids", strings.Join(ids, ","), "--yes"}
+	args = append(args, extra...)
+	root.SetArgs(args)
+	err := root.Execute()
+	return stdout.String(), err
+}
+
+func recordIDFixtures(count int) []string {
+	ids := make([]string, 0, count)
+	for index := 0; index < count; index++ {
+		ids = append(ids, fmt.Sprintf("r%03d", index))
+	}
+	return ids
+}
+
+func TestCrossPlatformCoverageRecordDeleteAutoChunksAndProvesAbsenceE2E(t *testing.T) {
+	ids := recordIDFixtures(101)
+	caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+		{text: `{"deletedCount":100}`},
+		{text: `{"data":{"records":[]}}`},
+		{text: `{"deletedCount":1}`},
+		{text: `{"data":{"records":[]}}`},
+	}}
+	out, err := runRecordDeleteCLI(t, caller, ids)
+	if err != nil {
+		t.Fatalf("record delete batches error = %v", err)
+	}
+	for _, want := range []string{`"deletedCount": 101`, `"batchCount": 2`, `"verifiedCount": 101`, `"status": "verified_absent"`} {
+		if !strings.Contains(out, want) {
+			t.Fatalf("delete output missing %s: %s", want, out)
+		}
+	}
+	if len(caller.calls) != 4 || caller.calls[0].tool != "delete_records" || caller.calls[1].tool != "query_records" {
+		t.Fatalf("delete call sequence = %#v", caller.calls)
+	}
+}
+
+func TestCrossPlatformCoverageRecordDeleteEmptyReplyRecoveredOnlyByAbsenceE2E(t *testing.T) {
+	t.Run("legal empty collection proves deletion", func(t *testing.T) {
+		caller := &upsertByKeyCaller{steps: []upsertByKeyStep{{text: ""}, {text: `{"records":[]}`}}}
+		out, err := runRecordDeleteCLI(t, caller, []string{"r1"})
+		if err != nil || !strings.Contains(out, `"status": "recovered"`) {
+			t.Fatalf("delete recovered = output:%q err:%v", out, err)
+		}
+	})
+
+	t.Run("missing records contract is unknown", func(t *testing.T) {
+		caller := &upsertByKeyCaller{steps: []upsertByKeyStep{{text: `{"deletedCount":1}`}, {text: `{"data":{}}`}}}
+		out, err := runRecordDeleteCLI(t, caller, []string{"r1"})
+		if err == nil || out != "" {
+			t.Fatalf("delete missing records = output:%q err:%v", out, err)
+		}
+		var typed *apperrors.Error
+		if !errors.As(err, &typed) || typed.Reason != "aitable_composite_unknown" {
+			t.Fatalf("delete missing records error = %#v", err)
+		}
+	})
+
+	t.Run("remaining record is not success", func(t *testing.T) {
+		caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+			{text: `{"deletedCount":0}`},
+			{text: `{"records":[{"recordId":"r1","cells":{"fld":"still here"}}]}`},
+		}}
+		out, err := runRecordDeleteCLI(t, caller, []string{"r1"})
+		if err == nil || out != "" {
+			t.Fatalf("delete remaining record = output:%q err:%v", out, err)
+		}
+	})
+}
+
+func TestCrossPlatformCoverageRecordDeletePartialCheckpointE2E(t *testing.T) {
+	ids := recordIDFixtures(101)
+	caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+		{text: `{"deletedCount":100}`},
+		{text: `{"records":[]}`},
+		{text: `{"deletedCount":0}`},
+		{text: `{"records":[{"recordId":"r100","cells":{}}]}`},
+	}}
+	out, err := runRecordDeleteCLI(t, caller, ids)
+	if err == nil || out != "" {
+		t.Fatalf("partial delete = output:%q err:%v", out, err)
+	}
+	var typed *apperrors.Error
+	if !errors.As(err, &typed) || typed.Reason != "aitable_composite_partial_success" {
