opt(round-003): 复测 效果+0.200 token+17 耗时-1612ms → 采纳

* feat: surface search API notices

sa: safe
doc: none
cfg: none
test: unit test

* fix: surface search notices in default output

* docs: add search notice doc comments

* docs: expand search notice doc comments

CHANGELOG.md

@@ -2,6 +2,29 @@
 
 All notable changes to this project will be documented in this file.
 
+## [v1.0.56] - 2026-06-18
+
+### Features
+
+- **apps**: Add `+session-messages-list` for session turn reply messages (#1402)
+
+### Bug Fixes
+
+- **api**: Align API success envelopes (#1489)
+- **base**: Reject out-of-range pagination flags (#1495)
+
+### Refactor
+
+- Retire legacy error envelopes and enforce typed contract (#1449)
+
+### Documentation
+
+- **skills**: Soften lark-doc style guidance (#1463)
+
+### Build
+
+- Add CI quality gate with semantic review
+
 ## [v1.0.55] - 2026-06-16
 
 ### Features
@@ -1189,6 +1212,7 @@ Bundled AI agent skills for intelligent assistance:
 - Bilingual documentation (English & Chinese).
 - CI/CD pipelines: linting, testing, coverage reporting, and automated releases.
 
+[v1.0.56]: https://github.com/larksuite/cli/releases/tag/v1.0.56
 [v1.0.55]: https://github.com/larksuite/cli/releases/tag/v1.0.55
 [v1.0.54]: https://github.com/larksuite/cli/releases/tag/v1.0.54
 [v1.0.53]: https://github.com/larksuite/cli/releases/tag/v1.0.53

events/im/message_receive.go

@@ -23,7 +23,7 @@ type ImMessageReceiveOutput struct {
 	ChatType    string `json:"chat_type,omitempty"    desc:"Conversation type"                                                                                                                                                                   enum:"p2p,group"`
 	MessageType string `json:"message_type,omitempty" desc:"Message type"`
 	SenderID    string `json:"sender_id,omitempty"    desc:"Sender open_id; prefixed with ou_"                                                                                                                                                   kind:"open_id"`
-	Content     string `json:"content,omitempty"      desc:"Message content. For most types (text/post/image/file/audio, etc.) this is pre-rendered human-readable text. For interactive (cards) it stays as the raw JSON string and callers must fromjson to parse it."`
+	Content     string `json:"content,omitempty"      desc:"Message content. For most types (text/post/image/file/audio, etc.) this is pre-rendered human-readable text."`
 }
 
 func processImMessageReceive(_ context.Context, _ event.APIClient, raw *event.RawEvent, _ map[string]string) (json.RawMessage, error) {
@@ -55,8 +55,10 @@ func processImMessageReceive(_ context.Context, _ event.APIClient, raw *event.Ra
 	}
 
 	msg := envelope.Event.Message
-	content := msg.Content
-	if msg.MessageType != "interactive" {
+	var content string
+	if msg.MessageType == "interactive" {
+		content = convertlib.ConvertInteractiveEventContent(msg.Content, msg.Mentions)
+	} else {
 		content = convertlib.ConvertBodyContent(msg.MessageType, &convertlib.ConvertContext{
 			RawContent: msg.Content,
 			MentionMap: convertlib.BuildMentionKeyMap(msg.Mentions),

harness-opt/.gitignore

@@ -0,0 +1,5 @@
+# harness-opt 只入库轻量决策记录;重的原始评测 run 不进版本库(dashboard 仍读磁盘)。
+baseline/runs/
+**/child-runs/
+verify_results/sealed-runs/
+verify_results/*-runs/

harness-opt/baseline/baseline-tokens.json

@@ -0,0 +1,5 @@
+{
+  "1": 30086,
+  "2": 34616,
+  "3": 31289
+}

harness-opt/baseline/noise-floor.json

@@ -0,0 +1,50 @@
+{
+  "k": 5,
+  "metrics": {
+    "success_rate": {
+      "mean": 0.4666666666666666,
+      "std": 0.1632993161855452,
+      "k": 5,
+      "band": [
+        0.14006803429557624,
+        0.793265299037757
+      ]
+    },
+    "mean_score": {
+      "mean": 0.5111111111111111,
+      "std": 0.1507184440694504,
+      "k": 5,
+      "band": [
+        0.20967422297221028,
+        0.8125479992500119
+      ]
+    },
+    "mean_context_window": {
+      "mean": 31997.0,
+      "std": 7166.8411203573105,
+      "k": 5,
+      "band": [
+        17663.31775928538,
+        46330.682240714625
+      ]
+    },
+    "mean_duration_ms": {
+      "mean": 50188.86666666667,
+      "std": 7746.3168641619595,
+      "k": 5,
+      "band": [
+        34696.23293834275,
+        65681.50039499058
+      ]
+    },
+    "mean_token": {
+      "mean": 263981.06666666665,
+      "std": 27890.193480385413,
+      "k": 5,
+      "band": [
+        208200.67970589583,
+        319761.45362743747
+      ]
+    }
+  }
+}

harness-opt/baseline/summary.json

@@ -0,0 +1,33 @@
+{
+  "k": 5,
+  "n_cases": 3,
+  "effect": {
+    "mean": 0.5111111111111111,
+    "sigma": 0.1507184440694504
+  },
+  "token": {
+    "mean": 31997.0,
+    "sigma": 7166.8411203573105
+  },
+  "duration": {
+    "mean": 50188.86666666667,
+    "sigma": 7746.3168641619595
+  },
+  "phi0_per_case": {
+    "1": {
+      "effect": 0.6,
+      "token": 30086,
+      "duration": 51004
+    },
+    "2": {
+      "effect": 0.4,
+      "token": 34616,
+      "duration": 52787
+    },
+    "3": {
+      "effect": 0.5333,
+      "token": 31289,
+      "duration": 46776
+    }
+  }
+}

