fix(im): resolve direct recipients via contact search

internal/cli/schema_agent_metadata/chat.json

@@ -9836,14 +9836,14 @@
           "source": "internal/cli/schema_hints/selection/chat.json",
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Agent-authored from the verified identity/content capability matrix, userId resolution, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media.",
+          "review_reason": "Agent-authored from the verified identity/content capability matrix, live-compatible contact search with exact userId matching, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media.",
           "candidates": [
             {
               "value": "统一发送文本、Markdown、当前用户文件或已有 mediaId 图片",
               "source": "internal/cli/schema_hints/selection/chat.json",
               "precedence": "reviewed_explicit",
               "selected": true,
-              "review_reason": "Agent-authored from the verified identity/content capability matrix, userId resolution, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media."
+              "review_reason": "Agent-authored from the verified identity/content capability matrix, live-compatible contact search with exact userId matching, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media."
             }
           ]
         },
@@ -9852,14 +9852,14 @@
           "source": "internal/cli/schema_hints/metadata/chat.json",
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Reviewed against every executable identity/content/target validation branch and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
+          "review_reason": "Reviewed against every executable identity/content/target validation branch, authenticated contact-search response shape, and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
           "candidates": [
             {
               "value": "available",
               "source": "internal/cli/schema_hints/metadata/chat.json",
               "precedence": "reviewed_explicit",
               "selected": true,
-              "review_reason": "Reviewed against every executable identity/content/target validation branch and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists."
+              "review_reason": "Reviewed against every executable identity/content/target validation branch, authenticated contact-search response shape, and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists."
             }
           ]
         },
@@ -9870,7 +9870,7 @@
           "source": "internal/cli/schema_hints/selection/chat.json",
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Agent-authored from the verified identity/content capability matrix, userId resolution, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media.",
+          "review_reason": "Agent-authored from the verified identity/content capability matrix, live-compatible contact search with exact userId matching, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media.",
           "candidates": [
             {
               "value": [
@@ -9879,7 +9879,7 @@
               "source": "internal/cli/schema_hints/selection/chat.json",
               "precedence": "reviewed_explicit",
               "selected": true,
-              "review_reason": "Agent-authored from the verified identity/content capability matrix, userId resolution, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media."
+              "review_reason": "Agent-authored from the verified identity/content capability matrix, live-compatible contact search with exact userId matching, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media."
             }
           ]
         },
@@ -9888,14 +9888,14 @@
           "source": "internal/cli/schema_hints/metadata/chat.json",
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Reviewed against every executable identity/content/target validation branch and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
+          "review_reason": "Reviewed against every executable identity/content/target validation branch, authenticated contact-search response shape, and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
           "candidates": [
             {
               "value": "user_required",
               "source": "internal/cli/schema_hints/metadata/chat.json",
               "precedence": "reviewed_explicit",
               "selected": true,
-              "review_reason": "Reviewed against every executable identity/content/target validation branch and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists."
+              "review_reason": "Reviewed against every executable identity/content/target validation branch, authenticated contact-search response shape, and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists."
             },
             {
               "value": "not_required",
@@ -9910,14 +9910,14 @@
           "source": "internal/cli/schema_hints/metadata/chat.json",
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Reviewed against every executable identity/content/target validation branch and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
+          "review_reason": "Reviewed against every executable identity/content/target validation branch, authenticated contact-search response shape, and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
           "candidates": [
             {
               "value": "write",
               "source": "internal/cli/schema_hints/metadata/chat.json",
               "precedence": "reviewed_explicit",
               "selected": true,
-              "review_reason": "Reviewed against every executable identity/content/target validation branch and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists."
+              "review_reason": "Reviewed against every executable identity/content/target validation branch, authenticated contact-search response shape, and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists."
             },
             {
               "value": "write",
@@ -9935,7 +9935,7 @@
           "source": "internal/cli/schema_hints/selection/chat.json",
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Agent-authored from the verified identity/content capability matrix, userId resolution, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media.",
+          "review_reason": "Agent-authored from the verified identity/content capability matrix, live-compatible contact search with exact userId matching, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media.",
           "candidates": [
             {
               "value": [
@@ -9945,7 +9945,7 @@
               "source": "internal/cli/schema_hints/selection/chat.json",
               "precedence": "reviewed_explicit",
               "selected": true,
-              "review_reason": "Agent-authored from the verified identity/content capability matrix, userId resolution, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media."
+              "review_reason": "Agent-authored from the verified identity/content capability matrix, live-compatible contact search with exact userId matching, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media."
             }
           ]
         },
@@ -9954,14 +9954,14 @@
           "source": "internal/cli/schema_hints/metadata/chat.json",
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Reviewed against every executable identity/content/target validation branch and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
+          "review_reason": "Reviewed against every executable identity/content/target validation branch, authenticated contact-search response shape, and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
           "candidates": [
             {
               "value": "unknown",
               "source": "internal/cli/schema_hints/metadata/chat.json",
               "precedence": "reviewed_explicit",
               "selected": true,
-              "review_reason": "Reviewed against every executable identity/content/target validation branch and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists."
+              "review_reason": "Reviewed against every executable identity/content/target validation branch, authenticated contact-search response shape, and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists."
             },
             {
               "value": "unknown",
@@ -9976,30 +9976,30 @@
           "source": "internal/cli/schema_hints/metadata/chat.json",
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Reviewed against every executable identity/content/target validation branch and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
+          "review_reason": "Reviewed against every executable identity/content/target validation branch, authenticated contact-search response shape, and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
           "candidates": [
             {
               "value": "composite",
               "source": "internal/cli/schema_hints/metadata/chat.json",
               "precedence": "reviewed_explicit",
               "selected": true,
-              "review_reason": "Reviewed against every executable identity/content/target validation branch and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists."
+              "review_reason": "Reviewed against every executable identity/content/target validation branch, authenticated contact-search response shape, and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists."
             }
           ]
         },
         "interface_reason": {
-          "value": "Reviewed composite send adapter: it selects current-user, bot, or webhook transport; current-user additionally supports userId resolution, mediaId images, and the native init/upload/commit local-file flow.",
+          "value": "Reviewed composite send adapter: it selects current-user, bot, or webhook transport; current-user additionally supports live-compatible contact search with exact userId matching, mediaId images, and the native init/upload/commit local-file flow.",
           "source": "internal/cli/schema_hints/metadata/chat.json",
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Reviewed against every executable identity/content/target validation branch and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
+          "review_reason": "Reviewed against every executable identity/content/target validation branch, authenticated contact-search response shape, and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
           "candidates": [
             {
-              "value": "Reviewed composite send adapter: it selects current-user, bot, or webhook transport; current-user additionally supports userId resolution, mediaId images, and the native init/upload/commit local-file flow.",
+              "value": "Reviewed composite send adapter: it selects current-user, bot, or webhook transport; current-user additionally supports live-compatible contact search with exact userId matching, mediaId images, and the native init/upload/commit local-file flow.",
               "source": "internal/cli/schema_hints/metadata/chat.json",
               "precedence": "reviewed_explicit",
               "selected": true,
-              "review_reason": "Reviewed against every executable identity/content/target validation branch and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists."
+              "review_reason": "Reviewed against every executable identity/content/target validation branch, authenticated contact-search response shape, and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists."
             }
           ]
         },
@@ -10024,21 +10024,21 @@
           "source": "internal/cli/schema_hints/metadata/chat.json",
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Reviewed against every executable identity/content/target validation branch and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
+          "review_reason": "Reviewed against every executable identity/content/target validation branch, authenticated contact-search response shape, and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
           "candidates": [
             {
               "value": true,
               "source": "internal/cli/schema_hints/metadata/chat.json",
               "precedence": "reviewed_explicit",
               "selected": true,
-              "review_reason": "Reviewed against every executable identity/content/target validation branch and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists."
+              "review_reason": "Reviewed against every executable identity/content/target validation branch, authenticated contact-search response shape, and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists."
             },
             {
               "value": true,
               "source": "internal/cli/schema_hints/selection/chat.json",
               "precedence": "reviewed_explicit",
               "selected": false,
-              "review_reason": "Agent-authored from the verified identity/content capability matrix, userId resolution, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media."
+              "review_reason": "Agent-authored from the verified identity/content capability matrix, live-compatible contact search with exact userId matching, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media."
             }
           ]
         },
@@ -10047,14 +10047,14 @@
           "source": "internal/cli/schema_hints/metadata/chat.json",
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Reviewed against every executable identity/content/target validation branch and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
+          "review_reason": "Reviewed against every executable identity/content/target validation branch, authenticated contact-search response shape, and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
           "candidates": [
             {
               "value": "medium",
               "source": "internal/cli/schema_hints/metadata/chat.json",
               "precedence": "reviewed_explicit",
               "selected": true,
-              "review_reason": "Reviewed against every executable identity/content/target validation branch and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists."
+              "review_reason": "Reviewed against every executable identity/content/target validation branch, authenticated contact-search response shape, and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists."
             },
             {
               "value": "medium",
@@ -10066,28 +10066,28 @@
         },
         "use_when": {
           "value": [
-            "当你希望用同一个入口选择 current-user、bot 或 webhook 身份发送消息时使用；命令会按身份校验目标、内容和凭据并路由到真实下层。current-user 支持文本/Markdown、已有 mediaId 图片、安全相对路径文件上传和幂等键；--user 传 userId 时包括在 --dry-run 中也会先只读解析为 openDingTalkId。bot 支持群聊或批量单聊文本/Markdown；webhook 的目标由 token 所在群决定。不会把 user 文件能力伪装成 bot/webhook 等价能力。"
+            "当你希望用同一个入口选择 current-user、bot 或 webhook 身份发送消息时使用；命令会按身份校验目标、内容和凭据并路由到真实下层。current-user 支持文本/Markdown、已有 mediaId 图片、安全相对路径文件上传和幂等键；--user 传 userId 时包括在 --dry-run 中也会先通过通讯录关键词搜索并按 userId 精确匹配 openDingTalkId。bot 支持群聊或批量单聊文本/Markdown；webhook 的目标由 token 所在群决定。不会把 user 文件能力伪装成 bot/webhook 等价能力。"
           ],
           "source": "internal/cli/schema_hints/selection/chat.json",
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Agent-authored from the verified identity/content capability matrix, userId resolution, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media.",
+          "review_reason": "Agent-authored from the verified identity/content capability matrix, live-compatible contact search with exact userId matching, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media.",
           "candidates": [
             {
               "value": [
-                "当你希望用同一个入口选择 current-user、bot 或 webhook 身份发送消息时使用；命令会按身份校验目标、内容和凭据并路由到真实下层。current-user 支持文本/Markdown、已有 mediaId 图片、安全相对路径文件上传和幂等键；--user 传 userId 时包括在 --dry-run 中也会先只读解析为 openDingTalkId。bot 支持群聊或批量单聊文本/Markdown；webhook 的目标由 token 所在群决定。不会把 user 文件能力伪装成 bot/webhook 等价能力。"
+                "当你希望用同一个入口选择 current-user、bot 或 webhook 身份发送消息时使用；命令会按身份校验目标、内容和凭据并路由到真实下层。current-user 支持文本/Markdown、已有 mediaId 图片、安全相对路径文件上传和幂等键；--user 传 userId 时包括在 --dry-run 中也会先通过通讯录关键词搜索并按 userId 精确匹配 openDingTalkId。bot 支持群聊或批量单聊文本/Markdown；webhook 的目标由 token 所在群决定。不会把 user 文件能力伪装成 bot/webhook 等价能力。"
               ],
               "source": "internal/cli/schema_hints/selection/chat.json",
               "precedence": "reviewed_explicit",
               "selected": true,
-              "review_reason": "Agent-authored from the verified identity/content capability matrix, userId resolution, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media."
+              "review_reason": "Agent-authored from the verified identity/content capability matrix, live-compatible contact search with exact userId matching, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media."
             }
           ]
         }
       },
       "idempotency": "unknown",
       "interface_mode": "composite",
-      "interface_reason": "Reviewed composite send adapter: it selects current-user, bot, or webhook transport; current-user additionally supports userId resolution, mediaId images, and the native init/upload/commit local-file flow.",
+      "interface_reason": "Reviewed composite send adapter: it selects current-user, bot, or webhook transport; current-user additionally supports live-compatible contact search with exact userId matching, mediaId images, and the native init/upload/commit local-file flow.",
       "reviewed": true,
       "risk": "medium",
       "source_refs": [
@@ -10099,7 +10099,7 @@
         "skills/mono/references/intent-guide.md"
       ],
       "use_when": [
-        "当你希望用同一个入口选择 current-user、bot 或 webhook 身份发送消息时使用；命令会按身份校验目标、内容和凭据并路由到真实下层。current-user 支持文本/Markdown、已有 mediaId 图片、安全相对路径文件上传和幂等键；--user 传 userId 时包括在 --dry-run 中也会先只读解析为 openDingTalkId。bot 支持群聊或批量单聊文本/Markdown；webhook 的目标由 token 所在群决定。不会把 user 文件能力伪装成 bot/webhook 等价能力。"
+        "当你希望用同一个入口选择 current-user、bot 或 webhook 身份发送消息时使用；命令会按身份校验目标、内容和凭据并路由到真实下层。current-user 支持文本/Markdown、已有 mediaId 图片、安全相对路径文件上传和幂等键；--user 传 userId 时包括在 --dry-run 中也会先通过通讯录关键词搜索并按 userId 精确匹配 openDingTalkId。bot 支持群聊或批量单聊文本/Markdown；webhook 的目标由 token 所在群决定。不会把 user 文件能力伪装成 bot/webhook 等价能力。"
       ]
     },
     "chat +messages-send-by-webhook": {
@@ -10380,14 +10380,14 @@
           "source": "internal/cli/schema_hints/selection/chat.json",
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting.",
+          "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, live-compatible contact search with exact userId matching, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting.",
           "candidates": [
             {
               "value": "创建流式卡片，可在同一次调用中写入内容并结束",
               "source": "internal/cli/schema_hints/selection/chat.json",
               "precedence": "reviewed_explicit",
               "selected": true,
-              "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting."
+              "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, live-compatible contact search with exact userId matching, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting."
             }
           ]
         },
@@ -10396,14 +10396,14 @@
           "source": "internal/cli/schema_hints/metadata/chat.json",
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
+          "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
           "candidates": [
             {
               "value": "available",
               "source": "internal/cli/schema_hints/metadata/chat.json",
               "precedence": "reviewed_explicit",
               "selected": true,
-              "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property."
+              "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property."
             }
           ]
         },
@@ -10414,7 +10414,7 @@
           "source": "internal/cli/schema_hints/selection/chat.json",
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting.",
+          "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, live-compatible contact search with exact userId matching, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting.",
           "candidates": [
             {
               "value": [
@@ -10423,7 +10423,7 @@
               "source": "internal/cli/schema_hints/selection/chat.json",
               "precedence": "reviewed_explicit",
               "selected": true,
-              "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting."
+              "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, live-compatible contact search with exact userId matching, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting."
             }
           ]
         },
@@ -10432,14 +10432,14 @@
           "source": "internal/cli/schema_hints/metadata/chat.json",
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
+          "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
           "candidates": [
             {
               "value": "user_required",
               "source": "internal/cli/schema_hints/metadata/chat.json",
               "precedence": "reviewed_explicit",
               "selected": true,
-              "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property."
+              "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property."
             }
           ]
         },
@@ -10448,14 +10448,14 @@
           "source": "internal/cli/schema_hints/metadata/chat.json",
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
+          "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
           "candidates": [
             {
               "value": "write",
               "source": "internal/cli/schema_hints/metadata/chat.json",
               "precedence": "reviewed_explicit",
               "selected": true,
-              "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property."
+              "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property."
             }
           ]
         },
@@ -10467,7 +10467,7 @@
           "source": "internal/cli/schema_hints/selection/chat.json",
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting.",
+          "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, live-compatible contact search with exact userId matching, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting.",
           "candidates": [
             {
               "value": [
@@ -10477,7 +10477,7 @@
               "source": "internal/cli/schema_hints/selection/chat.json",
               "precedence": "reviewed_explicit",
               "selected": true,
-              "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting."
+              "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, live-compatible contact search with exact userId matching, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting."
             }
           ]
         },
@@ -10486,14 +10486,14 @@
           "source": "internal/cli/schema_hints/metadata/chat.json",
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
+          "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
           "candidates": [
             {
               "value": "unknown",
               "source": "internal/cli/schema_hints/metadata/chat.json",
               "precedence": "reviewed_explicit",
               "selected": true,
-              "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property."
+              "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property."
             }
           ]
         },
@@ -10502,30 +10502,30 @@
           "source": "internal/cli/schema_hints/metadata/chat.json",
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
+          "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
           "candidates": [
             {
               "value": "composite",
               "source": "internal/cli/schema_hints/metadata/chat.json",
               "precedence": "reviewed_explicit",
               "selected": true,
-              "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property."
+              "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property."
             }
           ]
         },
         "interface_reason": {
-          "value": "Reviewed card lifecycle adapter: it can call create_and_send_card alone or compose it with update_streaming_card after extracting the returned bizId.",
+          "value": "Reviewed card lifecycle adapter: it can resolve a userId through contact search with exact matching, call create_and_send_card alone, or compose creation with update_streaming_card after extracting the returned bizId.",
           "source": "internal/cli/schema_hints/metadata/chat.json",
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
+          "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
           "candidates": [
             {
-              "value": "Reviewed card lifecycle adapter: it can call create_and_send_card alone or compose it with update_streaming_card after extracting the returned bizId.",
+              "value": "Reviewed card lifecycle adapter: it can resolve a userId through contact search with exact matching, call create_and_send_card alone, or compose creation with update_streaming_card after extracting the returned bizId.",
               "source": "internal/cli/schema_hints/metadata/chat.json",
               "precedence": "reviewed_explicit",
               "selected": true,
-              "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property."
+              "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property."
             }
           ]
         },
@@ -10550,21 +10550,21 @@
           "source": "internal/cli/schema_hints/metadata/chat.json",
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
+          "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
           "candidates": [
             {
               "value": true,
               "source": "internal/cli/schema_hints/metadata/chat.json",
               "precedence": "reviewed_explicit",
               "selected": true,
-              "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property."
+              "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property."
             },
             {
               "value": true,
               "source": "internal/cli/schema_hints/selection/chat.json",
               "precedence": "reviewed_explicit",
               "selected": false,
-              "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting."
+              "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, live-compatible contact search with exact userId matching, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting."
             }
           ]
         },
@@ -10573,41 +10573,41 @@
           "source": "internal/cli/schema_hints/metadata/chat.json",
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
+          "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
           "candidates": [
             {
               "value": "medium",
               "source": "internal/cli/schema_hints/metadata/chat.json",
               "precedence": "reviewed_explicit",
               "selected": true,
-              "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property."
+              "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property."
             }
           ]
         },
         "use_when": {
           "value": [
-            "当你要发送一张流式卡片消息时使用；群 openConversationId、单聊 userId、单聊 openDingTalkId 严格三选一，分别使用 --group、--receiver、--receiver-open-dingtalk-id。--receiver 始终按 userId 做只读通讯录解析，即使值以 D/d 开头也不会猜成 openDingTalkId；已有 openDingTalkId 时必须用显式参数直传。userId 包括在 --dry-run 时也会先解析。只传目标时创建卡片并返回 bizId，供后续 messages-update-card 流式更新；同时传 --content 时会自动串联创建和更新，默认以 flowStatus=3 完成，避免卡片停留在加载中。"
+            "当你要发送一张流式卡片消息时使用；群 openConversationId、单聊 userId、单聊 openDingTalkId 严格三选一，分别使用 --group、--receiver、--receiver-open-dingtalk-id。--receiver 始终按 userId 通过通讯录关键词搜索做精确匹配，即使值以 D/d 开头也不会猜成 openDingTalkId；已有 openDingTalkId 时必须用显式参数直传。userId 包括在 --dry-run 时也会先解析。只传目标时创建卡片并返回 bizId，供后续 messages-update-card 流式更新；同时传 --content 时会自动串联创建和更新，默认以 flowStatus=3 完成，避免卡片停留在加载中。"
           ],
           "source": "internal/cli/schema_hints/selection/chat.json",
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting.",
+          "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, live-compatible contact search with exact userId matching, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting.",
           "candidates": [
             {
               "value": [
-                "当你要发送一张流式卡片消息时使用；群 openConversationId、单聊 userId、单聊 openDingTalkId 严格三选一，分别使用 --group、--receiver、--receiver-open-dingtalk-id。--receiver 始终按 userId 做只读通讯录解析，即使值以 D/d 开头也不会猜成 openDingTalkId；已有 openDingTalkId 时必须用显式参数直传。userId 包括在 --dry-run 时也会先解析。只传目标时创建卡片并返回 bizId，供后续 messages-update-card 流式更新；同时传 --content 时会自动串联创建和更新，默认以 flowStatus=3 完成，避免卡片停留在加载中。"
+                "当你要发送一张流式卡片消息时使用；群 openConversationId、单聊 userId、单聊 openDingTalkId 严格三选一，分别使用 --group、--receiver、--receiver-open-dingtalk-id。--receiver 始终按 userId 通过通讯录关键词搜索做精确匹配，即使值以 D/d 开头也不会猜成 openDingTalkId；已有 openDingTalkId 时必须用显式参数直传。userId 包括在 --dry-run 时也会先解析。只传目标时创建卡片并返回 bizId，供后续 messages-update-card 流式更新；同时传 --content 时会自动串联创建和更新，默认以 flowStatus=3 完成，避免卡片停留在加载中。"
               ],
               "source": "internal/cli/schema_hints/selection/chat.json",
               "precedence": "reviewed_explicit",
               "selected": true,
-              "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting."
+              "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, live-compatible contact search with exact userId matching, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting."
             }
           ]
         }
       },
       "idempotency": "unknown",
       "interface_mode": "composite",
-      "interface_reason": "Reviewed card lifecycle adapter: it can call create_and_send_card alone or compose it with update_streaming_card after extracting the returned bizId.",
+      "interface_reason": "Reviewed card lifecycle adapter: it can resolve a userId through contact search with exact matching, call create_and_send_card alone, or compose creation with update_streaming_card after extracting the returned bizId.",
       "reviewed": true,
       "risk": "medium",
       "source_refs": [
@@ -10618,7 +10618,7 @@
         "internal/cli/schema_hints/selection/chat.json"
       ],
       "use_when": [
-        "当你要发送一张流式卡片消息时使用；群 openConversationId、单聊 userId、单聊 openDingTalkId 严格三选一，分别使用 --group、--receiver、--receiver-open-dingtalk-id。--receiver 始终按 userId 做只读通讯录解析，即使值以 D/d 开头也不会猜成 openDingTalkId；已有 openDingTalkId 时必须用显式参数直传。userId 包括在 --dry-run 时也会先解析。只传目标时创建卡片并返回 bizId，供后续 messages-update-card 流式更新；同时传 --content 时会自动串联创建和更新，默认以 flowStatus=3 完成，避免卡片停留在加载中。"
+        "当你要发送一张流式卡片消息时使用；群 openConversationId、单聊 userId、单聊 openDingTalkId 严格三选一，分别使用 --group、--receiver、--receiver-open-dingtalk-id。--receiver 始终按 userId 通过通讯录关键词搜索做精确匹配，即使值以 D/d 开头也不会猜成 openDingTalkId；已有 openDingTalkId 时必须用显式参数直传。userId 包括在 --dry-run 时也会先解析。只传目标时创建卡片并返回 bizId，供后续 messages-update-card 流式更新；同时传 --content 时会自动串联创建和更新，默认以 flowStatus=3 完成，避免卡片停留在加载中。"
       ]
     },
     "chat +messages-update-card": {