+		t.Fatalf("partial delete error = %#v", err)
+	}
+	result := typed.Details["result"].(compositeResult)
+	if result.Checkpoint["nextOffset"] != 100 || result.CompletedCount != 100 {
+		t.Fatalf("partial delete result = %#v", result)
+	}
+}

internal/shortcut/aitable/remaining_coverage_gap_test.go

@@ -0,0 +1,367 @@
+// Copyright 2026 Alibaba Group
+// SPDX-License-Identifier: Apache-2.0
+
+package aitable
+
+import (
+	"errors"
+	"fmt"
+	"strings"
+	"testing"
+
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/shortcut"
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/pkg/edition"
+)
+
+func TestCrossPlatformCoverageAITableProjectionShapeFailures(t *testing.T) {
+	if _, err := resolveNamedList("op", nil, []string{"data"}, []string{"items"}); err == nil {
+		t.Fatal("nil list response must fail")
+	}
+	if _, err := resolveNamedList("op", map[string]any{"data": "bad"}, []string{"data"}, []string{"items"}); err == nil {
+		t.Fatal("non-object envelope must fail")
+	}
+	if _, err := resolveNamedList("op", map[string]any{"data": map[string]any{"other": []any{}}}, []string{"data"}, []string{"items"}); err == nil {
+		t.Fatal("unrecognized envelope must fail")
+	}
+	for name, project := range map[string]func() error{
+		"base": func() error {
+			_, err := baseListProject("list_bases", map[string]any{"bases": []any{"bad"}})
+			return err
+		},
+		"template": func() error {
+			_, err := templateSearchProject(map[string]any{"templates": []any{"bad"}})
+			return err
+		},
+		"view": func() error {
+			_, err := viewGetProject(map[string]any{"views": []any{"bad"}})
+			return err
+		},
+		"form": func() error {
+			_, err := formListProject(map[string]any{"views": []any{"bad"}})
+			return err
+		},
+		"workflow": func() error {
+			_, err := workflowListProject(map[string]any{"workflows": []any{"bad"}})
+			return err
+		},
+	} {
+		t.Run(name, func(t *testing.T) {
+			if err := project(); err == nil {
+				t.Fatal("non-object projection item must fail")
+			}
+		})
+	}
+}
+
+func TestCrossPlatformCoverageAITableProjectionValidAndMissingIdentityShapes(t *testing.T) {
+	bases, err := baseListProject("list_bases", map[string]any{"bases": []any{map[string]any{"baseId": "b", "baseName": "Base"}}})
+	if err != nil || len(bases) != 1 {
+		t.Fatalf("base projection = %#v, %v", bases, err)
+	}
+	if _, err := templateSearchProject(map[string]any{"templates": []any{map[string]any{"name": "missing"}}}); err == nil {
+		t.Fatal("template missing ID succeeded")
+	}
+	templates, err := templateSearchProject(map[string]any{"templates": []any{map[string]any{"templateId": "t"}}})
+	if err != nil || len(templates) != 1 {
+		t.Fatalf("template projection = %#v, %v", templates, err)
+	}
+	views, err := viewGetProject(map[string]any{"views": []any{map[string]any{"viewId": "v"}}})
+	if err != nil || len(views) != 1 {
+		t.Fatalf("view projection = %#v, %v", views, err)
+	}
+	forms, err := formListProject(map[string]any{"views": []any{map[string]any{"viewId": "f"}}})
+	if err != nil || len(forms) != 1 {
+		t.Fatalf("form projection = %#v, %v", forms, err)
+	}