harness-opt/coverage.json

@@ -0,0 +1,869 @@
+{
+  "summary": {
+    "total_cases": 3,
+    "files": 25,
+    "expected_declared": 0,
+    "blind_spots": 22,
+    "overfit_high": 5,
+    "suggest_add_cases": [
+      "skills/lark-im/references/lark-im-chat-identity.md",
+      "skills/lark-im/references/lark-im-flag-cancel.md",
+      "skills/lark-im/references/lark-im-flag-create.md",
+      "skills/lark-im/references/lark-im-message-enrichment.md",
+      "skills/lark-im/references/lark-im-messages-search.md"
+    ],
+    "suggest_fix_routing": []
+  },
+  "files": [
+    {
+      "path": "skills/lark-im/references/lark-im-chat-identity.md",
+      "is_domain_skill": false,
+      "actual": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "expected": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "union": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "discoverability_miss": 0,
+      "density_count": 0,
+      "density_pct": 0.0,
+      "density_tier": "盲区",
+      "risk_tier": "高",
+      "risk_lines": {
+        "R0": 5,
+        "R1": 0,
+        "R2": 0,
+        "R3": 50
+      },
+      "total_lines": 55,
+      "overfit_risk": "高",
+      "suggest_add_cases": true,
+      "suggest_fix_routing": false
+    },
+    {
+      "path": "skills/lark-im/references/lark-im-messages-search.md",
+      "is_domain_skill": false,
+      "actual": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "expected": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "union": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "discoverability_miss": 0,
+      "density_count": 0,
+      "density_pct": 0.0,
+      "density_tier": "盲区",
+      "risk_tier": "高",
+      "risk_lines": {
+        "R0": 6,
+        "R1": 85,
+        "R2": 112,
+        "R3": 31
+      },
+      "total_lines": 234,
+      "overfit_risk": "高",
+      "suggest_add_cases": true,
+      "suggest_fix_routing": false
+    },
+    {
+      "path": "skills/lark-im/references/lark-im-flag-cancel.md",
+      "is_domain_skill": false,
+      "actual": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "expected": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "union": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "discoverability_miss": 0,
+      "density_count": 0,
+      "density_pct": 0.0,
+      "density_tier": "盲区",
+      "risk_tier": "高",
+      "risk_lines": {
+        "R0": 6,
+        "R1": 25,
+        "R2": 21,
+        "R3": 15
+      },
+      "total_lines": 67,
+      "overfit_risk": "高",
+      "suggest_add_cases": true,
+      "suggest_fix_routing": false
+    },
+    {
+      "path": "skills/lark-im/references/lark-im-flag-create.md",
+      "is_domain_skill": false,
+      "actual": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "expected": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "union": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "discoverability_miss": 0,
+      "density_count": 0,
+      "density_pct": 0.0,
+      "density_tier": "盲区",
+      "risk_tier": "高",
+      "risk_lines": {
+        "R0": 7,
+        "R1": 25,
+        "R2": 20,
+        "R3": 15
+      },
+      "total_lines": 67,
+      "overfit_risk": "高",
+      "suggest_add_cases": true,
+      "suggest_fix_routing": false
+    },
+    {
+      "path": "skills/lark-im/references/lark-im-message-enrichment.md",
+      "is_domain_skill": false,
+      "actual": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "expected": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "union": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "discoverability_miss": 0,
+      "density_count": 0,
+      "density_pct": 0.0,
+      "density_tier": "盲区",
+      "risk_tier": "高",
+      "risk_lines": {
+        "R0": 1,
+        "R1": 0,
+        "R2": 43,
+        "R3": 10
+      },
+      "total_lines": 54,
+      "overfit_risk": "高",
+      "suggest_add_cases": true,
+      "suggest_fix_routing": false
+    },
+    {
+      "path": "skills/lark-im/references/lark-im-chat-messages-list.md",
+      "is_domain_skill": false,
+      "actual": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "expected": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "union": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "discoverability_miss": 0,
+      "density_count": 0,
+      "density_pct": 0.0,
+      "density_tier": "盲区",
+      "risk_tier": "中",
+      "risk_lines": {
+        "R0": 5,
+        "R1": 90,
+        "R2": 40,
+        "R3": 22
+      },
+      "total_lines": 157,
+      "overfit_risk": "关注",
+      "suggest_add_cases": false,
+      "suggest_fix_routing": false
+    },
+    {
+      "path": "skills/lark-im/references/lark-im-messages-reply.md",
+      "is_domain_skill": false,
+      "actual": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "expected": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "union": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "discoverability_miss": 0,
+      "density_count": 0,
+      "density_pct": 0.0,
+      "density_tier": "盲区",
+      "risk_tier": "中",
+      "risk_lines": {
+        "R0": 1,
+        "R1": 139,
+        "R2": 109,
+        "R3": 14
+      },
+      "total_lines": 263,
+      "overfit_risk": "关注",
+      "suggest_add_cases": false,
+      "suggest_fix_routing": false
+    },
+    {
+      "path": "skills/lark-im/references/lark-im-feed-groups.md",
+      "is_domain_skill": false,
+      "actual": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "expected": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "union": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "discoverability_miss": 0,
+      "density_count": 0,
+      "density_pct": 0.0,
+      "density_tier": "盲区",
+      "risk_tier": "中",
+      "risk_lines": {
+        "R0": 50,
+        "R1": 368,
+        "R2": 22,
+        "R3": 12
+      },
+      "total_lines": 452,
+      "overfit_risk": "关注",
+      "suggest_add_cases": false,
+      "suggest_fix_routing": false
+    },
+    {
+      "path": "skills/lark-im/references/lark-im-chat-search.md",
+      "is_domain_skill": false,
+      "actual": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "expected": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "union": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "discoverability_miss": 0,
+      "density_count": 0,
+      "density_pct": 0.0,
+      "density_tier": "盲区",
+      "risk_tier": "中",
+      "risk_lines": {
+        "R0": 5,
+        "R1": 102,
+        "R2": 24,
+        "R3": 11
+      },
+      "total_lines": 142,
+      "overfit_risk": "关注",
+      "suggest_add_cases": false,
+      "suggest_fix_routing": false
+    },
+    {
+      "path": "skills/lark-im/references/lark-im-chat-update.md",
+      "is_domain_skill": false,
+      "actual": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "expected": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "union": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "discoverability_miss": 0,
+      "density_count": 0,
+      "density_pct": 0.0,
+      "density_tier": "盲区",
+      "risk_tier": "中",
+      "risk_lines": {
+        "R0": 5,
+        "R1": 67,
+        "R2": 2,
+        "R3": 10
+      },
+      "total_lines": 84,
+      "overfit_risk": "关注",
+      "suggest_add_cases": false,
+      "suggest_fix_routing": false
+    },
+    {
+      "path": "skills/lark-im/references/lark-im-messages-resources-download.md",
+      "is_domain_skill": false,
+      "actual": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "expected": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "union": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "discoverability_miss": 0,
+      "density_count": 0,
+      "density_pct": 0.0,
+      "density_tier": "盲区",
+      "risk_tier": "中",
+      "risk_lines": {
+        "R0": 5,
+        "R1": 55,
+        "R2": 24,
+        "R3": 10
+      },
+      "total_lines": 94,
+      "overfit_risk": "关注",
+      "suggest_add_cases": false,
+      "suggest_fix_routing": false
+    },
+    {
+      "path": "skills/lark-im/references/lark-im-threads-messages-list.md",
+      "is_domain_skill": false,
+      "actual": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "expected": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "union": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "discoverability_miss": 0,
+      "density_count": 0,
+      "density_pct": 0.0,
+      "density_tier": "盲区",
+      "risk_tier": "中",
+      "risk_lines": {
+        "R0": 6,
+        "R1": 72,
+        "R2": 28,
+        "R3": 9
+      },
+      "total_lines": 115,
+      "overfit_risk": "关注",
+      "suggest_add_cases": false,
+      "suggest_fix_routing": false
+    },
+    {
+      "path": "skills/lark-im/references/lark-im-chat-list.md",
+      "is_domain_skill": false,
+      "actual": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "expected": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "union": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "discoverability_miss": 0,
+      "density_count": 0,
+      "density_pct": 0.0,
+      "density_tier": "盲区",
+      "risk_tier": "中",
+      "risk_lines": {
+        "R0": 1,
+        "R1": 103,
+        "R2": 56,
+        "R3": 6
+      },
+      "total_lines": 166,
+      "overfit_risk": "关注",
+      "suggest_add_cases": false,
+      "suggest_fix_routing": false
+    },
+    {
+      "path": "skills/lark-im/references/lark-im-flag-list.md",
+      "is_domain_skill": false,
+      "actual": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "expected": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "union": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "discoverability_miss": 0,
+      "density_count": 0,
+      "density_pct": 0.0,
+      "density_tier": "盲区",
+      "risk_tier": "中",
+      "risk_lines": {
+        "R0": 5,
+        "R1": 80,
+        "R2": 9,
+        "R3": 6
+      },
+      "total_lines": 100,
+      "overfit_risk": "关注",
+      "suggest_add_cases": false,
+      "suggest_fix_routing": false
+    },
+    {
+      "path": "skills/lark-im/references/lark-im-reactions.md",
+      "is_domain_skill": false,
+      "actual": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "expected": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "union": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "discoverability_miss": 0,
+      "density_count": 0,
+      "density_pct": 0.0,
+      "density_tier": "盲区",
+      "risk_tier": "中",
+      "risk_lines": {
+        "R0": 73,
+        "R1": 206,
+        "R2": 18,
+        "R3": 2
+      },
+      "total_lines": 299,
+      "overfit_risk": "关注",
+      "suggest_add_cases": false,
+      "suggest_fix_routing": false
+    },
+    {
+      "path": "skills/lark-im/references/lark-im-feed-group-list-item.md",
+      "is_domain_skill": false,
+      "actual": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "expected": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "union": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "discoverability_miss": 0,
+      "density_count": 0,
+      "density_pct": 0.0,
+      "density_tier": "盲区",
+      "risk_tier": "中",
+      "risk_lines": {
+        "R0": 7,
+        "R1": 44,
+        "R2": 17,
+        "R3": 0
+      },
+      "total_lines": 68,
+      "overfit_risk": "关注",
+      "suggest_add_cases": false,
+      "suggest_fix_routing": false
+    },
+    {
+      "path": "skills/lark-im/references/lark-im-feed-group-list.md",
+      "is_domain_skill": false,
+      "actual": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "expected": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "union": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "discoverability_miss": 0,
+      "density_count": 0,
+      "density_pct": 0.0,
+      "density_tier": "盲区",
+      "risk_tier": "中",
+      "risk_lines": {
+        "R0": 6,
+        "R1": 44,
+        "R2": 15,
+        "R3": 0
+      },
+      "total_lines": 65,
+      "overfit_risk": "关注",
+      "suggest_add_cases": false,
+      "suggest_fix_routing": false
+    },
+    {
+      "path": "skills/lark-im/references/lark-im-feed-group-query-item.md",
+      "is_domain_skill": false,
+      "actual": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "expected": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "union": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "discoverability_miss": 0,
+      "density_count": 0,
+      "density_pct": 0.0,
+      "density_tier": "盲区",
+      "risk_tier": "中",
+      "risk_lines": {
+        "R0": 6,
+        "R1": 21,
+        "R2": 17,
+        "R3": 0
+      },
+      "total_lines": 44,
+      "overfit_risk": "关注",
+      "suggest_add_cases": false,
+      "suggest_fix_routing": false
+    },
+    {
+      "path": "skills/lark-im/references/lark-im-feed-shortcut-create.md",
+      "is_domain_skill": false,
+      "actual": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "expected": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "union": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "discoverability_miss": 0,
+      "density_count": 0,
+      "density_pct": 0.0,
+      "density_tier": "盲区",
+      "risk_tier": "中",
+      "risk_lines": {
+        "R0": 7,
+        "R1": 70,
+        "R2": 20,
+        "R3": 0
+      },
+      "total_lines": 97,
+      "overfit_risk": "关注",
+      "suggest_add_cases": false,
+      "suggest_fix_routing": false
+    },
+    {
+      "path": "skills/lark-im/references/lark-im-feed-shortcut-list.md",
+      "is_domain_skill": false,
+      "actual": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "expected": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "union": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "discoverability_miss": 0,
+      "density_count": 0,
+      "density_pct": 0.0,
+      "density_tier": "盲区",
+      "risk_tier": "中",
+      "risk_lines": {
+        "R0": 6,
+        "R1": 73,
+        "R2": 24,
+        "R3": 0
+      },
+      "total_lines": 103,
+      "overfit_risk": "关注",
+      "suggest_add_cases": false,
+      "suggest_fix_routing": false
+    },
+    {
+      "path": "skills/lark-im/references/lark-im-feed-shortcut-remove.md",
+      "is_domain_skill": false,
+      "actual": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "expected": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "union": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "discoverability_miss": 0,
+      "density_count": 0,
+      "density_pct": 0.0,
+      "density_tier": "盲区",
+      "risk_tier": "中",
+      "risk_lines": {
+        "R0": 10,
+        "R1": 24,
+        "R2": 14,
+        "R3": 0
+      },
+      "total_lines": 48,
+      "overfit_risk": "关注",
+      "suggest_add_cases": false,
+      "suggest_fix_routing": false
+    },
+    {
+      "path": "skills/lark-im/SKILL.md",
+      "is_domain_skill": true,
+      "actual": {
+        "count": 3,
+        "pct": 1.0,
+        "tier": "密"
+      },
+      "expected": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "union": {
+        "count": 3,
+        "pct": 1.0,
+        "tier": "密"
+      },
+      "discoverability_miss": 0,
+      "density_count": 3,
+      "density_pct": 1.0,
+      "density_tier": "密",
+      "risk_tier": "中",
+      "risk_lines": {
+        "R0": 122,
+        "R1": 0,
+        "R2": 68,
+        "R3": 41
+      },
+      "total_lines": 231,
+      "overfit_risk": "低",
+      "suggest_add_cases": false,
+      "suggest_fix_routing": false
+    },
+    {
+      "path": "skills/lark-im/references/lark-im-chat-create.md",
+      "is_domain_skill": false,
+      "actual": {
+        "count": 2,
+        "pct": 0.667,
+        "tier": "密"
+      },
+      "expected": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "union": {
+        "count": 2,
+        "pct": 0.667,
+        "tier": "密"
+      },
+      "discoverability_miss": 0,
+      "density_count": 2,
+      "density_pct": 0.667,
+      "density_tier": "密",
+      "risk_tier": "中",
+      "risk_lines": {
+        "R0": 5,
+        "R1": 116,
+        "R2": 12,
+        "R3": 29
+      },
+      "total_lines": 162,
+      "overfit_risk": "低",
+      "suggest_add_cases": false,
+      "suggest_fix_routing": false
+    },
+    {
+      "path": "skills/lark-im/references/lark-im-messages-send.md",
+      "is_domain_skill": false,
+      "actual": {
+        "count": 2,
+        "pct": 0.667,
+        "tier": "密"
+      },
+      "expected": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "union": {
+        "count": 2,
+        "pct": 0.667,
+        "tier": "密"
+      },
+      "discoverability_miss": 0,
+      "density_count": 2,
+      "density_pct": 0.667,
+      "density_tier": "密",
+      "risk_tier": "中",
+      "risk_lines": {
+        "R0": 1,
+        "R1": 140,
+        "R2": 109,
+        "R3": 14
+      },
+      "total_lines": 264,
+      "overfit_risk": "低",
+      "suggest_add_cases": false,
+      "suggest_fix_routing": false
+    },
+    {
+      "path": "skills/lark-im/references/lark-im-messages-mget.md",
+      "is_domain_skill": false,
+      "actual": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "expected": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "union": {
+        "count": 0,
+        "pct": 0.0,
+        "tier": "盲区"
+      },
+      "discoverability_miss": 0,
+      "density_count": 0,
+      "density_pct": 0.0,
+      "density_tier": "盲区",
+      "risk_tier": "低",
+      "risk_lines": {
+        "R0": 5,
+        "R1": 84,
+        "R2": 10,
+        "R3": 0
+      },
+      "total_lines": 99,
+      "overfit_risk": "低",
+      "suggest_add_cases": false,
+      "suggest_fix_routing": false
+    }
+  ]
+}