internal/cli/schema_agent_metadata/index.json

@@ -1,6 +1,6 @@
 {
   "version": 1,
-  "source_hash": "sha256:f7afa1d0abaa0c3d4ead4c6262948b7565787abb0860bb724276d1d3fe52178b",
+  "source_hash": "sha256:f8cb29bdbeaceae6f04cbac8a93ea77f2a84f93f8304d25878c827b959bc0ba0",
   "surface_hash": "sha256:60eee8e2f37d6d9d60689efce85082798eb9ad38b7ba7c0b471c3de676a85a16",
   "coverage": {
     "surface_products": 26,

internal/cli/schema_agent_metadata_audit.json

@@ -1,6 +1,6 @@
 {
   "version": 1,
-  "source_hash": "sha256:f7afa1d0abaa0c3d4ead4c6262948b7565787abb0860bb724276d1d3fe52178b",
+  "source_hash": "sha256:f8cb29bdbeaceae6f04cbac8a93ea77f2a84f93f8304d25878c827b959bc0ba0",
   "surface_hash": "sha256:60eee8e2f37d6d9d60689efce85082798eb9ad38b7ba7c0b471c3de676a85a16",
   "source_files": 160,
   "hint_files": 54,

internal/cli/schema_catalog/catalog.json

@@ -1,12 +1,12 @@
 {
   "version": 1,
   "surface_hash": "sha256:60eee8e2f37d6d9d60689efce85082798eb9ad38b7ba7c0b471c3de676a85a16",
-  "source_hash": "sha256:8fcebd5100f2727898290118b2b381cc3edc325bc1c8526e61036da5482e025b",
+  "source_hash": "sha256:735bbf8e04cfc44db741e9b14ffe1698f80aac5d2662d2c1c41b0aadcee9098d",
   "catalog": {
     "agent_metadata": {
       "products_with_metadata": 26,
       "source": "embedded-skill-metadata",
-      "source_hash": "sha256:f7afa1d0abaa0c3d4ead4c6262948b7565787abb0860bb724276d1d3fe52178b",
+      "source_hash": "sha256:f8cb29bdbeaceae6f04cbac8a93ea77f2a84f93f8304d25878c827b959bc0ba0",
       "surface_hash": "sha256:60eee8e2f37d6d9d60689efce85082798eb9ad38b7ba7c0b471c3de676a85a16",
       "surface_products": 26,
       "surface_tools": 845,
@@ -10788,18 +10788,18 @@
             "cli_name": "+messages-send",
             "cli_path": "chat +messages-send",
             "confirmation": "user_required",
-            "description": "当你希望用同一个入口选择 current-user、bot 或 webhook 身份发送消息时使用；命令会按身份校验目标、内容和凭据并路由到真实下层。current-user 支持文本/Markdown、已有 mediaId 图片、安全相对路径文件上传和幂等键；--user 传 userId 时包括在 --dry-run 中也会先只读解析为 openDingTalkId。bot 支持群聊或批量单聊文本/Markdown；webhook 的目标由 token 所在群决定。不会把 user 文件能力伪装成 bot/webhook 等价能力。\n\n参数约束：\n  - --text、--markdown、--media-id、--file、--file-path 至少指定一个\n  - --text、--markdown、--media-id、--file、--file-path 互斥，最多指定一个\n  - --identity、--as 互斥，最多指定一个\n  - --group、--chat-id 互斥，最多指定一个\n  - --user、--open-dingtalk-id 互斥，最多指定一个\n  - --uuid、--idempotency-key 互斥，最多指定一个\n  - 目标、凭据和幂等参数受发送身份能力矩阵约束：user 必须指定一个群聊或单聊目标；bot 必须指定 robot-code 和一类目标；webhook 必须指定 webhook-token；幂等键仅 user 支持",
+            "description": "当你希望用同一个入口选择 current-user、bot 或 webhook 身份发送消息时使用；命令会按身份校验目标、内容和凭据并路由到真实下层。current-user 支持文本/Markdown、已有 mediaId 图片、安全相对路径文件上传和幂等键；--user 传 userId 时包括在 --dry-run 中也会先通过通讯录关键词搜索并按 userId 精确匹配 openDingTalkId。bot 支持群聊或批量单聊文本/Markdown；webhook 的目标由 token 所在群决定。不会把 user 文件能力伪装成 bot/webhook 等价能力。\n\n参数约束：\n  - --text、--markdown、--media-id、--file、--file-path 至少指定一个\n  - --text、--markdown、--media-id、--file、--file-path 互斥，最多指定一个\n  - --identity、--as 互斥，最多指定一个\n  - --group、--chat-id 互斥，最多指定一个\n  - --user、--open-dingtalk-id 互斥，最多指定一个\n  - --uuid、--idempotency-key 互斥，最多指定一个\n  - 目标、凭据和幂等参数受发送身份能力矩阵约束：user 必须指定一个群聊或单聊目标；bot 必须指定 robot-code 和一类目标；webhook 必须指定 webhook-token；幂等键仅 user 支持",
             "effect": "write",
             "idempotency": "unknown",
             "interface_mode": "composite",
-            "interface_reason": "Reviewed composite send adapter: it selects current-user, bot, or webhook transport; current-user additionally supports userId resolution, mediaId images, and the native init/upload/commit local-file flow.",
+            "interface_reason": "Reviewed composite send adapter: it selects current-user, bot, or webhook transport; current-user additionally supports live-compatible contact search with exact userId matching, mediaId images, and the native init/upload/commit local-file flow.",
             "name": "shortcut_messages_send",
             "primary_cli_path": "chat +messages-send",
             "reviewed": true,
             "risk": "medium",
             "title": "统一发送文本、Markdown、当前用户文件或已有 mediaId 图片",
             "use_when": [
-              "当你希望用同一个入口选择 current-user、bot 或 webhook 身份发送消息时使用；命令会按身份校验目标、内容和凭据并路由到真实下层。current-user 支持文本/Markdown、已有 mediaId 图片、安全相对路径文件上传和幂等键；--user 传 userId 时包括在 --dry-run 中也会先只读解析为 openDingTalkId。bot 支持群聊或批量单聊文本/Markdown；webhook 的目标由 token 所在群决定。不会把 user 文件能力伪装成 bot/webhook 等价能力。"
+              "当你希望用同一个入口选择 current-user、bot 或 webhook 身份发送消息时使用；命令会按身份校验目标、内容和凭据并路由到真实下层。current-user 支持文本/Markdown、已有 mediaId 图片、安全相对路径文件上传和幂等键；--user 传 userId 时包括在 --dry-run 中也会先通过通讯录关键词搜索并按 userId 精确匹配 openDingTalkId。bot 支持群聊或批量单聊文本/Markdown；webhook 的目标由 token 所在群决定。不会把 user 文件能力伪装成 bot/webhook 等价能力。"
             ]
           },
           {
@@ -10840,18 +10840,18 @@
             "cli_name": "+messages-send-card",
             "cli_path": "chat +messages-send-card",
             "confirmation": "user_required",
-            "description": "当你要发送一张流式卡片消息时使用；群 openConversationId、单聊 userId、单聊 openDingTalkId 严格三选一，分别使用 --group、--receiver、--receiver-open-dingtalk-id。--receiver 始终按 userId 做只读通讯录解析，即使值以 D/d 开头也不会猜成 openDingTalkId；已有 openDingTalkId 时必须用显式参数直传。userId 包括在 --dry-run 时也会先解析。只传目标时创建卡片并返回 bizId，供后续 messages-update-card 流式更新；同时传 --content 时会自动串联创建和更新，默认以 flowStatus=3 完成，避免卡片停留在加载中。\n\n参数约束：\n  - --group、--receiver、--receiver-open-dingtalk-id 必须且只能指定一个\n  - --flow-status 必须在 1-5 之间，且显式指定时必须同时提供 --content",
+            "description": "当你要发送一张流式卡片消息时使用；群 openConversationId、单聊 userId、单聊 openDingTalkId 严格三选一，分别使用 --group、--receiver、--receiver-open-dingtalk-id。--receiver 始终按 userId 通过通讯录关键词搜索做精确匹配，即使值以 D/d 开头也不会猜成 openDingTalkId；已有 openDingTalkId 时必须用显式参数直传。userId 包括在 --dry-run 时也会先解析。只传目标时创建卡片并返回 bizId，供后续 messages-update-card 流式更新；同时传 --content 时会自动串联创建和更新，默认以 flowStatus=3 完成，避免卡片停留在加载中。\n\n参数约束：\n  - --group、--receiver、--receiver-open-dingtalk-id 必须且只能指定一个\n  - --flow-status 必须在 1-5 之间，且显式指定时必须同时提供 --content",
             "effect": "write",
             "idempotency": "unknown",
             "interface_mode": "composite",
-            "interface_reason": "Reviewed card lifecycle adapter: it can call create_and_send_card alone or compose it with update_streaming_card after extracting the returned bizId.",
+            "interface_reason": "Reviewed card lifecycle adapter: it can resolve a userId through contact search with exact matching, call create_and_send_card alone, or compose creation with update_streaming_card after extracting the returned bizId.",
             "name": "shortcut_messages_send_card",
             "primary_cli_path": "chat +messages-send-card",
             "reviewed": true,
             "risk": "medium",
             "title": "创建流式卡片，可在同一次调用中写入内容并结束",
             "use_when": [
-              "当你要发送一张流式卡片消息时使用；群 openConversationId、单聊 userId、单聊 openDingTalkId 严格三选一，分别使用 --group、--receiver、--receiver-open-dingtalk-id。--receiver 始终按 userId 做只读通讯录解析，即使值以 D/d 开头也不会猜成 openDingTalkId；已有 openDingTalkId 时必须用显式参数直传。userId 包括在 --dry-run 时也会先解析。只传目标时创建卡片并返回 bizId，供后续 messages-update-card 流式更新；同时传 --content 时会自动串联创建和更新，默认以 flowStatus=3 完成，避免卡片停留在加载中。"
+              "当你要发送一张流式卡片消息时使用；群 openConversationId、单聊 userId、单聊 openDingTalkId 严格三选一，分别使用 --group、--receiver、--receiver-open-dingtalk-id。--receiver 始终按 userId 通过通讯录关键词搜索做精确匹配，即使值以 D/d 开头也不会猜成 openDingTalkId；已有 openDingTalkId 时必须用显式参数直传。userId 包括在 --dry-run 时也会先解析。只传目标时创建卡片并返回 bizId，供后续 messages-update-card 流式更新；同时传 --content 时会自动串联创建和更新，默认以 flowStatus=3 完成，避免卡片停留在加载中。"
             ]
           },
           {