+	workflows, err := workflowListProject(map[string]any{"workflows": []any{map[string]any{"workflowId": "w"}}})
+	if err != nil || len(workflows) != 1 {
+		t.Fatalf("workflow projection = %#v, %v", workflows, err)
+	}
+}
+
+func TestCrossPlatformCoverageCompositeAndReviewedContractFallbacks(t *testing.T) {
+	err := compositeError(newCompositeResult("test"), errors.New("cause"), true)
+	if err == nil || !strings.Contains(err.Error(), "unknown") {
+		t.Fatalf("composite default status = %v", err)
+	}
+	unreviewed := shortcut.Shortcut{Command: "+not-reviewed"}
+	if got := withReviewedAITableShortcutContracts(unreviewed); len(got) != 1 || !got[0].Contract.Empty() {
+		t.Fatalf("unreviewed shortcut changed: %#v", got)
+	}
+	value := shortcut.Shortcut{
+		Command:     "+base-list",
+		Description: "fallback description",
+		Tips:        []string{"one", "two", "three"},
+	}
+	contract := reviewedAITableShortcutContract(value)
+	if contract.Description != "fallback description" || len(contract.Selection.Examples) != 2 {
+		t.Fatalf("reviewed contract fallback = %#v", contract)
+	}
+}
+
+func TestCrossPlatformCoverageURLResolveAllVerificationKindsE2E(t *testing.T) {
+	cases := []struct {
+		name string
+		url  string
+		tool string
+		text string
+	}{
+		{name: "record", url: "https://alidocs.dingtalk.com/i/nodes/base?iframeQuery=sheetId%3Dtable%26recordId%3Drecord", tool: "query_records", text: `{"records":[{"record_id":"record"}]}`},
+		{name: "view", url: "https://alidocs.dingtalk.com/i/nodes/base?iframeQuery=sheetId%3Dtable%26viewId%3Dview", tool: "get_views", text: `{"views":[{"view_id":"view"}]}`},
+		{name: "base by id", url: "https://alidocs.dingtalk.com/i/nodes/base", tool: "get_base", text: `{"base_id":"base"}`},
+		{name: "base structural fallback", url: "https://alidocs.dingtalk.com/i/nodes/base", tool: "get_base", text: `{"data":{"tables":[]}}`},
+	}
+	for _, tc := range cases {
+		t.Run(tc.name, func(t *testing.T) {
+			caller := &urlResolveE2ECaller{results: []*edition.ToolResult{urlResult(tc.text)}}
+			out, err := runURLResolveCLI(t, caller, "--url", tc.url, "--verify")
+			if err != nil || !strings.Contains(out, `"verified": true`) || caller.calls[0].tool != tc.tool {
+				t.Fatalf("verification = output:%q err:%v calls:%#v", out, err, caller.calls)
+			}
+		})
+	}
+
+	t.Run("tool error", func(t *testing.T) {
+		caller := &urlResolveE2ECaller{errors: []error{errors.New("verification failed")}}
+		out, err := runURLResolveCLI(t, caller, "--url", "https://alidocs.dingtalk.com/i/nodes/base", "--verify")
+		if err == nil || out != "" {
+			t.Fatalf("tool error = output:%q err:%v", out, err)
+		}
+	})
+
+	t.Run("base unknown response", func(t *testing.T) {
+		caller := &urlResolveE2ECaller{results: []*edition.ToolResult{urlResult(`{"success":true}`)}}
+		out, err := runURLResolveCLI(t, caller, "--url", "https://alidocs.dingtalk.com/i/nodes/base", "--verify")
+		if err == nil || out != "" {
+			t.Fatalf("unknown base = output:%q err:%v", out, err)
+		}
+	})
+}
+
+func TestCrossPlatformCoverageURLResolveRecursiveShapeHelpers(t *testing.T) {
+	value := map[string]any{"data": []any{map[string]any{"id": "wanted"}}}