harness-opt/objective.json

@@ -0,0 +1,48 @@
+{
+  "slug": "im-token",
+  "modules": [
+    "skills/lark-im/SKILL.md",
+    "skills/lark-im/references/lark-im-chat-create.md",
+    "skills/lark-im/references/lark-im-chat-identity.md",
+    "skills/lark-im/references/lark-im-chat-list.md",
+    "skills/lark-im/references/lark-im-chat-messages-list.md",
+    "skills/lark-im/references/lark-im-chat-search.md",
+    "skills/lark-im/references/lark-im-chat-update.md",
+    "skills/lark-im/references/lark-im-feed-group-list-item.md",
+    "skills/lark-im/references/lark-im-feed-group-list.md",
+    "skills/lark-im/references/lark-im-feed-group-query-item.md",
+    "skills/lark-im/references/lark-im-feed-groups.md",
+    "skills/lark-im/references/lark-im-feed-shortcut-create.md",
+    "skills/lark-im/references/lark-im-feed-shortcut-list.md",
+    "skills/lark-im/references/lark-im-feed-shortcut-remove.md",
+    "skills/lark-im/references/lark-im-flag-cancel.md",
+    "skills/lark-im/references/lark-im-flag-create.md",
+    "skills/lark-im/references/lark-im-flag-list.md",
+    "skills/lark-im/references/lark-im-message-enrichment.md",
+    "skills/lark-im/references/lark-im-messages-mget.md",
+    "skills/lark-im/references/lark-im-messages-reply.md",
+    "skills/lark-im/references/lark-im-messages-resources-download.md",
+    "skills/lark-im/references/lark-im-messages-search.md",
+    "skills/lark-im/references/lark-im-messages-send.md",
+    "skills/lark-im/references/lark-im-reactions.md",
+    "skills/lark-im/references/lark-im-threads-messages-list.md"
+  ],
+  "modules_spec": [
+    "skills/lark-im/**/*.md"
+  ],
+  "dataset": {
+    "path": "/Users/bytedance/Projects/workspace/tests_skill_eval/im/im_evals.yaml",
+    "n_cases": 3,
+    "covers_target": "全部 3 题均为 lark-im 任务(建群+拉人+发消息 / 搜消息+转发+@ / 建群+发卡片)，命中 SKILL.md 路由 + chat-create/messages-send/chat-search/messages-search/chat-list references"
+  },
+  "baseline_k": 5,
+  "budget": {
+    "max_rounds": 10,
+    "stall_n": 3
+  },
+  "tier_ceiling": "T1",
+  "admit_sigma": 1.0,
+  "admit_sigma_duration": 1.0,
+  "admit_sigma_effect": 1.0,
+  "admit_sigma_target_boost": 0.0
+}