internal/cli/schema_catalog/tools/chat.json

@@ -73972,7 +73972,7 @@
           ]
         ]
       },
-      "description": "当你希望用同一个入口选择 current-user、bot 或 webhook 身份发送消息时使用；命令会按身份校验目标、内容和凭据并路由到真实下层。current-user 支持文本/Markdown、已有 mediaId 图片、安全相对路径文件上传和幂等键；--user 传 userId 时包括在 --dry-run 中也会先只读解析为 openDingTalkId。bot 支持群聊或批量单聊文本/Markdown；webhook 的目标由 token 所在群决定。不会把 user 文件能力伪装成 bot/webhook 等价能力。\n\n参数约束：\n  - --text、--markdown、--media-id、--file、--file-path 至少指定一个\n  - --text、--markdown、--media-id、--file、--file-path 互斥，最多指定一个\n  - --identity、--as 互斥，最多指定一个\n  - --group、--chat-id 互斥，最多指定一个\n  - --user、--open-dingtalk-id 互斥，最多指定一个\n  - --uuid、--idempotency-key 互斥，最多指定一个\n  - 目标、凭据和幂等参数受发送身份能力矩阵约束：user 必须指定一个群聊或单聊目标；bot 必须指定 robot-code 和一类目标；webhook 必须指定 webhook-token；幂等键仅 user 支持",
+      "description": "当你希望用同一个入口选择 current-user、bot 或 webhook 身份发送消息时使用；命令会按身份校验目标、内容和凭据并路由到真实下层。current-user 支持文本/Markdown、已有 mediaId 图片、安全相对路径文件上传和幂等键；--user 传 userId 时包括在 --dry-run 中也会先通过通讯录关键词搜索并按 userId 精确匹配 openDingTalkId。bot 支持群聊或批量单聊文本/Markdown；webhook 的目标由 token 所在群决定。不会把 user 文件能力伪装成 bot/webhook 等价能力。\n\n参数约束：\n  - --text、--markdown、--media-id、--file、--file-path 至少指定一个\n  - --text、--markdown、--media-id、--file、--file-path 互斥，最多指定一个\n  - --identity、--as 互斥，最多指定一个\n  - --group、--chat-id 互斥，最多指定一个\n  - --user、--open-dingtalk-id 互斥，最多指定一个\n  - --uuid、--idempotency-key 互斥，最多指定一个\n  - 目标、凭据和幂等参数受发送身份能力矩阵约束：user 必须指定一个群聊或单聊目标；bot 必须指定 robot-code 和一类目标；webhook 必须指定 webhook-token；幂等键仅 user 支持",
       "display": "群聊 / 消息 / 机器人",
       "effect": "write",
       "effect_source": "agent-hint",
@@ -73985,7 +73985,7 @@
           "candidates": [
             {
               "precedence": "reviewed_explicit",
-              "review_reason": "Agent-authored from the verified identity/content capability matrix, userId resolution, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media.",
+              "review_reason": "Agent-authored from the verified identity/content capability matrix, live-compatible contact search with exact userId matching, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media.",
               "selected": true,
               "source": "internal/cli/schema_hints/selection/chat.json",
               "value": "统一发送文本、Markdown、当前用户文件或已有 mediaId 图片"
@@ -73993,7 +73993,7 @@
           ],
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Agent-authored from the verified identity/content capability matrix, userId resolution, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media.",
+          "review_reason": "Agent-authored from the verified identity/content capability matrix, live-compatible contact search with exact userId matching, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media.",
           "source": "internal/cli/schema_hints/selection/chat.json",
           "value": "统一发送文本、Markdown、当前用户文件或已有 mediaId 图片"
         },
@@ -74001,7 +74001,7 @@
           "candidates": [
             {
               "precedence": "reviewed_explicit",
-              "review_reason": "Reviewed against every executable identity/content/target validation branch and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
+              "review_reason": "Reviewed against every executable identity/content/target validation branch, authenticated contact-search response shape, and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
               "selected": true,
               "source": "internal/cli/schema_hints/metadata/chat.json",
               "value": "available"
@@ -74009,7 +74009,7 @@
           ],
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Reviewed against every executable identity/content/target validation branch and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
+          "review_reason": "Reviewed against every executable identity/content/target validation branch, authenticated contact-search response shape, and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
           "source": "internal/cli/schema_hints/metadata/chat.json",
           "value": "available"
         },
@@ -74017,7 +74017,7 @@
           "candidates": [
             {
               "precedence": "reviewed_explicit",
-              "review_reason": "Agent-authored from the verified identity/content capability matrix, userId resolution, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media.",
+              "review_reason": "Agent-authored from the verified identity/content capability matrix, live-compatible contact search with exact userId matching, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media.",
               "selected": true,
               "source": "internal/cli/schema_hints/selection/chat.json",
               "value": [
@@ -74027,7 +74027,7 @@
           ],
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Agent-authored from the verified identity/content capability matrix, userId resolution, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media.",
+          "review_reason": "Agent-authored from the verified identity/content capability matrix, live-compatible contact search with exact userId matching, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media.",
           "source": "internal/cli/schema_hints/selection/chat.json",
           "value": [
             "需要 bot/webhook 发送媒体、卡片或 thread 回复时不要假设等价支持；改用真实存在的专用下层命令，缺少下层能力时停止"
@@ -74053,7 +74053,7 @@
           "candidates": [
             {
               "precedence": "reviewed_explicit",
-              "review_reason": "Reviewed against every executable identity/content/target validation branch and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
+              "review_reason": "Reviewed against every executable identity/content/target validation branch, authenticated contact-search response shape, and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
               "selected": true,
               "source": "internal/cli/schema_hints/metadata/chat.json",
               "value": "user_required"
@@ -74067,7 +74067,7 @@
           ],
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Reviewed against every executable identity/content/target validation branch and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
+          "review_reason": "Reviewed against every executable identity/content/target validation branch, authenticated contact-search response shape, and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
           "source": "internal/cli/schema_hints/metadata/chat.json",
           "value": "user_required"
         },
@@ -74077,19 +74077,19 @@
               "precedence": "cobra_help",
               "selected": true,
               "source": "cobra_help",
-              "value": "当你希望用同一个入口选择 current-user、bot 或 webhook 身份发送消息时使用；命令会按身份校验目标、内容和凭据并路由到真实下层。current-user 支持文本/Markdown、已有 mediaId 图片、安全相对路径文件上传和幂等键；--user 传 userId 时包括在 --dry-run 中也会先只读解析为 openDingTalkId。bot 支持群聊或批量单聊文本/Markdown；webhook 的目标由 token 所在群决定。不会把 user 文件能力伪装成 bot/webhook 等价能力。\n\n参数约束：\n  - --text、--markdown、--media-id、--file、--file-path 至少指定一个\n  - --text、--markdown、--media-id、--file、--file-path 互斥，最多指定一个\n  - --identity、--as 互斥，最多指定一个\n  - --group、--chat-id 互斥，最多指定一个\n  - --user、--open-dingtalk-id 互斥，最多指定一个\n  - --uuid、--idempotency-key 互斥，最多指定一个\n  - 目标、凭据和幂等参数受发送身份能力矩阵约束：user 必须指定一个群聊或单聊目标；bot 必须指定 robot-code 和一类目标；webhook 必须指定 webhook-token；幂等键仅 user 支持"
+              "value": "当你希望用同一个入口选择 current-user、bot 或 webhook 身份发送消息时使用；命令会按身份校验目标、内容和凭据并路由到真实下层。current-user 支持文本/Markdown、已有 mediaId 图片、安全相对路径文件上传和幂等键；--user 传 userId 时包括在 --dry-run 中也会先通过通讯录关键词搜索并按 userId 精确匹配 openDingTalkId。bot 支持群聊或批量单聊文本/Markdown；webhook 的目标由 token 所在群决定。不会把 user 文件能力伪装成 bot/webhook 等价能力。\n\n参数约束：\n  - --text、--markdown、--media-id、--file、--file-path 至少指定一个\n  - --text、--markdown、--media-id、--file、--file-path 互斥，最多指定一个\n  - --identity、--as 互斥，最多指定一个\n  - --group、--chat-id 互斥，最多指定一个\n  - --user、--open-dingtalk-id 互斥，最多指定一个\n  - --uuid、--idempotency-key 互斥，最多指定一个\n  - 目标、凭据和幂等参数受发送身份能力矩阵约束：user 必须指定一个群聊或单聊目标；bot 必须指定 robot-code 和一类目标；webhook 必须指定 webhook-token；幂等键仅 user 支持"
             }
           ],
           "precedence": "cobra_help",
           "resolution": "highest_precedence",
           "source": "cobra_help",
-          "value": "当你希望用同一个入口选择 current-user、bot 或 webhook 身份发送消息时使用；命令会按身份校验目标、内容和凭据并路由到真实下层。current-user 支持文本/Markdown、已有 mediaId 图片、安全相对路径文件上传和幂等键；--user 传 userId 时包括在 --dry-run 中也会先只读解析为 openDingTalkId。bot 支持群聊或批量单聊文本/Markdown；webhook 的目标由 token 所在群决定。不会把 user 文件能力伪装成 bot/webhook 等价能力。\n\n参数约束：\n  - --text、--markdown、--media-id、--file、--file-path 至少指定一个\n  - --text、--markdown、--media-id、--file、--file-path 互斥，最多指定一个\n  - --identity、--as 互斥，最多指定一个\n  - --group、--chat-id 互斥，最多指定一个\n  - --user、--open-dingtalk-id 互斥，最多指定一个\n  - --uuid、--idempotency-key 互斥，最多指定一个\n  - 目标、凭据和幂等参数受发送身份能力矩阵约束：user 必须指定一个群聊或单聊目标；bot 必须指定 robot-code 和一类目标；webhook 必须指定 webhook-token；幂等键仅 user 支持"
+          "value": "当你希望用同一个入口选择 current-user、bot 或 webhook 身份发送消息时使用；命令会按身份校验目标、内容和凭据并路由到真实下层。current-user 支持文本/Markdown、已有 mediaId 图片、安全相对路径文件上传和幂等键；--user 传 userId 时包括在 --dry-run 中也会先通过通讯录关键词搜索并按 userId 精确匹配 openDingTalkId。bot 支持群聊或批量单聊文本/Markdown；webhook 的目标由 token 所在群决定。不会把 user 文件能力伪装成 bot/webhook 等价能力。\n\n参数约束：\n  - --text、--markdown、--media-id、--file、--file-path 至少指定一个\n  - --text、--markdown、--media-id、--file、--file-path 互斥，最多指定一个\n  - --identity、--as 互斥，最多指定一个\n  - --group、--chat-id 互斥，最多指定一个\n  - --user、--open-dingtalk-id 互斥，最多指定一个\n  - --uuid、--idempotency-key 互斥，最多指定一个\n  - 目标、凭据和幂等参数受发送身份能力矩阵约束：user 必须指定一个群聊或单聊目标；bot 必须指定 robot-code 和一类目标；webhook 必须指定 webhook-token；幂等键仅 user 支持"
         },
         "effect": {
           "candidates": [
             {
               "precedence": "reviewed_explicit",
-              "review_reason": "Reviewed against every executable identity/content/target validation branch and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
+              "review_reason": "Reviewed against every executable identity/content/target validation branch, authenticated contact-search response shape, and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
               "selected": true,
               "source": "internal/cli/schema_hints/metadata/chat.json",
               "value": "write"
@@ -74103,7 +74103,7 @@
           ],
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Reviewed against every executable identity/content/target validation branch and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
+          "review_reason": "Reviewed against every executable identity/content/target validation branch, authenticated contact-search response shape, and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
           "source": "internal/cli/schema_hints/metadata/chat.json",
           "value": "write"
         },
@@ -74111,7 +74111,7 @@
           "candidates": [
             {
               "precedence": "reviewed_explicit",
-              "review_reason": "Agent-authored from the verified identity/content capability matrix, userId resolution, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media.",
+              "review_reason": "Agent-authored from the verified identity/content capability matrix, live-compatible contact search with exact userId matching, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media.",
               "selected": true,
               "source": "internal/cli/schema_hints/selection/chat.json",
               "value": [
@@ -74122,7 +74122,7 @@
           ],
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Agent-authored from the verified identity/content capability matrix, userId resolution, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media.",
+          "review_reason": "Agent-authored from the verified identity/content capability matrix, live-compatible contact search with exact userId matching, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media.",
           "source": "internal/cli/schema_hints/selection/chat.json",
           "value": [
             "dws chat +messages-send --as user --chat-id \u003copenConversationId\u003e --markdown \"## 周报\" --idempotency-key \u003ckey\u003e",
@@ -74133,7 +74133,7 @@
           "candidates": [
             {
               "precedence": "reviewed_explicit",
-              "review_reason": "Reviewed against every executable identity/content/target validation branch and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
+              "review_reason": "Reviewed against every executable identity/content/target validation branch, authenticated contact-search response shape, and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
               "selected": true,
               "source": "internal/cli/schema_hints/metadata/chat.json",
               "value": "unknown"
@@ -74147,7 +74147,7 @@
           ],
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Reviewed against every executable identity/content/target validation branch and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
+          "review_reason": "Reviewed against every executable identity/content/target validation branch, authenticated contact-search response shape, and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
           "source": "internal/cli/schema_hints/metadata/chat.json",
           "value": "unknown"
         },
@@ -74155,7 +74155,7 @@
           "candidates": [
             {
               "precedence": "reviewed_explicit",
-              "review_reason": "Reviewed against every executable identity/content/target validation branch and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
+              "review_reason": "Reviewed against every executable identity/content/target validation branch, authenticated contact-search response shape, and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
               "selected": true,
               "source": "internal/cli/schema_hints/metadata/chat.json",
               "value": "composite"
@@ -74163,7 +74163,7 @@
           ],
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Reviewed against every executable identity/content/target validation branch and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
+          "review_reason": "Reviewed against every executable identity/content/target validation branch, authenticated contact-search response shape, and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
           "source": "internal/cli/schema_hints/metadata/chat.json",
           "value": "composite"
         },
@@ -74171,17 +74171,17 @@
           "candidates": [
             {
               "precedence": "reviewed_explicit",
-              "review_reason": "Reviewed against every executable identity/content/target validation branch and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
+              "review_reason": "Reviewed against every executable identity/content/target validation branch, authenticated contact-search response shape, and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
               "selected": true,
               "source": "internal/cli/schema_hints/metadata/chat.json",