+	if !responseContainsID(value, "wanted", "id") || responseContainsID(value, "other", "id") {
+		t.Fatal("responseContainsID recursion mismatch")
+	}
+	if !responseHasAnyKey(map[string]any{"data": []any{map[string]any{"tables": []any{}}}}, "tables") {
+		t.Fatal("responseHasAnyKey recursion mismatch")
+	}
+	if responseHasAnyKey([]any{"value"}, "tables") {
+		t.Fatal("unrelated key must not match")
+	}
+}
+
+func TestCrossPlatformCoverageViewPresetValidationAndDryRunE2E(t *testing.T) {
+	cases := []struct {
+		name   string
+		config string
+		steps  []upsertByKeyStep
+	}{
+		{name: "invalid config", config: `{`},
+		{name: "empty config", config: `{}`},
+		{name: "preflight error", config: `{"f":1}`, steps: []upsertByKeyStep{{err: errors.New("views failed")}}},
+		{name: "preflight missing list", config: `{"f":1}`, steps: []upsertByKeyStep{{text: `{}`}}},
+		{name: "matched missing id", config: `{"f":1}`, steps: []upsertByKeyStep{{text: `{"views":[{"viewName":"X","viewType":"Grid"}]}`}}},
+		{name: "matched type conflict", config: `{"f":1}`, steps: []upsertByKeyStep{{text: `{"views":[{"viewId":"v","viewName":"X","viewType":"Kanban"}]}`}}},
+	}
+	for _, tc := range cases {
+		t.Run(tc.name, func(t *testing.T) {
+			out, err := runAITableCompositeCLI(t, &upsertByKeyCaller{steps: tc.steps}, "+view-preset-apply",
+				"--base-id", "base", "--table-id", "table", "--name", "X", "--view-type", "Grid", "--config", tc.config, "--yes")
+			if err == nil || out != "" {
+				t.Fatalf("validation = output:%q err:%v", out, err)
+			}
+		})
+	}
+
+	caller := &upsertByKeyCaller{dryRun: true, steps: []upsertByKeyStep{{text: `{"views":[]}`}}}
+	out, err := runAITableCompositeCLI(t, caller, "+view-preset-apply",
+		"--base-id", "base", "--table-id", "table", "--name", "X", "--view-type", "Grid", "--config", `{"f":1}`, "--dry-run", "--yes")
+	if err != nil || !strings.Contains(out, `"status": "planned"`) {
+		t.Fatalf("dry run = output:%q err:%v", out, err)
+	}
+}
+
+func TestCrossPlatformCoverageViewPresetReadBackFailuresAndUpdateE2E(t *testing.T) {
+	preflight := `{"views":[{"viewId":"v","viewName":"X","viewType":"Grid","config":{"f":0}}]}`
+	cases := []struct {
+		name     string
+		write    upsertByKeyStep
+		readBack upsertByKeyStep
+	}{
+		{name: "read error", write: upsertByKeyStep{text: `{"updated":1}`}, readBack: upsertByKeyStep{err: errors.New("read failed")}},
+		{name: "missing list", write: upsertByKeyStep{text: `{"updated":1}`}, readBack: upsertByKeyStep{text: `{}`}},
+		{name: "missing exact name", write: upsertByKeyStep{text: `{"updated":1}`}, readBack: upsertByKeyStep{text: `{"views":[]}`}},
+		{name: "identity mismatch", write: upsertByKeyStep{text: `{"updated":1}`}, readBack: upsertByKeyStep{text: `{"views":[{"viewId":"other","viewName":"X","viewType":"Grid","config":{"f":1}}]}`}},
+		{name: "dual failure", write: upsertByKeyStep{err: errors.New("write failed")}, readBack: upsertByKeyStep{err: errors.New("read failed")}},
+	}
+	for _, tc := range cases {
+		t.Run(tc.name, func(t *testing.T) {
+			caller := &upsertByKeyCaller{steps: []upsertByKeyStep{{text: preflight}, tc.write, tc.readBack}}
+			out, err := runAITableCompositeCLI(t, caller, "+view-preset-apply",