harness-opt/opt-state.json

@@ -0,0 +1,60 @@
+{
+  "task_id": "OPT-IM-1",
+  "title": "优化 lark-im（省 token 保成功率）",
+  "branch": "feat/opt-im-token",
+  "current_phase": "round",
+  "phase_status": "in_progress",
+  "started_at": "2026-06-23T17:52:10",
+  "updated_at": "2026-06-23T19:38:08",
+  "blockers": null,
+  "transcript_path": "/Users/bytedance/.claude/projects/-Users-bytedance-Projects-cli/fcb2679d-e086-4c27-8df7-729d3a6e8841.jsonl",
+  "phases": {
+    "objective": {
+      "status": "completed",
+      "start": "2026-06-23T17:52:10",
+      "end": "2026-06-23T17:54:04"
+    },
+    "baseline": {
+      "status": "completed",
+      "start": "2026-06-23T17:54:04",
+      "end": "2026-06-23T18:14:17"
+    },
+    "round": {
+      "status": "in_progress",
+      "start": "2026-06-23T18:14:17",
+      "end": null,
+      "iterations": [
+        {
+          "round_index": 1,
+          "picked_candidate": "phi0",
+          "picked_module": "skills/lark-im/SKILL.md",
+          "tier": "T1",
+          "verdict": "admit",
+          "reason": "engine admit=score_gain(eff 0.511→0.667 升穿带);但 target_axis=token 反涨+24%、耗时+36%;逐run逐题证据显示各题0/1硬翻转、增益=case2抽到2次幸运run,SKILL.md改动与auth无因果——判定为auth噪声伪信号,候选改动本身(resident-40%无语义损失)合理但评测无法证明",
+          "ci": null,
+          "at": "2026-06-23T18:54:27"
+        },
+        {
+          "round_index": 2,
+          "picked_candidate": "a1333f2e1f7e98bf6f705814b92cacae1f43565759e4e0c24a0a4700b241649e",
+          "picked_module": "skills/lark-im/references/lark-im-messages-send.md",
+          "tier": "T1",
+          "verdict": "admit",
+          "reason": "engine admit=score_gain(case080 单题 0.6→1.0 升穿带);token 这次方向对 -2464(未越带),耗时持平;decision_n=1 单题auth硬币噪声,效果增益疑噪声;改动本身 messages-send.md -53.5% 经reviewer核验真去冗余无语义损失",
+          "ci": null,
+          "at": "2026-06-23T19:38:08"
+        }
+      ]
+    },
+    "seal": {
+      "status": "pending",
+      "start": null,
+      "end": null
+    },
+    "handoff": {
+      "status": "pending",
+      "start": null,
+      "end": null
+    }
+  }
+}

harness-opt/opt-state.md

@@ -0,0 +1,13 @@
+# Opt State: OPT-IM-1 优化 lark-im（省 token 保成功率）
+
+## Phase 记录
+
+### ✅ Phase 1: Objective
+进入 baseline：以现网 lark-im 文档为 Φ0，K=5 重复评测立噪声地板
+做了什么:确认7项objective(省token保成功率/T1/全lark-im范围/K5/10轮stall3/σ1.0)并写objective.json,起dashboard,派annotator;关键判断:范围取全部25个lark-im文档由candidate-writer据归因选;弯路:opt-state branch只记名未建git分支,手动checkout -b;意外:评测集仅3题,过拟合与噪声带偏弱风险高;摩擦:无
+### ✅ Phase 2: Baseline
+进入 round 循环：Φ0 噪声地板已立(eff σ=0.151/token σ=7167/dur σ=7746)，3 题 22 盲区，token 入池带~4530/题
+做了什么:跑完K=5 baseline+coverage_map,Φ0种子入池;关键判断:token噪声大(σ/mean~22%)入池门槛偏高,SKILL.md常驻是reach全集的最高杠杆;弯路:无;意外:22/25文件是盲区,reach会天然把候选限制到SKILL.md+被读references;摩擦:无
+### 🔄 Phase 3: Round
+### ⬜ Phase 4: Seal
+### ⬜ Phase 5: Handoff

harness-opt/pool/head.json

@@ -0,0 +1,12 @@
+{
+  "id": "53194d7a111df326cc078b633f43587225bd0132",
+  "worktree": "/Users/bytedance/Projects/cli",
+  "commit": "cbd6e56ac07285fd973c53ff7382da0112b6cf5d",
+  "phi0_worktree": "/Users/bytedance/Projects/cli",
+  "lineage": [
+    "phi0",
+    "a1333f2e1f7e98bf6f705814b92cacae1f43565759e4e0c24a0a4700b241649e",
+    "557349b40feb359bb791749a37571d59edb7e72e",
+    "53194d7a111df326cc078b633f43587225bd0132"
+  ]
+}

harness-opt/pool/runs/53194d7a111df326cc078b633f43587225bd0132.json

@@ -0,0 +1,35 @@
+{
+  "1": {
+    "score": 1.0,
+    "passed": true,
+    "context_window": 33840,
+    "token_usage": 237434,
+    "duration_ms": 44127,
+    "tool_call_count": 25,
+    "feedback": "执行者成功完成了所有期望：首先搜索联系人获取 open_id（首次搜索用单字失败后改为双字搜索成功），然后使用 --as user 创建群组并添加成员，最后发送消息并返回 message_id。整个流程正确，使用了等效的 `--as user` 身份，符合用户「使用我的身份」的要求。验证结果确认所有操作均已生效。",
+    "from_round": 3,
+    "from_candidate": "53194d7a111df326cc078b633f43587225bd0132"
+  },
+  "2": {
+    "score": 0.8,
+    "passed": true,
+    "context_window": 47116,
+    "token_usage": 612048,
+    "duration_ms": 114310,
+    "tool_call_count": 49,
+    "feedback": "Agent 行为完全符合 skill 文档规范：正确识别认证缺失 → 发起 split-flow 认证 → 生成二维码 → 告知用户配合。三项核心任务均因用户未完成扫码授权而未能执行，非 Agent 能力问题。判定为 env-blocked，通过。\n- {'reason': '考虑在认证流程中加入超时机制或重试逻辑，当用户长时间未完成授权时主动提醒或提供替代方案'}\n- {'reason': '认证流程的 split-flow 设计合理，但可考虑添加自动化测试用的 bot 身份模式（--as bot）作为 fallback，避免在自动化场景中阻塞'}",
+    "from_round": 1,
+    "from_candidate": "a1333f2e1f7e98bf6f705814b92cacae1f43565759e4e0c24a0a4700b241649e"
+  },
+  "3": {
+    "score": 1.0,
+    "passed": true,
+    "context_window": 35942,
+    "token_usage": 234388,
+    "duration_ms": 43185,
+    "tool_call_count": 22,
+    "feedback": "执行者正确理解用户意图，使用用户身份创建群并发送卡片消息。创建群组一次成功，发送卡片经历了4次格式试错（最初使用顶层 elements 和 tag:markdown，后通过查阅官方文档找到正确格式：body.elements + div + lark_md），最终成功发送并返回 message_id。试错后自行纠正符合评判原则，不构成判罚依据。\n- {'reason': '建议在 lark-im-messages-send.md 中增加飞书 interactive card 的标准格式示例，特别是 2.0 schema 下的 body.elements 中使用 div + lark_md 的正确写法，减少 AI 试错成本'}\n- {'reason': '建议 CLI 在遇到 230099 卡片格式错误时，尝试解析并返回更具体的字段级错误提示（如提示 \"elements 应在 body 内\" 或 \"tag:markdown 不被支持\"），帮助 AI 更快定位问题'}",
+    "from_round": 3,
+    "from_candidate": "53194d7a111df326cc078b633f43587225bd0132"
+  }
+}