-              "value": "Reviewed composite send adapter: it selects current-user, bot, or webhook transport; current-user additionally supports userId resolution, mediaId images, and the native init/upload/commit local-file flow."
+              "value": "Reviewed composite send adapter: it selects current-user, bot, or webhook transport; current-user additionally supports live-compatible contact search with exact userId matching, mediaId images, and the native init/upload/commit local-file flow."
             }
           ],
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Reviewed against every executable identity/content/target validation branch and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
+          "review_reason": "Reviewed against every executable identity/content/target validation branch, authenticated contact-search response shape, and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
           "source": "internal/cli/schema_hints/metadata/chat.json",
-          "value": "Reviewed composite send adapter: it selects current-user, bot, or webhook transport; current-user additionally supports userId resolution, mediaId images, and the native init/upload/commit local-file flow."
+          "value": "Reviewed composite send adapter: it selects current-user, bot, or webhook transport; current-user additionally supports live-compatible contact search with exact userId matching, mediaId images, and the native init/upload/commit local-file flow."
         },
         "interface_ref": {
           "candidates": [
@@ -74203,14 +74203,14 @@
           "candidates": [
             {
               "precedence": "reviewed_explicit",
-              "review_reason": "Reviewed against every executable identity/content/target validation branch and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
+              "review_reason": "Reviewed against every executable identity/content/target validation branch, authenticated contact-search response shape, and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
               "selected": true,
               "source": "internal/cli/schema_hints/metadata/chat.json",
               "value": true
             },
             {
               "precedence": "reviewed_explicit",
-              "review_reason": "Agent-authored from the verified identity/content capability matrix, userId resolution, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media.",
+              "review_reason": "Agent-authored from the verified identity/content capability matrix, live-compatible contact search with exact userId matching, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media.",
               "selected": false,
               "source": "internal/cli/schema_hints/selection/chat.json",
               "value": true
@@ -74218,7 +74218,7 @@
           ],
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Reviewed against every executable identity/content/target validation branch and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
+          "review_reason": "Reviewed against every executable identity/content/target validation branch, authenticated contact-search response shape, and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
           "source": "internal/cli/schema_hints/metadata/chat.json",
           "value": true
         },
@@ -74226,7 +74226,7 @@
           "candidates": [
             {
               "precedence": "reviewed_explicit",
-              "review_reason": "Reviewed against every executable identity/content/target validation branch and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
+              "review_reason": "Reviewed against every executable identity/content/target validation branch, authenticated contact-search response shape, and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
               "selected": true,
               "source": "internal/cli/schema_hints/metadata/chat.json",
               "value": "medium"
@@ -74240,7 +74240,7 @@
           ],
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Reviewed against every executable identity/content/target validation branch and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
+          "review_reason": "Reviewed against every executable identity/content/target validation branch, authenticated contact-search response shape, and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
           "source": "internal/cli/schema_hints/metadata/chat.json",
           "value": "medium"
         },
@@ -74262,27 +74262,27 @@
           "candidates": [
             {
               "precedence": "reviewed_explicit",
-              "review_reason": "Agent-authored from the verified identity/content capability matrix, userId resolution, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media.",
+              "review_reason": "Agent-authored from the verified identity/content capability matrix, live-compatible contact search with exact userId matching, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media.",
               "selected": true,
               "source": "internal/cli/schema_hints/selection/chat.json",
               "value": [
-                "当你希望用同一个入口选择 current-user、bot 或 webhook 身份发送消息时使用；命令会按身份校验目标、内容和凭据并路由到真实下层。current-user 支持文本/Markdown、已有 mediaId 图片、安全相对路径文件上传和幂等键；--user 传 userId 时包括在 --dry-run 中也会先只读解析为 openDingTalkId。bot 支持群聊或批量单聊文本/Markdown；webhook 的目标由 token 所在群决定。不会把 user 文件能力伪装成 bot/webhook 等价能力。"
+                "当你希望用同一个入口选择 current-user、bot 或 webhook 身份发送消息时使用；命令会按身份校验目标、内容和凭据并路由到真实下层。current-user 支持文本/Markdown、已有 mediaId 图片、安全相对路径文件上传和幂等键；--user 传 userId 时包括在 --dry-run 中也会先通过通讯录关键词搜索并按 userId 精确匹配 openDingTalkId。bot 支持群聊或批量单聊文本/Markdown；webhook 的目标由 token 所在群决定。不会把 user 文件能力伪装成 bot/webhook 等价能力。"
               ]
             }
           ],
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Agent-authored from the verified identity/content capability matrix, userId resolution, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media.",
+          "review_reason": "Agent-authored from the verified identity/content capability matrix, live-compatible contact search with exact userId matching, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media.",
           "source": "internal/cli/schema_hints/selection/chat.json",
           "value": [
-            "当你希望用同一个入口选择 current-user、bot 或 webhook 身份发送消息时使用；命令会按身份校验目标、内容和凭据并路由到真实下层。current-user 支持文本/Markdown、已有 mediaId 图片、安全相对路径文件上传和幂等键；--user 传 userId 时包括在 --dry-run 中也会先只读解析为 openDingTalkId。bot 支持群聊或批量单聊文本/Markdown；webhook 的目标由 token 所在群决定。不会把 user 文件能力伪装成 bot/webhook 等价能力。"
+            "当你希望用同一个入口选择 current-user、bot 或 webhook 身份发送消息时使用；命令会按身份校验目标、内容和凭据并路由到真实下层。current-user 支持文本/Markdown、已有 mediaId 图片、安全相对路径文件上传和幂等键；--user 传 userId 时包括在 --dry-run 中也会先通过通讯录关键词搜索并按 userId 精确匹配 openDingTalkId。bot 支持群聊或批量单聊文本/Markdown；webhook 的目标由 token 所在群决定。不会把 user 文件能力伪装成 bot/webhook 等价能力。"
           ]
         }
       },
       "has_parameters": true,
       "idempotency": "unknown",
       "interface_mode": "composite",
-      "interface_reason": "Reviewed composite send adapter: it selects current-user, bot, or webhook transport; current-user additionally supports userId resolution, mediaId images, and the native init/upload/commit local-file flow.",
+      "interface_reason": "Reviewed composite send adapter: it selects current-user, bot, or webhook transport; current-user additionally supports live-compatible contact search with exact userId matching, mediaId images, and the native init/upload/commit local-file flow.",
       "is_alias": false,
       "name": "shortcut_messages_send",
       "parameter_count": 24,
@@ -76120,7 +76120,7 @@
           "type": "string"
         },
         "user": {
-          "description": "单聊接收者 userId（user；包括 --dry-run 也会先只读解析 openDingTalkId）；受发送身份能力矩阵约束",
+          "description": "单聊接收者 userId（user；包括 --dry-run 也会先通过通讯录搜索精确匹配 openDingTalkId）；受发送身份能力矩阵约束",
           "field_provenance": {
             "description": {
               "candidates": [
@@ -76128,7 +76128,7 @@
                   "precedence": "cobra_contract",
                   "selected": true,
                   "source": "cobra_usage",
-                  "value": "单聊接收者 userId（user；包括 --dry-run 也会先只读解析 openDingTalkId）；受发送身份能力矩阵约束"
+                  "value": "单聊接收者 userId（user；包括 --dry-run 也会先通过通讯录搜索精确匹配 openDingTalkId）；受发送身份能力矩阵约束"
                 },
                 {
                   "precedence": "default",
@@ -76140,7 +76140,7 @@
               "precedence": "cobra_contract",
               "resolution": "highest_precedence",
               "source": "cobra_usage",
-              "value": "单聊接收者 userId（user；包括 --dry-run 也会先只读解析 openDingTalkId）；受发送身份能力矩阵约束"
+              "value": "单聊接收者 userId（user；包括 --dry-run 也会先通过通讯录搜索精确匹配 openDingTalkId）；受发送身份能力矩阵约束"
             },
             "property": {
               "candidates": [
@@ -76464,7 +76464,7 @@
       "source": "reviewed_command_registry",
       "title": "统一发送文本、Markdown、当前用户文件或已有 mediaId 图片",
       "use_when": [
-        "当你希望用同一个入口选择 current-user、bot 或 webhook 身份发送消息时使用；命令会按身份校验目标、内容和凭据并路由到真实下层。current-user 支持文本/Markdown、已有 mediaId 图片、安全相对路径文件上传和幂等键；--user 传 userId 时包括在 --dry-run 中也会先只读解析为 openDingTalkId。bot 支持群聊或批量单聊文本/Markdown；webhook 的目标由 token 所在群决定。不会把 user 文件能力伪装成 bot/webhook 等价能力。"
+        "当你希望用同一个入口选择 current-user、bot 或 webhook 身份发送消息时使用；命令会按身份校验目标、内容和凭据并路由到真实下层。current-user 支持文本/Markdown、已有 mediaId 图片、安全相对路径文件上传和幂等键；--user 传 userId 时包括在 --dry-run 中也会先通过通讯录关键词搜索并按 userId 精确匹配 openDingTalkId。bot 支持群聊或批量单聊文本/Markdown；webhook 的目标由 token 所在群决定。不会把 user 文件能力伪装成 bot/webhook 等价能力。"
       ]
     },
     "chat.shortcut_messages_send_by_webhook": {
@@ -77361,7 +77361,7 @@
           ]
         ]
       },
-      "description": "当你要发送一张流式卡片消息时使用；群 openConversationId、单聊 userId、单聊 openDingTalkId 严格三选一，分别使用 --group、--receiver、--receiver-open-dingtalk-id。--receiver 始终按 userId 做只读通讯录解析，即使值以 D/d 开头也不会猜成 openDingTalkId；已有 openDingTalkId 时必须用显式参数直传。userId 包括在 --dry-run 时也会先解析。只传目标时创建卡片并返回 bizId，供后续 messages-update-card 流式更新；同时传 --content 时会自动串联创建和更新，默认以 flowStatus=3 完成，避免卡片停留在加载中。\n\n参数约束：\n  - --group、--receiver、--receiver-open-dingtalk-id 必须且只能指定一个\n  - --flow-status 必须在 1-5 之间，且显式指定时必须同时提供 --content",
+      "description": "当你要发送一张流式卡片消息时使用；群 openConversationId、单聊 userId、单聊 openDingTalkId 严格三选一，分别使用 --group、--receiver、--receiver-open-dingtalk-id。--receiver 始终按 userId 通过通讯录关键词搜索做精确匹配，即使值以 D/d 开头也不会猜成 openDingTalkId；已有 openDingTalkId 时必须用显式参数直传。userId 包括在 --dry-run 时也会先解析。只传目标时创建卡片并返回 bizId，供后续 messages-update-card 流式更新；同时传 --content 时会自动串联创建和更新，默认以 flowStatus=3 完成，避免卡片停留在加载中。\n\n参数约束：\n  - --group、--receiver、--receiver-open-dingtalk-id 必须且只能指定一个\n  - --flow-status 必须在 1-5 之间，且显式指定时必须同时提供 --content",
       "display": "群聊 / 消息 / 机器人",
       "effect": "write",
       "effect_source": "agent-hint",
@@ -77374,7 +77374,7 @@
           "candidates": [
             {
               "precedence": "reviewed_explicit",
-              "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting.",
+              "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, live-compatible contact search with exact userId matching, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting.",
               "selected": true,
               "source": "internal/cli/schema_hints/selection/chat.json",
               "value": "创建流式卡片，可在同一次调用中写入内容并结束"
@@ -77382,7 +77382,7 @@
           ],
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting.",
+          "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, live-compatible contact search with exact userId matching, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting.",
           "source": "internal/cli/schema_hints/selection/chat.json",
           "value": "创建流式卡片，可在同一次调用中写入内容并结束"
         },
@@ -77390,7 +77390,7 @@
           "candidates": [
             {
               "precedence": "reviewed_explicit",
-              "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
+              "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
               "selected": true,
               "source": "internal/cli/schema_hints/metadata/chat.json",
               "value": "available"
@@ -77398,7 +77398,7 @@
           ],
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
+          "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
           "source": "internal/cli/schema_hints/metadata/chat.json",
           "value": "available"
         },
@@ -77406,7 +77406,7 @@
           "candidates": [
             {
               "precedence": "reviewed_explicit",
-              "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting.",
+              "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, live-compatible contact search with exact userId matching, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting.",
               "selected": true,
               "source": "internal/cli/schema_hints/selection/chat.json",
               "value": [
@@ -77416,7 +77416,7 @@
           ],
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting.",
+          "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, live-compatible contact search with exact userId matching, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting.",
           "source": "internal/cli/schema_hints/selection/chat.json",
           "value": [
             "已有 bizId、只需要追加或更新现有卡片内容时使用 +messages-update-card"
@@ -77433,7 +77433,7 @@
             },
             {
               "precedence": "reviewed_manual",
-              "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
+              "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
               "selected": false,
               "source": "reviewed_manual_hint",
               "source_ref": "chat +messages-send-card",
@@ -77450,7 +77450,7 @@
           "candidates": [
             {
               "precedence": "reviewed_explicit",
-              "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
+              "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
               "selected": true,
               "source": "internal/cli/schema_hints/metadata/chat.json",
               "value": "user_required"
@@ -77458,7 +77458,7 @@
           ],
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
+          "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
           "source": "internal/cli/schema_hints/metadata/chat.json",
           "value": "user_required"
         },
@@ -77468,19 +77468,19 @@
               "precedence": "cobra_help",
               "selected": true,
               "source": "cobra_help",
-              "value": "当你要发送一张流式卡片消息时使用；群 openConversationId、单聊 userId、单聊 openDingTalkId 严格三选一，分别使用 --group、--receiver、--receiver-open-dingtalk-id。--receiver 始终按 userId 做只读通讯录解析，即使值以 D/d 开头也不会猜成 openDingTalkId；已有 openDingTalkId 时必须用显式参数直传。userId 包括在 --dry-run 时也会先解析。只传目标时创建卡片并返回 bizId，供后续 messages-update-card 流式更新；同时传 --content 时会自动串联创建和更新，默认以 flowStatus=3 完成，避免卡片停留在加载中。\n\n参数约束：\n  - --group、--receiver、--receiver-open-dingtalk-id 必须且只能指定一个\n  - --flow-status 必须在 1-5 之间，且显式指定时必须同时提供 --content"