+				"--base-id", "base", "--table-id", "table", "--name", "X", "--view-type", "Grid", "--config", `{"f":1}`, "--yes")
+			if err == nil || out != "" {
+				t.Fatalf("read-back failure = output:%q err:%v", out, err)
+			}
+		})
+	}
+
+	caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+		{text: preflight}, {text: `{"updated":1}`}, {text: `{"views":[{"viewId":"v","viewName":"X","viewType":"Grid","config":{"f":1}}]}`},
+	}}
+	out, err := runAITableCompositeCLI(t, caller, "+view-preset-apply",
+		"--base-id", "base", "--table-id", "table", "--name", "X", "--view-type", "Grid", "--config", `{"f":1}`, "--yes")
+	if err != nil || !strings.Contains(out, `"action": "update"`) {
+		t.Fatalf("update success = output:%q err:%v", out, err)
+	}
+}
+
+func TestCrossPlatformCoverageViewPresetPureMatching(t *testing.T) {
+	if presetViewMatches(map[string]any{"viewType": "Kanban", "config": map[string]any{}}, "Grid", map[string]any{}) {
+		t.Fatal("type mismatch must fail")
+	}
+	if mapContains(map[string]any{}, map[string]any{"missing": 1}) {
+		t.Fatal("missing key must fail")
+	}
+	if mapContains(map[string]any{"nested": 1}, map[string]any{"nested": map[string]any{"x": 1}}) {
+		t.Fatal("nested type mismatch must fail")
+	}
+	if mapContains(map[string]any{"nested": map[string]any{}}, map[string]any{"nested": map[string]any{"x": 1}}) {
+		t.Fatal("nested value mismatch must fail")
+	}
+	if !mapContains(map[string]any{"nested": map[string]any{"x": 1, "extra": true}}, map[string]any{"nested": map[string]any{"x": 1}}) {
+		t.Fatal("nested subset must match")
+	}
+	if mapContains(map[string]any{"x": 1}, map[string]any{"x": 2}) {
+		t.Fatal("scalar mismatch must fail")
+	}
+}
+
+func TestCrossPlatformCoverageWorkflowDeployValidationAndFailureStagesE2E(t *testing.T) {
+	cases := []struct {
+		name  string
+		dsl   string
+		extra []string
+		steps []upsertByKeyStep
+	}{
+		{name: "invalid DSL", dsl: `{`},
+		{name: "wrong DSL version", dsl: `{"version":"v0"}`},
+		{name: "update preflight error", dsl: workflowDSLFixture, extra: []string{"--workflow-id", "w"}, steps: []upsertByKeyStep{{err: errors.New("preflight failed")}}},
+		{name: "update target not found", dsl: workflowDSLFixture, extra: []string{"--workflow-id", "w"}, steps: []upsertByKeyStep{{text: `{"flowId":"other"}`}}},
+		{name: "publish missing ID", dsl: workflowDSLFixture, steps: []upsertByKeyStep{{text: `{"valid":true}`}}},
+		{name: "publish ID wrong type", dsl: workflowDSLFixture, steps: []upsertByKeyStep{{text: `{"valid":true,"flowId":1}`}}},
+		{name: "read-back error", dsl: workflowDSLFixture, steps: []upsertByKeyStep{{text: `{"valid":true,"flowId":"w"}`}, {err: errors.New("read failed")}}},
+		{name: "read-back wrong ID", dsl: workflowDSLFixture, steps: []upsertByKeyStep{{text: `{"valid":true,"flowId":"w"}`}, {text: `{"flowId":"other"}`}}},
+	}
+	for _, tc := range cases {
+		t.Run(tc.name, func(t *testing.T) {
+			args := []string{"--base-id", "base", "--dsl", tc.dsl}
+			args = append(args, tc.extra...)
+			args = append(args, "--yes")
+			out, err := runAITableCompositeCLI(t, &upsertByKeyCaller{steps: tc.steps}, "+workflow-deploy", args...)