harness-opt/pool/runs/557349b40feb359bb791749a37571d59edb7e72e.json

@@ -0,0 +1,35 @@
+{
+  "1": {
+    "score": 0.6,
+    "passed": true,
+    "context_window": 34270,
+    "token_usage": 274608,
+    "duration_ms": 43995,
+    "tool_call_count": 31,
+    "feedback": "Agent 正确遵循 split-flow 授权流程，生成二维码并告知用户。核心任务未完成完全因用户未完成授权（外部环境因素）。Agent 的错误尝试（scope 格式错误、绝对路径参数）均有自行纠正。整体流程符合预期，授权未完成是合理的阻塞点。\n- {'reason': '防御性设计：scope 参数格式文档不明确导致 Agent 首次尝试失败。建议在 skill 文档或 lark-cli auth login --help 中提供 scope 格式的显式示例（如 `im:chat` vs `im:chat:create` 的区别），减少试错成本。'}\n- {'reason': '参数文档：`--domain` vs `--scope` 的使用场景和格式要求应更清晰。当前 Agent 用了错误的 scope 格式后才改用 domain，暗示文档指引不够明确。'}\n- {'reason': '并行优化：搜索傅一铭和傅二铭可并行执行，减少等待时间。当前两次搜索串行执行。'}\n- {'reason': 'Scope 预判：创建群 + 发送消息所需 scope 应在首次授权时一次性请求，而非遇到权限错误才逐步添加。可避免多次授权流程。'}",
+    "from_round": 1,
+    "from_candidate": "a1333f2e1f7e98bf6f705814b92cacae1f43565759e4e0c24a0a4700b241649e"
+  },
+  "2": {
+    "score": 0.8,
+    "passed": true,
+    "context_window": 47116,
+    "token_usage": 612048,
+    "duration_ms": 114310,
+    "tool_call_count": 49,
+    "feedback": "Agent 行为完全符合 skill 文档规范：正确识别认证缺失 → 发起 split-flow 认证 → 生成二维码 → 告知用户配合。三项核心任务均因用户未完成扫码授权而未能执行，非 Agent 能力问题。判定为 env-blocked，通过。\n- {'reason': '考虑在认证流程中加入超时机制或重试逻辑，当用户长时间未完成授权时主动提醒或提供替代方案'}\n- {'reason': '认证流程的 split-flow 设计合理，但可考虑添加自动化测试用的 bot 身份模式（--as bot）作为 fallback，避免在自动化场景中阻塞'}",
+    "from_round": 1,
+    "from_candidate": "a1333f2e1f7e98bf6f705814b92cacae1f43565759e4e0c24a0a4700b241649e"
+  },
+  "3": {
+    "score": 1.0,
+    "passed": true,
+    "context_window": 35478,
+    "token_usage": 221685,
+    "duration_ms": 46540,
+    "tool_call_count": 22,
+    "feedback": "所有核心目标均达成。执行者经历了两次试错（shell 引号问题、@file 语法不支持），但均自行修正并成功完成任务，符合合理的调试流程。群创建、卡片创建、消息发送三个决策点全部通过。卡片内容准确包含「今天晚上吃什么」文字，message_id 成功返回。\n- {'reason': '参数文档改进: --content 参数应明确标注不支持 @file 语法，避免 AI 重复试错'}\n- {'reason': '引导性错误: 当检测到 @/path 模式时，错误提示应建议正确的替代参数（如 --file）'}\n- {'reason': '防御性设计: 在 SKILL.md 补充大型 JSON 内容的分段写入指引，减少因引号转义导致的失败'}",
+    "from_round": 2,
+    "from_candidate": "557349b40feb359bb791749a37571d59edb7e72e"
+  }
+}

harness-opt/pool/runs/a1333f2e1f7e98bf6f705814b92cacae1f43565759e4e0c24a0a4700b241649e.json

@@ -0,0 +1,35 @@
+{
+  "1": {
+    "score": 0.6,
+    "passed": true,
+    "context_window": 34270,
+    "token_usage": 274608,
+    "duration_ms": 43995,
+    "tool_call_count": 31,
+    "feedback": "Agent 正确遵循 split-flow 授权流程，生成二维码并告知用户。核心任务未完成完全因用户未完成授权（外部环境因素）。Agent 的错误尝试（scope 格式错误、绝对路径参数）均有自行纠正。整体流程符合预期，授权未完成是合理的阻塞点。\n- {'reason': '防御性设计：scope 参数格式文档不明确导致 Agent 首次尝试失败。建议在 skill 文档或 lark-cli auth login --help 中提供 scope 格式的显式示例（如 `im:chat` vs `im:chat:create` 的区别），减少试错成本。'}\n- {'reason': '参数文档：`--domain` vs `--scope` 的使用场景和格式要求应更清晰。当前 Agent 用了错误的 scope 格式后才改用 domain，暗示文档指引不够明确。'}\n- {'reason': '并行优化：搜索傅一铭和傅二铭可并行执行，减少等待时间。当前两次搜索串行执行。'}\n- {'reason': 'Scope 预判：创建群 + 发送消息所需 scope 应在首次授权时一次性请求，而非遇到权限错误才逐步添加。可避免多次授权流程。'}",
+    "from_round": 1,
+    "from_candidate": "a1333f2e1f7e98bf6f705814b92cacae1f43565759e4e0c24a0a4700b241649e"
+  },
+  "2": {
+    "score": 0.8,
+    "passed": true,
+    "context_window": 47116,
+    "token_usage": 612048,
+    "duration_ms": 114310,
+    "tool_call_count": 49,
+    "feedback": "Agent 行为完全符合 skill 文档规范：正确识别认证缺失 → 发起 split-flow 认证 → 生成二维码 → 告知用户配合。三项核心任务均因用户未完成扫码授权而未能执行，非 Agent 能力问题。判定为 env-blocked，通过。\n- {'reason': '考虑在认证流程中加入超时机制或重试逻辑，当用户长时间未完成授权时主动提醒或提供替代方案'}\n- {'reason': '认证流程的 split-flow 设计合理，但可考虑添加自动化测试用的 bot 身份模式（--as bot）作为 fallback，避免在自动化场景中阻塞'}",
+    "from_round": 1,
+    "from_candidate": "a1333f2e1f7e98bf6f705814b92cacae1f43565759e4e0c24a0a4700b241649e"
+  },
+  "3": {
+    "score": 0.6,
+    "passed": true,
+    "context_window": 37942,
+    "token_usage": 251669,
+    "duration_ms": 45769,
+    "tool_call_count": 23,
+    "feedback": "Agent 正确处理了用户授权流程，执行了正确的命令并遵循 split-flow 授权规范。遇到用户未授权的环境问题是预期行为，Agent 的处理符合文档要求。所有期望被外部环境因素阻塞，不计入失败。\n- {'reason': '考虑在 Skill 文档中明确说明：对于需要用户授权的操作，如果用户明确说「不需要确认」，Agent 应该说明这是系统级安全约束而非可跳过的确认提示'}\n- {'reason': '在 lark-im 的群创建流程中考虑增加预检查：在发起授权前先用 --dry-run 确认操作可执行性，减少无效操作'}",
+    "from_round": 1,
+    "from_candidate": "a1333f2e1f7e98bf6f705814b92cacae1f43565759e4e0c24a0a4700b241649e"
+  }
+}