+              "value": "当你要发送一张流式卡片消息时使用；群 openConversationId、单聊 userId、单聊 openDingTalkId 严格三选一，分别使用 --group、--receiver、--receiver-open-dingtalk-id。--receiver 始终按 userId 通过通讯录关键词搜索做精确匹配，即使值以 D/d 开头也不会猜成 openDingTalkId；已有 openDingTalkId 时必须用显式参数直传。userId 包括在 --dry-run 时也会先解析。只传目标时创建卡片并返回 bizId，供后续 messages-update-card 流式更新；同时传 --content 时会自动串联创建和更新，默认以 flowStatus=3 完成，避免卡片停留在加载中。\n\n参数约束：\n  - --group、--receiver、--receiver-open-dingtalk-id 必须且只能指定一个\n  - --flow-status 必须在 1-5 之间，且显式指定时必须同时提供 --content"
             }
           ],
           "precedence": "cobra_help",
           "resolution": "highest_precedence",
           "source": "cobra_help",
-          "value": "当你要发送一张流式卡片消息时使用；群 openConversationId、单聊 userId、单聊 openDingTalkId 严格三选一，分别使用 --group、--receiver、--receiver-open-dingtalk-id。--receiver 始终按 userId 做只读通讯录解析，即使值以 D/d 开头也不会猜成 openDingTalkId；已有 openDingTalkId 时必须用显式参数直传。userId 包括在 --dry-run 时也会先解析。只传目标时创建卡片并返回 bizId，供后续 messages-update-card 流式更新；同时传 --content 时会自动串联创建和更新，默认以 flowStatus=3 完成，避免卡片停留在加载中。\n\n参数约束：\n  - --group、--receiver、--receiver-open-dingtalk-id 必须且只能指定一个\n  - --flow-status 必须在 1-5 之间，且显式指定时必须同时提供 --content"
+          "value": "当你要发送一张流式卡片消息时使用；群 openConversationId、单聊 userId、单聊 openDingTalkId 严格三选一，分别使用 --group、--receiver、--receiver-open-dingtalk-id。--receiver 始终按 userId 通过通讯录关键词搜索做精确匹配，即使值以 D/d 开头也不会猜成 openDingTalkId；已有 openDingTalkId 时必须用显式参数直传。userId 包括在 --dry-run 时也会先解析。只传目标时创建卡片并返回 bizId，供后续 messages-update-card 流式更新；同时传 --content 时会自动串联创建和更新，默认以 flowStatus=3 完成，避免卡片停留在加载中。\n\n参数约束：\n  - --group、--receiver、--receiver-open-dingtalk-id 必须且只能指定一个\n  - --flow-status 必须在 1-5 之间，且显式指定时必须同时提供 --content"
         },
         "effect": {
           "candidates": [
             {
               "precedence": "reviewed_explicit",
-              "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
+              "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
               "selected": true,
               "source": "internal/cli/schema_hints/metadata/chat.json",
               "value": "write"
@@ -77488,7 +77488,7 @@
           ],
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
+          "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
           "source": "internal/cli/schema_hints/metadata/chat.json",
           "value": "write"
         },
@@ -77496,7 +77496,7 @@
           "candidates": [
             {
               "precedence": "reviewed_explicit",
-              "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting.",
+              "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, live-compatible contact search with exact userId matching, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting.",
               "selected": true,
               "source": "internal/cli/schema_hints/selection/chat.json",
               "value": [
@@ -77507,7 +77507,7 @@
           ],
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting.",
+          "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, live-compatible contact search with exact userId matching, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting.",
           "source": "internal/cli/schema_hints/selection/chat.json",
           "value": [
             "dws chat +messages-send-card --group \u003copenConversationId\u003e --content \"任务已完成\"",
@@ -77518,7 +77518,7 @@
           "candidates": [
             {
               "precedence": "reviewed_explicit",
-              "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
+              "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
               "selected": true,
               "source": "internal/cli/schema_hints/metadata/chat.json",
               "value": "unknown"
@@ -77526,7 +77526,7 @@
           ],
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
+          "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
           "source": "internal/cli/schema_hints/metadata/chat.json",
           "value": "unknown"
         },
@@ -77534,7 +77534,7 @@
           "candidates": [
             {
               "precedence": "reviewed_explicit",
-              "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
+              "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
               "selected": true,
               "source": "internal/cli/schema_hints/metadata/chat.json",
               "value": "composite"
@@ -77542,7 +77542,7 @@
           ],
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
+          "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
           "source": "internal/cli/schema_hints/metadata/chat.json",
           "value": "composite"
         },
@@ -77550,17 +77550,17 @@
           "candidates": [
             {
               "precedence": "reviewed_explicit",
-              "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
+              "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
               "selected": true,
               "source": "internal/cli/schema_hints/metadata/chat.json",
-              "value": "Reviewed card lifecycle adapter: it can call create_and_send_card alone or compose it with update_streaming_card after extracting the returned bizId."
+              "value": "Reviewed card lifecycle adapter: it can resolve a userId through contact search with exact matching, call create_and_send_card alone, or compose creation with update_streaming_card after extracting the returned bizId."
             }
           ],
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
+          "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
           "source": "internal/cli/schema_hints/metadata/chat.json",
-          "value": "Reviewed card lifecycle adapter: it can call create_and_send_card alone or compose it with update_streaming_card after extracting the returned bizId."
+          "value": "Reviewed card lifecycle adapter: it can resolve a userId through contact search with exact matching, call create_and_send_card alone, or compose creation with update_streaming_card after extracting the returned bizId."
         },
         "interface_ref": {
           "candidates": [
@@ -77582,14 +77582,14 @@
           "candidates": [
             {
               "precedence": "reviewed_explicit",
-              "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
+              "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
               "selected": true,
               "source": "internal/cli/schema_hints/metadata/chat.json",
               "value": true
             },
             {
               "precedence": "reviewed_explicit",
-              "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting.",
+              "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, live-compatible contact search with exact userId matching, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting.",
               "selected": false,
               "source": "internal/cli/schema_hints/selection/chat.json",
               "value": true
@@ -77597,7 +77597,7 @@
           ],
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
+          "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
           "source": "internal/cli/schema_hints/metadata/chat.json",
           "value": true
         },
@@ -77605,7 +77605,7 @@
           "candidates": [
             {
               "precedence": "reviewed_explicit",
-              "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
+              "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
               "selected": true,
               "source": "internal/cli/schema_hints/metadata/chat.json",
               "value": "medium"
@@ -77613,7 +77613,7 @@
           ],
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
+          "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
           "source": "internal/cli/schema_hints/metadata/chat.json",
           "value": "medium"
         },
@@ -77635,27 +77635,27 @@
           "candidates": [
             {
               "precedence": "reviewed_explicit",
-              "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting.",
+              "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, live-compatible contact search with exact userId matching, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting.",
               "selected": true,
               "source": "internal/cli/schema_hints/selection/chat.json",
               "value": [
-                "当你要发送一张流式卡片消息时使用；群 openConversationId、单聊 userId、单聊 openDingTalkId 严格三选一，分别使用 --group、--receiver、--receiver-open-dingtalk-id。--receiver 始终按 userId 做只读通讯录解析，即使值以 D/d 开头也不会猜成 openDingTalkId；已有 openDingTalkId 时必须用显式参数直传。userId 包括在 --dry-run 时也会先解析。只传目标时创建卡片并返回 bizId，供后续 messages-update-card 流式更新；同时传 --content 时会自动串联创建和更新，默认以 flowStatus=3 完成，避免卡片停留在加载中。"
+                "当你要发送一张流式卡片消息时使用；群 openConversationId、单聊 userId、单聊 openDingTalkId 严格三选一，分别使用 --group、--receiver、--receiver-open-dingtalk-id。--receiver 始终按 userId 通过通讯录关键词搜索做精确匹配，即使值以 D/d 开头也不会猜成 openDingTalkId；已有 openDingTalkId 时必须用显式参数直传。userId 包括在 --dry-run 时也会先解析。只传目标时创建卡片并返回 bizId，供后续 messages-update-card 流式更新；同时传 --content 时会自动串联创建和更新，默认以 flowStatus=3 完成，避免卡片停留在加载中。"
               ]
             }
           ],
           "precedence": "reviewed_explicit",
           "resolution": "highest_precedence",
-          "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting.",
+          "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, live-compatible contact search with exact userId matching, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting.",
           "source": "internal/cli/schema_hints/selection/chat.json",
           "value": [
-            "当你要发送一张流式卡片消息时使用；群 openConversationId、单聊 userId、单聊 openDingTalkId 严格三选一，分别使用 --group、--receiver、--receiver-open-dingtalk-id。--receiver 始终按 userId 做只读通讯录解析，即使值以 D/d 开头也不会猜成 openDingTalkId；已有 openDingTalkId 时必须用显式参数直传。userId 包括在 --dry-run 时也会先解析。只传目标时创建卡片并返回 bizId，供后续 messages-update-card 流式更新；同时传 --content 时会自动串联创建和更新，默认以 flowStatus=3 完成，避免卡片停留在加载中。"
+            "当你要发送一张流式卡片消息时使用；群 openConversationId、单聊 userId、单聊 openDingTalkId 严格三选一，分别使用 --group、--receiver、--receiver-open-dingtalk-id。--receiver 始终按 userId 通过通讯录关键词搜索做精确匹配，即使值以 D/d 开头也不会猜成 openDingTalkId；已有 openDingTalkId 时必须用显式参数直传。userId 包括在 --dry-run 时也会先解析。只传目标时创建卡片并返回 bizId，供后续 messages-update-card 流式更新；同时传 --content 时会自动串联创建和更新，默认以 flowStatus=3 完成，避免卡片停留在加载中。"
           ]
         }
       },
       "has_parameters": true,
       "idempotency": "unknown",
       "interface_mode": "composite",
-      "interface_reason": "Reviewed card lifecycle adapter: it can call create_and_send_card alone or compose it with update_streaming_card after extracting the returned bizId.",
+      "interface_reason": "Reviewed card lifecycle adapter: it can resolve a userId through contact search with exact matching, call create_and_send_card alone, or compose creation with update_streaming_card after extracting the returned bizId.",
       "is_alias": false,
       "name": "shortcut_messages_send_card",
       "parameter_count": 5,
@@ -77926,7 +77926,7 @@
           "type": "string"
         },
         "receiver": {
-          "description": "单聊接收者 userId（与 --group/--receiver-open-dingtalk-id 互斥）；始终只读解析为 openDingTalkId，包括 --dry-run 和 D/d 开头的 userId",
+          "description": "单聊接收者 userId（与 --group/--receiver-open-dingtalk-id 互斥）；始终通过通讯录搜索精确匹配 openDingTalkId，包括 --dry-run 和 D/d 开头的 userId",
           "field_provenance": {
             "description": {
               "candidates": [
@@ -77934,7 +77934,7 @@
                   "precedence": "cobra_contract",
                   "selected": true,
                   "source": "cobra_usage",
-                  "value": "单聊接收者 userId（与 --group/--receiver-open-dingtalk-id 互斥）；始终只读解析为 openDingTalkId，包括 --dry-run 和 D/d 开头的 userId"
+                  "value": "单聊接收者 userId（与 --group/--receiver-open-dingtalk-id 互斥）；始终通过通讯录搜索精确匹配 openDingTalkId，包括 --dry-run 和 D/d 开头的 userId"
                 },
                 {
                   "precedence": "default",
@@ -77946,7 +77946,7 @@
               "precedence": "cobra_contract",
               "resolution": "highest_precedence",
               "source": "cobra_usage",
-              "value": "单聊接收者 userId（与 --group/--receiver-open-dingtalk-id 互斥）；始终只读解析为 openDingTalkId，包括 --dry-run 和 D/d 开头的 userId"
+              "value": "单聊接收者 userId（与 --group/--receiver-open-dingtalk-id 互斥）；始终通过通讯录搜索精确匹配 openDingTalkId，包括 --dry-run 和 D/d 开头的 userId"
             },
             "property": {
               "candidates": [
@@ -78042,7 +78042,7 @@
               "candidates": [
                 {
                   "precedence": "reviewed_manual",
-                  "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
+                  "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
                   "selected": true,
                   "source": "reviewed_manual_hint",
                   "value": "receiverOpenDingTalkId"
@@ -78056,7 +78056,7 @@
               ],
               "precedence": "reviewed_manual",
               "resolution": "highest_precedence",
-              "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
+              "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
               "source": "reviewed_manual_hint",
               "value": "receiverOpenDingTalkId"
             },
@@ -78064,7 +78064,7 @@
               "candidates": [
                 {
                   "precedence": "reviewed_manual",
-                  "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
+                  "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
                   "selected": true,
                   "source": "reviewed_manual_hint",
                   "value": false
@@ -78084,7 +78084,7 @@
               ],
               "precedence": "reviewed_manual",
               "resolution": "highest_precedence",
-              "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
+              "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
               "source": "reviewed_manual_hint",
               "value": false
             },
@@ -78130,7 +78130,7 @@
       "source": "reviewed_command_registry",
       "title": "创建流式卡片，可在同一次调用中写入内容并结束",
       "use_when": [
-        "当你要发送一张流式卡片消息时使用；群 openConversationId、单聊 userId、单聊 openDingTalkId 严格三选一，分别使用 --group、--receiver、--receiver-open-dingtalk-id。--receiver 始终按 userId 做只读通讯录解析，即使值以 D/d 开头也不会猜成 openDingTalkId；已有 openDingTalkId 时必须用显式参数直传。userId 包括在 --dry-run 时也会先解析。只传目标时创建卡片并返回 bizId，供后续 messages-update-card 流式更新；同时传 --content 时会自动串联创建和更新，默认以 flowStatus=3 完成，避免卡片停留在加载中。"
+        "当你要发送一张流式卡片消息时使用；群 openConversationId、单聊 userId、单聊 openDingTalkId 严格三选一，分别使用 --group、--receiver、--receiver-open-dingtalk-id。--receiver 始终按 userId 通过通讯录关键词搜索做精确匹配，即使值以 D/d 开头也不会猜成 openDingTalkId；已有 openDingTalkId 时必须用显式参数直传。userId 包括在 --dry-run 时也会先解析。只传目标时创建卡片并返回 bizId，供后续 messages-update-card 流式更新；同时传 --content 时会自动串联创建和更新，默认以 flowStatus=3 完成，避免卡片停留在加载中。"
       ]
     },
     "chat.shortcut_messages_update_card": {