+			if err == nil || out != "" {
+				t.Fatalf("workflow failure = output:%q err:%v", out, err)
+			}
+		})
+	}
+
+	caller := &upsertByKeyCaller{dryRun: true}
+	out, err := runAITableCompositeCLI(t, caller, "+workflow-deploy", "--base-id", "base", "--dsl", workflowDSLFixture, "--enable", "--dry-run", "--yes")
+	if err != nil || !strings.Contains(out, `"status": "planned"`) || !strings.Contains(out, "enable and verify") {
+		t.Fatalf("workflow dry run = output:%q err:%v", out, err)
+	}
+}
+
+func TestCrossPlatformCoverageWorkflowEnableErrorCanStillBeVerifiedE2E(t *testing.T) {
+	caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+		{text: `{"valid":true,"flowId":"w"}`}, {text: `{"flowId":"w"}`},
+		{err: errors.New("enable reply failed")}, {text: `{"list":[{"flowId":"w","enabled":true}]}`},
+	}}
+	out, err := runAITableCompositeCLI(t, caller, "+workflow-deploy", "--base-id", "base", "--dsl", workflowDSLFixture, "--enable", "--yes")
+	if err != nil || !strings.Contains(out, "enable response error") {
+		t.Fatalf("enable recovery = output:%q err:%v", out, err)
+	}
+}
+
+func TestCrossPlatformCoverageWorkflowListFailureShapesE2E(t *testing.T) {
+	for name, step := range map[string]upsertByKeyStep{
+		"list error":   {err: errors.New("list failed")},
+		"missing list": {text: `{}`},
+		"not present":  {text: `{"list":[]}`},
+	} {
+		t.Run(name, func(t *testing.T) {
+			caller := &upsertByKeyCaller{steps: []upsertByKeyStep{
+				{text: `{"valid":true,"flowId":"w"}`}, {text: `{"flowId":"w"}`}, {text: `{"enabled":true}`}, step,
+			}}
+			out, err := runAITableCompositeCLI(t, caller, "+workflow-deploy", "--base-id", "base", "--dsl", workflowDSLFixture, "--enable", "--yes")
+			if err == nil || out != "" {
+				t.Fatalf("list failure = output:%q err:%v", out, err)
+			}
+		})
+	}
+}
+
+func TestCrossPlatformCoverageWorkflowListPaginationAndBoundE2E(t *testing.T) {
+	page := make([]any, 100)
+	for index := range page {
+		page[index] = map[string]any{"flowId": fmt.Sprintf("other-%d", index)}
+	}
+	caller := &upsertByKeyCaller{callFn: func(index int, _, tool string, _ map[string]any) (string, error) {
+		switch index {
+		case 0:
+			return `{"valid":true,"flowId":"w"}`, nil
+		case 1:
+			return `{"flowId":"w"}`, nil
+		case 2:
+			return `{"enabled":true}`, nil
+		case 3:
+			return mustJSONText(t, map[string]any{"list": page}), nil
+		default:
+			return `{"list":[{"flowId":"w","status":"ACTIVE"}]}`, nil
+		}
+	}}
+	out, err := runAITableCompositeCLI(t, caller, "+workflow-deploy", "--base-id", "base", "--dsl", workflowDSLFixture, "--enable", "--yes")
+	if err != nil || !strings.Contains(out, `"running": true`) || len(caller.calls) != 5 {
+		t.Fatalf("paginated list = output:%q err:%v calls:%d", out, err, len(caller.calls))
+	}
+
+	fullPage := mustJSONText(t, map[string]any{"list": page})
+	caller = &upsertByKeyCaller{callFn: func(index int, _, _ string, _ map[string]any) (string, error) {
+		if index == 0 {
+			return `{"valid":true,"flowId":"w"}`, nil
+		}
+		if index == 1 {
+			return `{"flowId":"w"}`, nil
+		}
+		if index == 2 {
+			return `{"enabled":true}`, nil
+		}
+		return fullPage, nil
+	}}
+	out, err = runAITableCompositeCLI(t, caller, "+workflow-deploy", "--base-id", "base", "--dsl", workflowDSLFixture, "--enable", "--yes")
+	if err == nil || out != "" || len(caller.calls) != 103 {
+		t.Fatalf("workflow list bound = output:%q err:%v calls:%d", out, err, len(caller.calls))
+	}
+}
+
+func TestCrossPlatformCoverageWorkflowPureResponseHelpers(t *testing.T) {