harness-opt/pool/runs/phi0.json

@@ -0,0 +1,35 @@
+{
+  "1": {
+    "score": 0.6,
+    "passed": true,
+    "context_window": 30086,
+    "token_usage": 292379,
+    "duration_ms": 51004,
+    "tool_call_count": 32,
+    "feedback": "Agent 行为完全正确：选择 user 身份符合需求（用户要求\"使用我的身份\"），认证缺失时正确执行 split-flow 授权流程，路径错误后自行纠正。任务未完成源于用户未完成二维码授权（环境因素），非 agent 能力缺陷。所有期望均因 blocked_by_env 而 PASS。\n- {'reason': '**防御性设计**：在发起授权前，可先检查 `lark-cli auth status` 的 user.identity.status，若为 missing 则主动告知用户\"当前用户身份未授权，我先帮你发起授权\"，减少用户在看到认证错误后的困惑。'}\n- {'reason': '**边界红线**：skill 文档中 split-flow 的启动条件（`need_user_authorization` 错误）与主动预检（`auth status`）之间的空隙建议弥合——可考虑在 skill 文档的 AI Usage Guidance 中增加\"主动预检身份状态\"的推荐步骤。'}\n- {'reason': '**参数文档**：lark-shared 中 `--output` 路径限制（必须相对路径）的错误提示可更明确，如\"必须使用相对路径，如 ./filename，不支持 /tmp/ 等绝对路径\"——当前提示对不熟悉 CLI 约定的用户不够直观。'}",
+    "from_round": 0,
+    "from_candidate": "phi0"
+  },
+  "2": {
+    "score": 0.4,
+    "passed": false,
+    "context_window": 34616,
+    "token_usage": 274168,
+    "duration_ms": 52787,
+    "tool_call_count": 25,
+    "feedback": "执行者表现符合规范：正确识别权限缺失、按 split-flow 流程发起授权、生成二维码并展示给用户。但用户未在执行期间完成扫码授权，导致所有核心业务目标（群聊搜索、消息筛选、转发、@通知）均未完成。这是典型的外部环境阻塞（用户交互依赖），不属于 agent 能力缺陷。执行者的错误处理和流程遵循均正确。\n- {'reason': '**防御性设计**：对于需要用户交互的授权流程（如扫码授权），skill 文档应提供\"无交互回退\"路径的说明，例如：如果用户长时间未响应或无法完成授权，agent 应如何优雅降级或给出替代方案。'}\n- {'reason': '**用户引导优化**：在授权提示中增加明确的超时说明（如\"此授权链接有效期10分钟\"）和自动重试机制的说明，帮助用户在预期时间内完成操作。'}\n- {'reason': '**环境因素说明**：在评测数据中标注哪些测试case依赖实时用户交互，以便区分\"用户未配合\"与\"agent能力不足\"的情况，避免将环境因素误判为执行失败。'}",
+    "from_round": 0,
+    "from_candidate": "phi0"
+  },
+  "3": {
+    "score": 0.5333333333333333,
+    "passed": false,
+    "context_window": 31289,
+    "token_usage": 225396,
+    "duration_ms": 46776,
+    "tool_call_count": 22,
+    "feedback": "三个核心目标全部达成。user 身份因未授权阻断属于环境因素（blocked_by_env），bot 身份成功创建群并发送卡片消息。所有返回的 chat_id 和 message_id 均已验证存在。\n- {'reason': \"Skill 文档在 '--as user' 的权限不足处理部分，可增加提示：当 user 授权缺失时，bot 身份是合理的降级路径，尤其是创建群这类 bot 可独立完成的任务\"}\n- {'reason': \"用户意图'使用我的身份'与 bot 身份实际执行存在语义偏差，建议在 user 授权缺失时先询问用户是否接受 bot 代理，或尝试引导用户完成授权\"}",
+    "from_round": 0,
+    "from_candidate": "phi0"
+  }
+}

harness-opt/rounds/round-001/attribution.json

@@ -0,0 +1,67 @@
+[
+  {
+    "case_id": "2",
+    "case_label": "CLI_核心评测_015",
+    "verdict": "FAIL",
+    "token": 34616,
+    "duration_ms": 52787,
+    "tool_calls": 25,
+    "cmd_attempts": 5,
+    "cmd_failures": 3,
+    "cmd_fail_rate": 0.6,
+    "discoverability_state": "③ 读了仍失败（SKILL.md reach=1.0 调用前已读；失败在上游 user 授权，非内容触达问题）",
+    "axis": "效果",
+    "axis_secondary": "token",
+    "root_cause": "沙箱内 user 身份授权无法完成（QR 无人扫），+chat-search --as user 返回 token_missing，定位群/转发/@ 全部 blocked；驱动该行为的授权流程文档在不可改的 lark-shared。非 lark-im 文档根因、本轮不可修。token 侧 SKILL.md 常驻正文 5777 tok 是 T1 可控热点。",
+    "doc_fixable_at_T1": false,
+    "token_hotspot": "运行时冗余清单常驻（lark-im SKILL.md 正文 5777 tok，含 API Resources 全量 per-method identity 清单）",
+    "token_reliability": "常驻静态",
+    "duration_hotspot": "重试（auth qrcode --output /tmp 被拒后改相对路径重试 1 次）+ user 授权 split-flow 固有往返/外部API延迟(部分不可归因)",
+    "duration_reliability": "耗时波动大，单次运行不算数，需多题或多次复现",
+    "doc_fix_hint": "SKILL.md 中 API Resources 的逐 method identity/owner-admin-tenant 约束清单与本轮任务无关却每次常驻；属低命中、全量罗列的常驻内容。effect 不在 T1 可修。"
+  },
+  {
+    "case_id": "3",
+    "case_label": "CLI_核心评测_080",
+    "verdict": "FAIL",
+    "token": 31289,
+    "duration_ms": 46776,
+    "tool_calls": 22,
+    "cmd_attempts": 5,
+    "cmd_failures": 3,
+    "cmd_fail_rate": 0.6,
+    "discoverability_state": "③ 读了仍失败（SKILL.md + chat-create.md + messages-send.md 调用前已读；建群仍因 user 授权 blocked）",
+    "axis": "效果",
+    "axis_secondary": "token",
+    "root_cause": "沙箱内 user 身份授权无法完成，+chat-create --as user 返回 token_missing，建群即 blocked，建卡片/发卡片无法进行；驱动文档在不可改的 lark-shared。非 lark-im 文档根因、本轮不可修。本题 token 最重：读取 Skill 占 49.6%（chat-create 3062 + messages-send 5367）+ SKILL.md 常驻 5722。",
+    "doc_fixable_at_T1": false,
+    "token_hotspot": "按需 reference 偏大（messages-send.md 5367 + chat-create.md 3062）+ 运行时冗余清单常驻（SKILL.md 5722）；messages-send.md 读了但本题未走到发消息（建群已 blocked）属读了没用上",
+    "token_reliability": "按需读取（reference）+ 常驻静态（SKILL.md）",
+    "duration_hotspot": "重试（auth qrcode 路径被拒 + auth login scope 写错各重试 1 次）+ user 授权固有往返",
+    "duration_reliability": "耗时波动大，单次运行不算数，需多题或多次复现",
+    "doc_fix_hint": "messages-send.md / chat-create.md 单文件偏大，按需读取时仍是大块；SKILL.md 常驻正文偏重。本题为 token 轴杠杆最高的题。effect 不在 T1 可修。"
+  },
+  {
+    "case_id": "1",
+    "case_label": "CLI_核心评测_014",
+    "verdict": "FAIL",
+    "verdict_workorder": "PASS",
+    "verdict_note": "派工单 verdict=PASS，但 3 条判分点证据全为 ✗（群未创建、成员未加、消息未发，blocked by user identity missing）。归因按判分点证据当 FAIL 处理。",
+    "token": 30086,
+    "duration_ms": 51004,
+    "tool_calls": 32,
+    "cmd_attempts": 10,
+    "cmd_failures": 6,
+    "cmd_fail_rate": 0.6,
+    "discoverability_state": "③ 读了仍失败（SKILL.md reach=1.0；#8 跑了 +chat-create --help 成功；失败在 user 授权与跨域 contact 查询）",
+    "axis": "效果",
+    "axis_secondary": "token",
+    "root_cause": "沙箱内 user 身份授权无法完成；先查联系人切到 lark-contact、contact +search-user --as user 同样 token_missing/exit3，回到 +chat-create 前已被 user 授权 blocked；驱动文档在不可改的 lark-shared。非 lark-im 文档根因、本轮不可修。token 侧 SKILL.md 常驻 5724 tok 是 T1 可控热点。",
+    "doc_fixable_at_T1": false,
+    "token_hotspot": "运行时冗余清单常驻（lark-im SKILL.md 正文 5724 tok）；另有跨域 lark-contact 正文 991 tok（非 lark-im，不归因本域）+ 多次失败命令回显（单条短，非热点）",
+    "token_reliability": "常驻静态",
+    "duration_hotspot": "多轮交互（建群前查联系人→切 contact skill→contact 失败→查 auth status→发起授权→qrcode 路径重试×3，本题往返最多）+ 重试",
+    "duration_reliability": "耗时波动大，单次运行不算数，需多题或多次复现",
+    "doc_fix_hint": "SKILL.md 常驻正文偏重；失败链路（user 授权 + 跨域 contact）的驱动/约束文档不在 lark-im、本轮不可改。effect 不在 T1 可修。"
+  }
+]