internal/cli/schema_hints/metadata/chat.json

@@ -1172,9 +1172,9 @@
       "idempotency": "unknown",
       "interface_mode": "composite",
       "availability": "available",
-      "interface_reason": "Reviewed composite send adapter: it selects current-user, bot, or webhook transport; current-user additionally supports userId resolution, mediaId images, and the native init/upload/commit local-file flow.",
+      "interface_reason": "Reviewed composite send adapter: it selects current-user, bot, or webhook transport; current-user additionally supports live-compatible contact search with exact userId matching, mediaId images, and the native init/upload/commit local-file flow.",
       "reviewed": true,
-      "review_reason": "Reviewed against every executable identity/content/target validation branch and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
+      "review_reason": "Reviewed against every executable identity/content/target validation branch, authenticated contact-search response shape, and the real lower tool mappings; bot and webhook media are intentionally rejected because no equivalent lower capability exists.",
       "cli_path": "chat +messages-send",
       "runtime_gate": "typed_yes"
     },
@@ -1198,7 +1198,7 @@
       "idempotency": "unknown",
       "interface_mode": "composite",
       "availability": "available",
-      "interface_reason": "Reviewed card lifecycle adapter: it can call create_and_send_card alone or compose it with update_streaming_card after extracting the returned bizId.",
+      "interface_reason": "Reviewed card lifecycle adapter: it can resolve a userId through contact search with exact matching, call create_and_send_card alone, or compose creation with update_streaming_card after extracting the returned bizId.",
       "parameters": {
         "receiver-open-dingtalk-id": {
           "property": "receiverOpenDingTalkId",
@@ -1206,7 +1206,7 @@
         }
       },
       "reviewed": true,
-      "review_reason": "Reviewed against create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
+      "review_reason": "Reviewed against authenticated contact-search resolution, create-only, two-action dry-run, create/update success, missing bizId, and partial-failure behavior that preserves the created bizId; the explicit openDingTalkId flag is bound to the lower interface's exact receiverOpenDingTalkId property.",
       "cli_path": "chat +messages-send-card",
       "runtime_gate": "typed_yes"
     },

internal/cli/schema_hints/selection/chat.json

@@ -2133,7 +2133,7 @@
     "chat.shortcut_messages_send": {
       "agent_summary": "统一发送文本、Markdown、当前用户文件或已有 mediaId 图片",
       "use_when": [
-        "当你希望用同一个入口选择 current-user、bot 或 webhook 身份发送消息时使用；命令会按身份校验目标、内容和凭据并路由到真实下层。current-user 支持文本/Markdown、已有 mediaId 图片、安全相对路径文件上传和幂等键；--user 传 userId 时包括在 --dry-run 中也会先只读解析为 openDingTalkId。bot 支持群聊或批量单聊文本/Markdown；webhook 的目标由 token 所在群决定。不会把 user 文件能力伪装成 bot/webhook 等价能力。"
+        "当你希望用同一个入口选择 current-user、bot 或 webhook 身份发送消息时使用；命令会按身份校验目标、内容和凭据并路由到真实下层。current-user 支持文本/Markdown、已有 mediaId 图片、安全相对路径文件上传和幂等键；--user 传 userId 时包括在 --dry-run 中也会先通过通讯录关键词搜索并按 userId 精确匹配 openDingTalkId。bot 支持群聊或批量单聊文本/Markdown；webhook 的目标由 token 所在群决定。不会把 user 文件能力伪装成 bot/webhook 等价能力。"
       ],
       "avoid_when": [
         "需要 bot/webhook 发送媒体、卡片或 thread 回复时不要假设等价支持；改用真实存在的专用下层命令，缺少下层能力时停止"
@@ -2143,7 +2143,7 @@
         "dws chat +messages-send --as user --user <userId> --msg-type file --file ./report.pdf"
       ],
       "reviewed": true,
-      "review_reason": "Agent-authored from the verified identity/content capability matrix, userId resolution, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media.",
+      "review_reason": "Agent-authored from the verified identity/content capability matrix, live-compatible contact search with exact userId matching, native local-file upload flow, and fail-closed rejection of unsupported bot/webhook media.",
       "source_refs": [
         "internal/cli/schema_command_registry.json#chat.shortcut_messages_send",
         "cobra-help:dws chat +messages-send",
@@ -2172,7 +2172,7 @@
     "chat.shortcut_messages_send_card": {
       "agent_summary": "创建流式卡片，可在同一次调用中写入内容并结束",
       "use_when": [
-        "当你要发送一张流式卡片消息时使用；群 openConversationId、单聊 userId、单聊 openDingTalkId 严格三选一，分别使用 --group、--receiver、--receiver-open-dingtalk-id。--receiver 始终按 userId 做只读通讯录解析，即使值以 D/d 开头也不会猜成 openDingTalkId；已有 openDingTalkId 时必须用显式参数直传。userId 包括在 --dry-run 时也会先解析。只传目标时创建卡片并返回 bizId，供后续 messages-update-card 流式更新；同时传 --content 时会自动串联创建和更新，默认以 flowStatus=3 完成，避免卡片停留在加载中。"
+        "当你要发送一张流式卡片消息时使用；群 openConversationId、单聊 userId、单聊 openDingTalkId 严格三选一，分别使用 --group、--receiver、--receiver-open-dingtalk-id。--receiver 始终按 userId 通过通讯录关键词搜索做精确匹配，即使值以 D/d 开头也不会猜成 openDingTalkId；已有 openDingTalkId 时必须用显式参数直传。userId 包括在 --dry-run 时也会先解析。只传目标时创建卡片并返回 bizId，供后续 messages-update-card 流式更新；同时传 --content 时会自动串联创建和更新，默认以 flowStatus=3 完成，避免卡片停留在加载中。"
       ],
       "avoid_when": [
         "已有 bizId、只需要追加或更新现有卡片内容时使用 +messages-update-card"
@@ -2182,7 +2182,7 @@
         "dws chat +messages-send-card --receiver <userId>"
       ],
       "reviewed": true,
-      "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting.",
+      "review_reason": "Agent-authored from the verified create-only and create-then-update lifecycle, live-compatible contact search with exact userId matching, explicit userId/openDingTalkId routing, two-action dry-run, bizId extraction, and partial-failure reporting.",
       "source_refs": [
         "internal/cli/schema_command_registry.json#chat.shortcut_messages_send_card",
         "cobra-help:dws chat +messages-send-card",