+	if _, _, _, err := workflowPublishResult(nil); err == nil {
+		t.Fatal("nil publish result must fail")
+	}
+	if value, found := findBoolByKeys(map[string]any{"data": []any{map[string]any{"enabled": true}}}, "enabled"); !found || !value {
+		t.Fatal("nested bool not found")
+	}
+	if _, found := findBoolByKeys([]any{"value"}, "enabled"); found {
+		t.Fatal("unrelated bool unexpectedly found")
+	}
+	for _, workflow := range []map[string]any{
+		{"status": "enabled"}, {"state": "active"}, {"isEnabled": true},
+	} {
+		if !workflowIsRunning(workflow) {
+			t.Fatalf("workflow should be running: %#v", workflow)
+		}
+	}
+	if workflowIsRunning(map[string]any{"status": "STOP", "enabled": false}) {
+		t.Fatal("stopped workflow must not be running")
+	}
+}

internal/shortcut/aitable/schema_contracts.go

@@ -0,0 +1,147 @@
+// Copyright 2026 Alibaba Group
+// Licensed under the Apache License, Version 2.0
+
+package aitable
+
+import (
+	"strings"
+
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/corecmd"
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/corecmd/contract"
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/shortcut"
+)
+
+const reviewedAITableShortcutInterfaceReason = "Reviewed AI Table Shortcut adapter: the executable CLI owns validation, orchestration, output projection, and confirmation; no single pinned MCP interface represents the complete command contract."
+
+// reviewedAITableShortcutContractCommands is the exact migration ledger for
+// the 53 executable AI Table shortcuts that were hidden only because they
+// predated typed Contract declarations. New shortcuts must declare Safety and
+// Contract directly and must never be added to this migration ledger.
+var reviewedAITableShortcutContractCommands = map[string]struct{}{
+	"+advperm-disable":           {},
+	"+advperm-enable":            {},
+	"+attachment-upload":         {},
+	"+base-copy":                 {},
+	"+base-delete":               {},
+	"+base-get-primary-doc-id":   {},
+	"+base-update":               {},
+	"+chart-delete":              {},
+	"+chart-share-get":           {},
+	"+chart-share-update":        {},
+	"+chart-update":              {},
+	"+dashboard-arrange":         {},
+	"+dashboard-delete":          {},
+	"+dashboard-share-get":       {},
+	"+dashboard-share-update":    {},
+	"+dashboard-update":          {},
+	"+export-data":               {},
+	"+field-delete":              {},
+	"+field-update":              {},
+	"+form-delete":               {},
+	"+form-field-hide":           {},
+	"+form-field-update":         {},
+	"+form-share-update":         {},
+	"+form-update":               {},
+	"+import-data":               {},
+	"+import-upload":             {},
+	"+record-delete":             {},
+	"+record-primary-doc-create": {},
+	"+record-primary-doc-get":    {},
+	"+record-update":             {},
+	"+record-upsert":             {},
+	"+role-create":               {},
+	"+role-delete":               {},
+	"+role-get":                  {},
+	"+role-update":               {},
+	"+section-create":            {},
+	"+section-delete":            {},
+	"+section-move-node":         {},
+	"+section-rename":            {},
+	"+section-reorder":           {},
+	"+table-delete":              {},
+	"+table-update":              {},
+	"+view-delete":               {},
+	"+view-duplicate":            {},
+	"+view-lock":                 {},
+	"+view-set-fill-color-rule":  {},
+	"+view-set-frozen-cols":      {},
+	"+view-set-row-height":       {},