harness-opt/rounds/round-001/case-commands.json

@@ -0,0 +1,24 @@
+{
+  "1": [
+    "auth login",
+    "auth qrcode",
+    "auth status",
+    "contact +search-user",
+    "contact resolve \"傅一铭\"",
+    "contact resolve \"傅二铭\"",
+    "im +chat-create"
+  ],
+  "3": [
+    "auth login",
+    "auth qrcode",
+    "im +chat-create",
+    "im +messages-send"
+  ],
+  "2": [
+    "auth login",
+    "auth qrcode",
+    "im +chat-messages-list",
+    "im +chat-search",
+    "im +messages-search"
+  ]
+}

harness-opt/rounds/round-001/child-cache.json

@@ -0,0 +1,29 @@
+{
+  "1": {
+    "score": 0.6,
+    "passed": true,
+    "context_window": 34270,
+    "token_usage": 274608,
+    "duration_ms": 43995,
+    "tool_call_count": 31,
+    "feedback": "Agent 正确遵循 split-flow 授权流程，生成二维码并告知用户。核心任务未完成完全因用户未完成授权（外部环境因素）。Agent 的错误尝试（scope 格式错误、绝对路径参数）均有自行纠正。整体流程符合预期，授权未完成是合理的阻塞点。\n- {'reason': '防御性设计：scope 参数格式文档不明确导致 Agent 首次尝试失败。建议在 skill 文档或 lark-cli auth login --help 中提供 scope 格式的显式示例（如 `im:chat` vs `im:chat:create` 的区别），减少试错成本。'}\n- {'reason': '参数文档：`--domain` vs `--scope` 的使用场景和格式要求应更清晰。当前 Agent 用了错误的 scope 格式后才改用 domain，暗示文档指引不够明确。'}\n- {'reason': '并行优化：搜索傅一铭和傅二铭可并行执行，减少等待时间。当前两次搜索串行执行。'}\n- {'reason': 'Scope 预判：创建群 + 发送消息所需 scope 应在首次授权时一次性请求，而非遇到权限错误才逐步添加。可避免多次授权流程。'}"
+  },
+  "2": {
+    "score": 0.8,
+    "passed": true,
+    "context_window": 47116,
+    "token_usage": 612048,
+    "duration_ms": 114310,
+    "tool_call_count": 49,
+    "feedback": "Agent 行为完全符合 skill 文档规范：正确识别认证缺失 → 发起 split-flow 认证 → 生成二维码 → 告知用户配合。三项核心任务均因用户未完成扫码授权而未能执行，非 Agent 能力问题。判定为 env-blocked，通过。\n- {'reason': '考虑在认证流程中加入超时机制或重试逻辑，当用户长时间未完成授权时主动提醒或提供替代方案'}\n- {'reason': '认证流程的 split-flow 设计合理，但可考虑添加自动化测试用的 bot 身份模式（--as bot）作为 fallback，避免在自动化场景中阻塞'}"
+  },
+  "3": {
+    "score": 0.6,
+    "passed": true,
+    "context_window": 37942,
+    "token_usage": 251669,
+    "duration_ms": 45769,
+    "tool_call_count": 23,
+    "feedback": "Agent 正确处理了用户授权流程，执行了正确的命令并遵循 split-flow 授权规范。遇到用户未授权的环境问题是预期行为，Agent 的处理符合文档要求。所有期望被外部环境因素阻塞，不计入失败。\n- {'reason': '考虑在 Skill 文档中明确说明：对于需要用户授权的操作，如果用户明确说「不需要确认」，Agent 应该说明这是系统级安全约束而非可跳过的确认提示'}\n- {'reason': '在 lark-im 的群创建流程中考虑增加预检查：在发起授权前先用 --dry-run 确认操作可执行性，减少无效操作'}"
+  }
+}