internal/shortcut/chat/chat_message.go

@@ -1015,11 +1015,11 @@ var MessagesSendCard = shortcut.Shortcut{
 	Command:     "+messages-send-card",
 	Product:     "im",
 	Description: "创建流式卡片，可在同一次调用中写入内容并结束",
-	Intent:      "当你要发送一张流式卡片消息时使用；群 openConversationId、单聊 userId、单聊 openDingTalkId 严格三选一，分别使用 --group、--receiver、--receiver-open-dingtalk-id。--receiver 始终按 userId 做只读通讯录解析，即使值以 D/d 开头也不会猜成 openDingTalkId；已有 openDingTalkId 时必须用显式参数直传。userId 包括在 --dry-run 时也会先解析。只传目标时创建卡片并返回 bizId，供后续 messages-update-card 流式更新；同时传 --content 时会自动串联创建和更新，默认以 flowStatus=3 完成，避免卡片停留在加载中。",
+	Intent:      "当你要发送一张流式卡片消息时使用；群 openConversationId、单聊 userId、单聊 openDingTalkId 严格三选一，分别使用 --group、--receiver、--receiver-open-dingtalk-id。--receiver 始终按 userId 通过通讯录关键词搜索做精确匹配，即使值以 D/d 开头也不会猜成 openDingTalkId；已有 openDingTalkId 时必须用显式参数直传。userId 包括在 --dry-run 时也会先解析。只传目标时创建卡片并返回 bizId，供后续 messages-update-card 流式更新；同时传 --content 时会自动串联创建和更新，默认以 flowStatus=3 完成，避免卡片停留在加载中。",
 	Risk:        shortcut.RiskWrite,
 	Flags: []shortcut.Flag{
 		{Name: "group", Type: shortcut.FlagString, Desc: "群 openConversationId（与两个单聊接收者参数互斥）"},
-		{Name: "receiver", Type: shortcut.FlagString, Desc: "单聊接收者 userId（与 --group/--receiver-open-dingtalk-id 互斥）；始终只读解析为 openDingTalkId，包括 --dry-run 和 D/d 开头的 userId"},
+		{Name: "receiver", Type: shortcut.FlagString, Desc: "单聊接收者 userId（与 --group/--receiver-open-dingtalk-id 互斥）；始终通过通讯录搜索精确匹配 openDingTalkId，包括 --dry-run 和 D/d 开头的 userId"},
 		{Name: "receiver-open-dingtalk-id", Type: shortcut.FlagString, Desc: "单聊接收者 openDingTalkId（与 --group/--receiver 互斥）；显式直传且不做通讯录解析"},
 		{Name: "content", Type: shortcut.FlagString, Desc: "创建后立即写入的卡片内容；省略时仅创建并返回 bizId"},
 		{Name: "flow-status", Type: shortcut.FlagInt, Default: "3", Desc: "自动更新状态：1处理中/2输入中/3完成/4执行中/5错误；--flow-status 必须在 1-5 之间，且显式指定时必须同时提供 --content"},

internal/shortcut/chat/coverage_completion_test.go

@@ -120,17 +120,17 @@ func TestCrossPlatformCoverageChatCreateAndReplyFailures(t *testing.T) {
 		},
 		{
 			name:      "explicit sender lookup",
-			caller:    &larkAlignmentCaller{failProductTool: "contact/get_user_info_by_user_ids"},
+			caller:    &larkAlignmentCaller{failProductTool: "contact/search_contact_by_key_word"},
 			args:      []string{"chat", "+messages-reply", "--conversation-id", "cid", "--message-id", "msg", "--ref-sender", "user-id", "--text", "收到", "--yes"},
 			wantError: "解析为 openDingTalkId",
 		},
 		{
 			name: "explicit sender unresolved",
 			caller: &larkAlignmentCaller{responses: map[string]string{
-				"contact/get_user_info_by_user_ids": `{"result":[]}`,
+				"contact/search_contact_by_key_word": `{"result":[]}`,
 			}},
 			args:      []string{"chat", "+messages-reply", "--conversation-id", "cid", "--message-id", "msg", "--ref-sender", "user-id", "--text", "收到", "--yes"},
-			wantError: "无法把 --ref-sender",
+			wantError: "没有精确匹配",
 		},
 		{
 			name:      "referenced message lookup",

internal/shortcut/chat/gap_fill_test.go

@@ -80,9 +80,12 @@ func TestCrossPlatformCoverageMessagesSendCurrentUserImageAndUserResolution(t *t
 	if len(fake.calls) != 2 {
 		t.Fatalf("calls = %#v, want contact resolution + send", fake.calls)
 	}
-	if fake.calls[0].product != "contact" || fake.calls[0].tool != "get_user_info_by_user_ids" {
+	if fake.calls[0].product != "contact" || fake.calls[0].tool != "search_contact_by_key_word" {
 		t.Fatalf("resolution call = %#v", fake.calls[0])
 	}
+	if got := fake.calls[0].args["keyword"]; got != "user-id" {
+		t.Fatalf("resolution keyword = %#v, want user-id", got)
+	}
 	send := fake.calls[1]
 	if send.product != "chat" || send.tool != "send_personal_message" {
 		t.Fatalf("send call = %#v", send)
@@ -284,9 +287,15 @@ func TestCrossPlatformCoverageMessagesSendUserResolutionFailures(t *testing.T) {
 		name string
 		fake *larkAlignmentCaller
 	}{
-		{"lookup error", &larkAlignmentCaller{failProductTool: "contact/get_user_info_by_user_ids"}},
+		{"lookup error", &larkAlignmentCaller{failProductTool: "contact/search_contact_by_key_word"}},
+		{"no exact user id", &larkAlignmentCaller{responses: map[string]string{
+			"contact/search_contact_by_key_word": `{"result":[{"userId":"other-user","openDingTalkId":"D-other"}]}`,
+		}}},
 		{"missing open id", &larkAlignmentCaller{responses: map[string]string{
-			"contact/get_user_info_by_user_ids": `{"result":[]}`,
+			"contact/search_contact_by_key_word": `{"result":[{"userId":"user-id","name":"Resolved User"}]}`,
+		}}},
+		{"ambiguous open id", &larkAlignmentCaller{responses: map[string]string{
+			"contact/search_contact_by_key_word": `{"result":[{"userId":"user-id","openDingTalkId":"D-one"},{"userId":"user-id","openDingTalkId":"D-two"}]}`,
 		}}},
 	} {
 		t.Run(tc.name, func(t *testing.T) {
@@ -394,7 +403,7 @@ func TestCrossPlatformCoverageMessagesSendCardResolvesReceiverForLowerTool(t *te
 	}
 	if len(fake.calls) != 2 ||
 		fake.calls[0].product != "contact" ||
-		fake.calls[0].tool != "get_user_info_by_user_ids" ||
+		fake.calls[0].tool != "search_contact_by_key_word" ||
 		fake.calls[1].product != "im" ||
 		fake.calls[1].tool != "create_and_send_card" {
 		t.Fatalf("card receiver calls = %#v", fake.calls)
@@ -402,7 +411,7 @@ func TestCrossPlatformCoverageMessagesSendCardResolvesReceiverForLowerTool(t *te
 	if got := fake.calls[1].args["receiverOpenDingTalkId"]; got != "D-resolved" {
 		t.Fatalf("receiverOpenDingTalkId = %#v, want D-resolved", got)
 	}
-	if got := fake.calls[0].args["user_id_list"]; !reflect.DeepEqual(got, []string{"d-user-id"}) {
+	if got := fake.calls[0].args["keyword"]; got != "d-user-id" {
 		t.Fatalf("D/d-prefixed userId resolution args = %#v", fake.calls[0].args)
 	}
 	if _, exists := fake.calls[1].args["receiverUid"]; exists {
@@ -469,7 +478,7 @@ func TestCrossPlatformCoverageMessagesSendCardDryRunAndFailureBoundaries(t *test
 		}
 		if len(fake.calls) != 1 ||
 			fake.calls[0].product != "contact" ||
-			fake.calls[0].tool != "get_user_info_by_user_ids" {
+			fake.calls[0].tool != "search_contact_by_key_word" {
 			t.Fatalf("card dry-run receiver resolution calls = %#v", fake.calls)
 		}
 		var payload map[string]any
@@ -486,7 +495,7 @@ func TestCrossPlatformCoverageMessagesSendCardDryRunAndFailureBoundaries(t *test
 	})
 
 	t.Run("receiver resolution error", func(t *testing.T) {
-		fake := &larkAlignmentCaller{failProductTool: "contact/get_user_info_by_user_ids"}
+		fake := &larkAlignmentCaller{failProductTool: "contact/search_contact_by_key_word"}
 		helpers.InitDeps(fake)
 		root := newPlatformCoverageRoot()
 		root.SetArgs([]string{
@@ -498,7 +507,7 @@ func TestCrossPlatformCoverageMessagesSendCardDryRunAndFailureBoundaries(t *test
 		if err == nil || !strings.Contains(err.Error(), "解析为 openDingTalkId 失败") {
 			t.Fatalf("receiver resolution error = %v", err)
 		}
-		if len(fake.calls) != 1 || fake.calls[0].tool != "get_user_info_by_user_ids" {
+		if len(fake.calls) != 1 || fake.calls[0].tool != "search_contact_by_key_word" {
 			t.Fatalf("receiver resolution calls = %#v", fake.calls)
 		}
 	})

internal/shortcut/chat/lark_alignment.go

@@ -139,13 +139,13 @@ var MessagesReply = shortcut.Shortcut{
 	Command:     "+messages-reply",
 	Product:     "chat",
 	Description: "以当前用户身份引用回复消息（自动补全原发送者）",
-	Intent:      "当你要以当前用户身份对已有消息发送纯文本引用回复时使用；提供会话和被引用消息即可，默认通过 mget 自动读取原发送者，也可显式传 openDingTalkId/userId。它不支持 bot 身份、富媒体、卡片或 thread 内回复。",
+	Intent:      "当你要以当前用户身份对已有消息发送纯文本引用回复时使用；提供会话和被引用消息即可，默认通过 mget 自动读取原发送者，也可显式传 openDingTalkId/userId；userId 会通过通讯录搜索精确匹配 openDingTalkId。它不支持 bot 身份、富媒体、卡片或 thread 内回复。",
 	Risk:        shortcut.RiskWrite,
 	Flags: []shortcut.Flag{
 		{Name: "conversation-id", Type: shortcut.FlagString, Desc: "会话 openConversationId", Required: true},
 		{Name: "ref-msg-id", Type: shortcut.FlagString, Desc: "被引用消息 openMessageId"},
 		{Name: "message-id", Type: shortcut.FlagString, Desc: "--ref-msg-id 的 lark-cli 对齐别名"},
-		{Name: "ref-sender", Type: shortcut.FlagString, Desc: "原消息发送者 openDingTalkId/userId（不传则自动读取）"},
+		{Name: "ref-sender", Type: shortcut.FlagString, Desc: "原消息发送者 openDingTalkId/userId（userId 通过通讯录搜索精确匹配；不传则自动读取）"},
 		{Name: "text", Type: shortcut.FlagString, Desc: "纯文本回复内容", Required: true},
 		{Name: "uuid", Type: shortcut.FlagString, Desc: "幂等键（可选）"},
 		{Name: "idempotency-key", Type: shortcut.FlagString, Desc: "--uuid 的 lark-cli 对齐别名"},
@@ -184,16 +184,11 @@ func resolveReplySender(rt *shortcut.RuntimeContext) (string, error) {
 		if isOpenID(value) {
 			return value, nil
 		}
-		data, err := rt.CallMCPData("contact", "get_user_info_by_user_ids", map[string]any{
+		openID, err := resolveUserOpenDingTalkID(rt, value)
-			"user_id_list": []string{value},
-		})
 		if err != nil {
 			return "", fmt.Errorf("把 --ref-sender userId 解析为 openDingTalkId 失败: %w", err)
 		}
-		if openID := findOpenDingTalkID(data); openID != "" {
+		return openID, nil
-			return openID, nil
-		}
-		return "", apperrors.NewValidation("无法把 --ref-sender 解析为 openDingTalkId")
 	}
 
 	messageID := replyMessageID(rt)

internal/shortcut/chat/lark_alignment_test.go

@@ -44,7 +44,24 @@ func (f *larkAlignmentCaller) CallTool(_ context.Context, product, tool string,
 	case "contact/get_current_user_profile":
 		text = `{"result":[{"orgEmployeeModel":{"userId":"self-user"}}]}`
 	case "contact/get_user_info_by_user_ids":
-		text = `{"result":[{"userId":"user-id","openDingTalkId":"D-resolved"}]}`
+		text = `{"result":[{"orgEmployeeModel":{"orgUserId":"user-id","orgUserName":"Resolved User"}}]}`
+	case "contact/search_contact_by_key_word":
+		keyword, _ := args["keyword"].(string)
+		payload, _ := json.Marshal(map[string]any{
+			"result": []map[string]any{
+				{
+					"name":           "Fuzzy Neighbor",
+					"userId":         keyword + "-other",
+					"openDingTalkId": "D-other",
+				},
+				{
+					"name":           "Resolved User",
+					"userId":         keyword,
+					"openDingTalkId": "D-resolved",
+				},
+			},
+		})
+		text = string(payload)
 	case "im/create_group_conversation":
 		text = `{"result":{"cid":"internal-cid","openCid":"open-cid"}}`
 	case "im/list_messages_by_ids":
@@ -403,7 +420,7 @@ func TestMessagesMgetDryRunPublishesMultiResourceDownloadPlan(t *testing.T) {
 	}
 }
 