+	"+view-update":               {},
+	"+workflow-disable":          {},
+	"+workflow-enable":           {},
+	"+workflow-get":              {},
+	"+workflow-list":             {},
+}
+
+func withReviewedAITableShortcutContracts(values ...shortcut.Shortcut) []shortcut.Shortcut {
+	out := make([]shortcut.Shortcut, len(values))
+	for index, value := range values {
+		out[index] = value
+		if !value.Contract.Empty() {
+			continue
+		}
+		if _, reviewed := reviewedAITableShortcutContractCommands[value.Command]; !reviewed {
+			continue
+		}
+		out[index].Safety = reviewedAITableShortcutSafety(value.Risk)
+		out[index].Contract = reviewedAITableShortcutContract(value)
+	}
+	return out
+}
+
+func reviewedAITableShortcutSafety(risk shortcut.Risk) contract.SafetySpec {
+	switch risk {
+	case shortcut.RiskWrite:
+		return contract.SafetySpec{
+			Effect: "write", Risk: "medium",
+			Confirmation: "user_required", Idempotency: "unknown",
+		}
+	case shortcut.RiskHighWrite:
+		return contract.SafetySpec{
+			Effect: "destructive", Risk: "high",
+			Confirmation: "user_required", Idempotency: "unknown",
+		}
+	default:
+		return contract.SafetySpec{
+			Effect: "read", Risk: "low",
+			Confirmation: "not_required", Idempotency: "idempotent",
+		}
+	}
+}
+
+func reviewedAITableShortcutContract(value shortcut.Shortcut) corecmd.ContractDecl {
+	name := "shortcut_" + strings.ReplaceAll(strings.TrimPrefix(value.Command, "+"), "-", "_")
+	cliPath := "aitable " + value.Command
+	description := strings.TrimSpace(value.Intent)
+	if description == "" {
+		description = strings.TrimSpace(value.Description)
+	}
+	examples := append([]string(nil), value.Tips...)
+	if len(examples) > 2 {
+		examples = examples[:2]
+	}
+	return corecmd.ContractDecl{
+		Title:       value.Description,
+		Description: description,
+		Interface: &contract.InterfaceSpec{
+			Mode:         contract.InterfaceModeComposite,
+			Availability: contract.InterfaceAvailable,
+			Reason:       reviewedAITableShortcutInterfaceReason,
+		},
+		Selection: contract.SelectionSpec{
+			AgentSummary: value.Description,
+			UseWhen:      []string{description},
+			AvoidWhen: []string{
+				"只需未经适配的底层参数或原始响应时改用对应 aitable 原子命令；电子表格单元格操作改用 sheet",
+			},
+			Examples: examples,
+		},
+		Identity: contract.ToolIdentitySpec{
+			ProductID:      "aitable",
+			Name:           name,
+			CanonicalPath:  "aitable." + name,
+			CLIPath:        cliPath,
+			PrimaryCLIPath: cliPath,
+		},
+	}
+}

internal/shortcut/aitable/schema_contracts_test.go

@@ -0,0 +1,34 @@
+// Copyright 2026 Alibaba Group
+// SPDX-License-Identifier: Apache-2.0
+
+package aitable
+
+import (
+	"testing"
+
+	"github.com/DingTalk-Real-AI/dingtalk-workspace-cli/internal/shortcut"
+)
+
+func TestCrossPlatformCoverageAITableLegacyContractLedgerIsExact(t *testing.T) {
+	if got, want := len(reviewedAITableShortcutContractCommands), 53; got != want {
+		t.Fatalf("legacy contract migration ledger = %d, want %d", got, want)
+	}
+	for command := range reviewedAITableShortcutContractCommands {
+		found := false
+		for _, item := range shortcut.All() {
+			if item.Service == "aitable" && item.Command == command {
+				found = true
+				if item.Contract.Empty() {
+					t.Errorf("%s: migration did not deliver Contract", command)
+				}
+				if item.Safety.Confirmation == "" {
+					t.Errorf("%s: migration did not deliver explicit Safety", command)
+				}
+				break
+			}
+		}
+		if !found {
+			t.Errorf("stale migration ledger entry %s", command)
+		}
+	}
+}