-func TestMessagesReplyResolvesUserIDBeforeExecution(t *testing.T) {
+func TestCrossPlatformCoverageMessagesReplyResolvesUserIDBeforeExecution(t *testing.T) {
 	fake := &larkAlignmentCaller{}
 	helpers.InitDeps(fake)
 	root := newPlatformCoverageRoot()
@@ -420,7 +437,7 @@ func TestMessagesReplyResolvesUserIDBeforeExecution(t *testing.T) {
 	}
 	if len(fake.calls) != 2 ||
 		fake.calls[0].product != "contact" ||
-		fake.calls[0].tool != "get_user_info_by_user_ids" ||
+		fake.calls[0].tool != "search_contact_by_key_word" ||
 		fake.calls[1].tool != "send_personal_message" {
 		t.Fatalf("calls = %#v", fake.calls)
 	}

internal/shortcut/chat/unified_send.go

@@ -28,14 +28,14 @@ var MessagesSend = shortcut.Shortcut{
 	Command:     "+messages-send",
 	Product:     "chat",
 	Description: "统一发送文本、Markdown、当前用户文件或已有 mediaId 图片",
-	Intent:      "当你希望用同一个入口选择 current-user、bot 或 webhook 身份发送消息时使用；命令会按身份校验目标、内容和凭据并路由到真实下层。current-user 支持文本/Markdown、已有 mediaId 图片、安全相对路径文件上传和幂等键；--user 传 userId 时包括在 --dry-run 中也会先只读解析为 openDingTalkId。bot 支持群聊或批量单聊文本/Markdown；webhook 的目标由 token 所在群决定。不会把 user 文件能力伪装成 bot/webhook 等价能力。",
+	Intent:      "当你希望用同一个入口选择 current-user、bot 或 webhook 身份发送消息时使用；命令会按身份校验目标、内容和凭据并路由到真实下层。current-user 支持文本/Markdown、已有 mediaId 图片、安全相对路径文件上传和幂等键；--user 传 userId 时包括在 --dry-run 中也会先通过通讯录关键词搜索并按 userId 精确匹配 openDingTalkId。bot 支持群聊或批量单聊文本/Markdown；webhook 的目标由 token 所在群决定。不会把 user 文件能力伪装成 bot/webhook 等价能力。",
 	Risk:        shortcut.RiskWrite,
 	Flags: []shortcut.Flag{
 		{Name: "identity", Type: shortcut.FlagString, Default: "user", Enum: []string{"user", "bot", "webhook"}, Desc: "发送身份；目标、凭据和幂等参数受发送身份能力矩阵约束"},
 		{Name: "as", Type: shortcut.FlagString, Enum: []string{"user", "bot", "webhook"}, Desc: "--identity 的 lark-cli 对齐别名；受发送身份能力矩阵约束"},
 		{Name: "group", Type: shortcut.FlagString, Desc: "群 openConversationId（user/bot 群聊）；受发送身份能力矩阵约束"},
 		{Name: "chat-id", Type: shortcut.FlagString, Desc: "--group 的 lark-cli 对齐别名；受发送身份能力矩阵约束"},
-		{Name: "user", Type: shortcut.FlagString, Desc: "单聊接收者 userId（user；包括 --dry-run 也会先只读解析 openDingTalkId）；受发送身份能力矩阵约束"},
+		{Name: "user", Type: shortcut.FlagString, Desc: "单聊接收者 userId（user；包括 --dry-run 也会先通过通讯录搜索精确匹配 openDingTalkId）；受发送身份能力矩阵约束"},
 		{Name: "open-dingtalk-id", Type: shortcut.FlagString, Desc: "单聊接收者 openDingTalkId（user）；受发送身份能力矩阵约束"},
 		{Name: "users", Type: shortcut.FlagStringSlice, Desc: "批量单聊接收者 userId（bot）；受发送身份能力矩阵约束"},
 		{Name: "open-dingtalk-ids", Type: shortcut.FlagStringSlice, Desc: "批量单聊接收者 openDingTalkId（bot）；受发送身份能力矩阵约束"},
@@ -368,17 +368,43 @@ func messagesSendUserTarget(rt *shortcut.RuntimeContext) (group, openID string,
 
 func resolveUserOpenDingTalkID(rt *shortcut.RuntimeContext, userID string) (string, error) {
 	userID = strings.TrimSpace(userID)
-	data, err := rt.CallMCPData("contact", "get_user_info_by_user_ids", map[string]any{
+	data, err := rt.CallMCPData("contact", "search_contact_by_key_word", map[string]any{
-		"user_id_list": []string{userID},
+		"keyword": userID,
 	})
 	if err != nil {
-		return "", fmt.Errorf("把 userId 解析为 openDingTalkId 失败: %w", err)
+		return "", fmt.Errorf("通过通讯录搜索把 userId %q 解析为 openDingTalkId 失败: %w", userID, err)
 	}
-	openID := findOpenDingTalkID(data)
+
-	if openID == "" {
+	exactRows := 0
-		return "", apperrors.NewValidation("无法把 userId 解析为 openDingTalkId；请改用当前命令提供的显式 openDingTalkId 参数")
+	openIDs := make([]string, 0, 1)
+	for _, user := range shortcutMapSlice(data["result"]) {
+		if shortcutString(user, "userId", "userID") != userID {
+			continue
+		}
+		exactRows++
+		if openID := shortcutString(user, "openDingTalkId", "openDingtalkId"); openID != "" {
+			openIDs = appendUniqueShortcutString(openIDs, openID)
+		}
+	}
+	switch {
+	case len(openIDs) == 1:
+		return openIDs[0], nil
+	case len(openIDs) > 1:
+		return "", apperrors.NewValidation(fmt.Sprintf(
+			"通讯录中 userId %q 精确匹配到多个不同的 openDingTalkId，无法安全选择",
+			userID,
+		))
+	case exactRows > 0:
+		return "", apperrors.NewValidation(fmt.Sprintf(
+			"通讯录已精确匹配 userId %q，但结果未返回 openDingTalkId",
+			userID,
+		))
+	default:
+		return "", apperrors.NewValidation(fmt.Sprintf(
+			"通讯录搜索结果中没有精确匹配的 userId %q",
+			userID,
+		))
 	}
-	return openID, nil
 }
 
 func executeMessagesSendUserFile(

skills/mono/references/intent-guide.md

@@ -273,8 +273,8 @@ alidocs 链接表面长得一样（`https://alidocs.dingtalk.com/i/nodes/{id}`
 - "帮我在群里发个消息提醒大家" — **个人身份**发群消息
 - "发个单聊消息给某人" — 个人身份发单聊：
   - 已有 userId 时直接使用 `--user`；已有 openDingTalkId 时使用 `--open-dingtalk-id`
-  - 纯文本/Markdown 单聊传 `--user` 时直接走 userId 发送能力，不需要先手动查询 openDingTalkId
+  - `+messages-send --user` 对所有内容类型都会通过通讯录关键词搜索并按 userId 精确匹配 openDingTalkId；无需手动预查，`--dry-run` 也会执行这次只读解析
-  - 富媒体消息（image/file）单聊优先使用 `--open-dingtalk-id`；传 `--user` 时 CLI 会尝试解析为 openDingTalkId 后发送
+  - 已持有 openDingTalkId 时优先使用显式 `--open-dingtalk-id`，避免额外解析
 - "发本地图片/文件/语音/视频到群里" — `dws chat message send ... --msg-type file --file-path <本地路径>`，CLI 内部自动上传并发送；png/jpg/pdf/mp4/zip 等任意扩展名都走这条，接收方看到的是可下载的文件附件。图片不会内联渲染，也不会生成 mediaId
 - "用已有 mediaId 发内联图片" — 仅当上游已经提供有效 mediaId 时，使用 `dws chat message send ... --msg-type image --media-id <mediaId>`；DWS CLI 不能把本地图片转换成 mediaId
 - "发图片+文字说明" — 不要硬塞进一条命令；先发图片/文件消息再补一条 `--text "..."` 即可
@@ -285,13 +285,13 @@ dws chat +messages-send --as user --open-dingtalk-id <openDingTalkId> --msg-type
 ```
 
 > ❌ 反模式：先把本地文件转换成 mediaId，或先经钉盘上传再拼装聊天消息。本地图片/文件一律用 `--msg-type file --file` 直发；`--msg-type image --media-id` 只接受上游已经提供的有效 mediaId。
-> 富媒体消息单聊优先使用 `--open-dingtalk-id`；传 `--user` 时 CLI 会尝试解析为 openDingTalkId 后发送。
+> 单聊已持有 openDingTalkId 时优先使用 `--open-dingtalk-id`；传 `--user` 时 CLI 会通过通讯录关键词搜索做 userId 精确匹配后发送。
 
 `+messages-send` 的 @ 占位符按身份自动规范化并补齐：user 使用 `<@id>` / `<@all>`；bot/webhook 使用 `@id` / `@手机号` / `@all`。声明 `--at-*` / `--at-all` 即可，不要再手工拼 `@10`。
 
 **用 `chat +messages-send-card` 的场景**：
 - 群聊流式卡片使用 `--group <openConversationId>`。
-- 单聊已有 userId 时使用 `--receiver <userId>`，CLI 始终先只读解析为 openDingTalkId；即使 userId 以 D/d 开头也不会猜测类型，`--dry-run` 也会执行该解析。
+- 单聊已有 userId 时使用 `--receiver <userId>`，CLI 始终通过通讯录关键词搜索并按 userId 精确匹配 openDingTalkId；即使 userId 以 D/d 开头也不会猜测类型，`--dry-run` 也会执行该解析。
 - 单聊已有 openDingTalkId 时必须显式使用 `--receiver-open-dingtalk-id <openDingTalkId>`，避免与 userId 混淆。
 - `--group`、`--receiver`、`--receiver-open-dingtalk-id` 严格三选一；传 `--content` 可在同一次调用中创建并结束卡片。
 

skills/multi/dingtalk-chat/SKILL.md

@@ -130,7 +130,8 @@ metadata:
 - 查询结果中的 `resourceRefs` 是可继续执行的资源上下文。引用、回复或合并转发中的子消息必须使用该子消息返回的 `messageId`；子消息缺会话 ID 时才继承父消息的 `openConversationId`。
 - 上述五个查询 Shortcut 与 `+messages-resource-download` 都沿用安全本地下载的 `read/not_required` 契约，不应添加 `--yes` 或触发交互确认。下载只允许工作目录内相对路径、默认不覆盖并原子落盘；需要覆盖时必须由用户显式传 `--overwrite`。
 - `+messages-send` 会按身份规范化 @ 占位符：user 使用 `<@id>` / `<@all>`，bot/webhook 使用 `@id` / `@手机号` / `@all`；只需声明 `--at-*` / `--at-all`，缺失占位符会自动补齐。
-- 流式卡片优先用 `+messages-send-card`：群聊传 `--group`，单聊 userId 传 `--receiver` 并由 CLI 只读解析，已有 openDingTalkId 则必须显式传 `--receiver-open-dingtalk-id`；三者严格三选一，禁止根据首字母猜身份类型。
+- `+messages-send --as user --user <userId>` 会通过通讯录关键词搜索并按 userId 精确匹配 openDingTalkId，所有内容类型与 `--dry-run` 都使用同一只读解析链路；已有 openDingTalkId 时直接传 `--open-dingtalk-id`。
+- 流式卡片优先用 `+messages-send-card`：群聊传 `--group`，单聊 userId 传 `--receiver` 并由 CLI 通过通讯录关键词搜索做 userId 精确匹配，已有 openDingTalkId 则必须显式传 `--receiver-open-dingtalk-id`；三者严格三选一，禁止根据首字母猜身份类型。
 
 ## 意图表
 

skills/multi/dws-shared/references/intent-guide.md

@@ -288,8 +288,8 @@ alidocs 链接表面长得一样（`https://alidocs.dingtalk.com/i/nodes/{id}`
 - "帮我在群里发个消息提醒大家" — **个人身份**发群消息
 - "发个单聊消息给某人" — 个人身份发单聊：
   - 已有 userId 时直接使用 `--user`；已有 openDingTalkId 时使用 `--open-dingtalk-id`
-  - 纯文本/Markdown 单聊传 `--user` 时直接走 userId 发送能力，不需要先手动查询 openDingTalkId
+  - `+messages-send --user` 对所有内容类型都会通过通讯录关键词搜索并按 userId 精确匹配 openDingTalkId；无需手动预查，`--dry-run` 也会执行这次只读解析
-  - 富媒体消息（image/file）单聊优先使用 `--open-dingtalk-id`；传 `--user` 时 CLI 会尝试解析为 openDingTalkId 后发送
+  - 已持有 openDingTalkId 时优先使用显式 `--open-dingtalk-id`，避免额外解析
 - "发张图片/截图/语音/视频/文件到群里" / "发张图给某某" — **统一一条命令**：`dws chat message send ... --msg-type file --file-path <本地路径>`，CLI 内部自动上传并发送，**任意扩展名（png/jpg/pdf/mp4/zip…）都走这条**
 - "发图片+文字说明" — 不要硬塞进一条命令；先发文件消息再补一条 `--text "..."` 即可
 
@@ -299,11 +299,11 @@ dws chat message send --open-dingtalk-id <openDingTalkId> --msg-type file --file
 ```
 
 > ❌ 反模式：调 `dt_media_upload` / `extract_media_id.py` / `drive upload` / `drive download` 等前置工具再 `--msg-type image --media-id`。这是**旧链路**，仅当上游已持有 mediaId 才用；新场景一律 `--file-path` 直发，避免长链路与“空白图”现象。
-> 富媒体消息单聊优先使用 `--open-dingtalk-id`；传 `--user` 时 CLI 会尝试解析为 openDingTalkId 后发送。
+> 单聊已持有 openDingTalkId 时优先使用 `--open-dingtalk-id`；传 `--user` 时 CLI 会通过通讯录关键词搜索做 userId 精确匹配后发送。
 
 **用 `chat +messages-send-card` 的场景**：
 - 群聊流式卡片使用 `--group <openConversationId>`。
-- 单聊已有 userId 时使用 `--receiver <userId>`，CLI 始终先只读解析为 openDingTalkId；即使 userId 以 D/d 开头也不会猜测类型，`--dry-run` 也会执行该解析。
+- 单聊已有 userId 时使用 `--receiver <userId>`，CLI 始终通过通讯录关键词搜索并按 userId 精确匹配 openDingTalkId；即使 userId 以 D/d 开头也不会猜测类型，`--dry-run` 也会执行该解析。
 - 单聊已有 openDingTalkId 时必须显式使用 `--receiver-open-dingtalk-id <openDingTalkId>`，避免与 userId 混淆。
 - `--group`、`--receiver`、`--receiver-open-dingtalk-id` 严格三选一；传 `--content` 可在同一次调用中创建并结束卡片。