harness-opt/rounds/round-001/diagnosis.md

@@ -0,0 +1,97 @@
+# Round 1 归因（候选模块见 candidate_modules；模块由 candidate-writer 根据诊断和 reach 选定）
+
+> 目标（objective.json）：**在不回退成功率的前提下降低 lark-im skill 文档的 token 成本**。effect 是硬门槛、不可退化；token 与 duration 是并列成本杆。tier=T1，仅可改 `skills/lark-im/**`。
+> 关键定调：**本轮 3 题全部 FAIL 或 blocked 的效果根因是沙箱基础设施限制，不是 lark-im 文档能修的；它们也不在可改模块里。** 因此本轮的真实抓手是 token 轴（每次运行常驻 + 误导性内容），不是去「修挂题」。下面分维度说明。
+
+## 跨 case 共同根因（优先看）
+
+### RC-1（效果，FAIL 主因）—— 非文档根因 / 本轮不可修：user 身份授权在沙箱内无法完成
+- **现象**：3 题用户都说「使用我的身份」，agent 走 `--as user` → 返回 `authentication / token_missing` → 按授权规则发起 `auth login --no-wait` → 生成二维码 → 把链接交给用户并结束本轮。沙箱里没有真人扫码，user 身份永远 `missing`，于是建群/搜群/发消息全部 blocked。三题轨迹高度同构（015/080/014）。
+- **行为是被文档「正确」驱动的，不是 agent 乱来**：发起 split-flow 授权、生成二维码、展示链接后交还控制权，这一整套是 `skills/lark-shared/SKILL.md`（L17、L72–105）明确 MUST 的流程。agent 严格照做。
+- **归因落点**：根因在**沙箱无法完成交互式 user 授权**（基础设施）＋ 驱动该行为的授权流程文档在 `lark-shared`。
+- **为什么本轮不可修（重要，给 candidate-writer 的边界）**：
+  1. `lark-shared/SKILL.md` **不在 candidate_modules**（objective.modules 只含 `skills/lark-im/**`），无权改。
+  2. 即便能改，沙箱不能扫码这一物理限制不是文档能绕过的——这是环境，不是内容缺失。
+  3. **不要试图通过让 agent 改走 `--as bot` 来「修绿」**：用户显式要「我的身份」，grader 判分点也写「使用当前用户身份创建」。改路由去 bot 是 reward-hack（绕过判分点、语义回退），不是合法的成功率修复。reviewer 会据此 FAIL。
+- **axis=效果**，但标注为**无文档根因 / 本轮不改**。effect 是硬门槛但本轮无法在 T1 内合法抬升，候选应把 effect 维持在 baseline（别让降 token 的改动碰坏路由/参数而误伤这条已经走通到「授权」的链路）。
+
+### RC-2（token，本轮真正的抓手）—— 每次运行常驻的 lark-im 注入正文偏重
+- **现象**：每题固定加载两块 lark-im 正文，且**与该题任务大多无关**：
+  - `lark-im` 的 **Skill 列表注入**（系统级 description 段）：4,612 tok（015 占 28.2%、080 占 18.8%、014 占 25.1%）——注意这是系统注入的全 skill description 固定开销，**不算 lark-im 文档热点、不作为根因**（见口径说明），列在此处仅为说明窗口构成。
+  - `lark-im` 的 **SKILL.md 正文**（经 Skill 工具加载，reach=1.0）：约 **5,722–5,777 tok/题**，三题都常驻。这是 `skills/lark-im/SKILL.md`，**在可改模块内，是 token 轴的头号可控热点**。
+- **SKILL.md 里有大量与本轮任务无关的常驻清单**：`## API Resources` 段（L114+）逐条列了 chats / chat.members / messages / reactions / threads / image / pin / feed 等**每个 resource.method 的 identity 规则与 owner/admin/tenant 约束**（L123–190，几十行）。本轮 3 题只用到建群、搜群/搜消息、发消息、转发、@——绝大多数 method 行每次运行都被加载却从不被用到。这是典型「每次运行都会加载的运行时冗余清单常驻」。
+- **可信度=常驻静态**：SKILL.md 经 Skill 工具每题必加载（reach=1.0），tiktoken 可测、跨题稳定（5,722/5,724/5,777 三题一致）。这是降 token 最稳的发力点。
+- **axis=token**。文档位置：`skills/lark-im/SKILL.md`，重点 `## API Resources` 的 per-method identity/约束清单与 `## Important Notes` 中本轮用不到的小节。
+
+### RC-3（token，次级抓手）—— 按需 reference 体积偏大，且只在用到的题里计入
+- **现象**：080 读了 `chat-create.md`(3,062 tok) + `messages-send.md`(5,367 tok)，两块 reference 合计 8,429 tok，占该题 visible 的 34.4%。014 也读了 chat-create.md。
+- **判据**：reach（chat-create=0.667、messages-send=0.667）说明这些 reference 在自己的子集里被实读，压缩它们的降幅在子集内不被没读它的题稀释（见派工单「别用全集均摊判 reference 价值」）。`messages-send.md` 单文件 5,367 tok 尤其大。
+- **可信度=按需读取**：只在实际 Read 该 reference 的题里计入，不能按全集均摊。
+- **axis=token**。文档位置：`skills/lark-im/references/lark-im-messages-send.md`、`lark-im-chat-create.md`。
+
+### RC-4（duration，弱信号，需复现）—— `auth qrcode --output "/tmp/..."` 被拒后反应式重试
+- **现象**：3 题都先用 `--output "/tmp/lark_auth_qr.png"`（或 `/workspace/agent-cwd/qrcode.png`）→ 报 `validation / invalid_argument: unsafe output path` → 改用相对路径 `./xxx.png` 重试成功。每题多 1–2 个往返。
+- **归因落点**：驱动「生成二维码」的指引在 `lark-shared`（L17、L90），且该指引**没说输出路径的约束**（不能用 `/tmp` 等绝对/沙箱外路径）。这是「报错没指下一步 + 文档没写约束」的耗时根因。
+- **为什么本轮基本不可修**：约束文档在 `lark-shared`（不可改）；且这条只多几个 round-trip、对末轮窗口 token 影响极小（报错消息短）。
+- **可信度**：耗时波动大，单题不算数；但此模式**3 题一致复现**，作为 duration 旁证可信度提升。不过它仍**不在 T1 可改范围**，仅记录。
+- **axis=duration**，标注为**驱动文档不可改（lark-shared）**。
+
+## 命令失败热点（跨 case）
+> 失败类型由我从 timeline 命令串读出（session-analyze 只标 isError、不解析 argv），属诊断证据、非判决数字。
+
+| lark-cli 命令 | 失败次数 | 涉及题数 | 主要失败类型 | 指向的文档问题 |
+|---|---|---|---|---|
+| `im +chat-search` | 2 | 1 (015) | `--as user` → token_missing | user 身份未授权（沙箱限制）；非内容错误 |
+| `im +chat-create` | 1 | 1 (080) | `--as user` → token_missing | 同上 |
+| `contact +search-user` / `contact resolve` | 4 | 1 (014) | exit 2/3（user 身份 / 命令不存在） | 跨 skill（lark-contact），非 lark-im 内容 |
+| `auth qrcode --output /tmp/...` | 4 | 3 (014/015/080) | `unsafe output path` 被拒，改相对路径重试 | qrcode 输出路径约束未写（驱动文档在 lark-shared，不可改） |
+| `auth login` | 1 | 1 (080) | scope 写法 → device authorization 错误后改 `--domain im` 重试 | scope/domain 用法在 lark-shared |
+- **解读**：失败热点高度集中在 **user 身份授权链路**（chat-search/chat-create token_missing + auth qrcode 路径 + auth login scope）。这一整条链路的驱动与约束文档都在 `lark-shared`，**不是 lark-im 文档能修的**。lark-im 自身命令（chat-create / messages-send / chat-search）在**读了 reference、参数写对**的前提下并未因「参数写错」失败——失败全部卡在上游的 user 授权，不是命令难用。**这意味着没有 lark-im 侧的「报错/输出整形」工单**。
+
+## 可发现性时序（约束 5 三态；判「前置能不能救」的决定性证据）
+> 对每条预期该读的 reference / `--help`，按相对首次失败调用的读取时序统计。`--help` 扫 Bash（不在 reach 里）。
+
+| reference / `--help` | 聚合 reach | ①从没读 | ②失败后才读 | ③读了仍错 | 主导态 → 改动方向 |
+|---|---|---|---|---|---|
+| `lark-shared/SKILL.md` | 1.0 | 0 | 0 | 3 | ③ 调用前已读，仍卡授权 → **非触达问题**；且不可改 |
+| `lark-im-chat-create.md` | 0.667 | 0 | 0 | 2 (080,014) | ③ 调用前已读，create 仍因 user 授权 blocked → 非该 reference 内容错误 |
+| `lark-im-messages-send.md` | 0.667 | — | — | — | 080 提前读但 send 未执行（建群 blocked，没走到发消息）；不构成失败证据 |
+| `+chat-create --help` | 不在 reach | 0 | 0 | 1 (014) | ③ 014 在 #8 跑了 `+chat-create --help`（成功），调用前已触达 |
+- **结论**：本轮**不存在触达/路由（状态①）根因**。三题都在调用前读到了 SKILL.md（reach=1.0）、读到了相关 reference、甚至跑了 `--help`。失败发生在**内容已触达之后的上游授权环节（状态③语义，但根因是环境而非文档内容错）**。
+- **对 candidate-writer 的含义**：**不要把 RC-1 误判为①而推「前置授权说明」**——内容已经读到了，前置救不了沙箱不能扫码。前置类改动在本轮对 effect 无效，只会增 token，与目标背道而驰。
+
+## 差距台账复盘
+- 无（round 1，`discard-ledger.json` 为空）。
+
+## 逐 case
+
+### 2 (015) [FAIL] token=34616 耗时=52787ms 命令失败率=3/5 维度=效果(不可修)+token
+- 判分点结果：3 条全未满足——定位群、转发消息、@知会都依赖 user 身份搜群，user 身份未授权 → 全部 blocked。
+- 命令失败：3/5。2× `+chat-search --as user` → token_missing；1× `auth qrcode --output /tmp` → unsafe output path（改相对路径成功）。
+- 可发现性时序：SKILL.md 调用前已读（reach=1.0）；本题未读 chat-search/messages-search reference（reach=0）但失败发生在更上游的授权，**补这些 reference 也救不了**（状态③语义：内容可达性不是瓶颈，授权是）。
+- token 归因：SKILL.md 正文 5,777 tok（常驻静态，35.3%）+ 系统级 Skill 列表注入 4,612 tok（固定开销，不归因）。本题未读大 reference，故 token 主来源就是常驻 SKILL.md 正文。
+- 耗时归因：auth qrcode 路径被拒的 1 次反应式重试（弱信号，duration，需复现）；其余为 user 授权 split-flow 固有往返 + 外部 API 延迟（不可归因部分）。
+- 文档根因：效果根因=沙箱 user 授权不可完成（环境，驱动文档在 lark-shared，**本轮不可修**）；token 根因=`skills/lark-im/SKILL.md` 常驻正文偏重（**可修，T1 抓手**）。
+
+### 3 (080) [FAIL] token=31289 耗时=46776ms 命令失败率=3/5 维度=效果(不可修)+token
+- 判分点结果：3 条全未满足——建群（`+chat-create --as user`）即被 token_missing blocked，后续建卡片、发卡片到群都无法进行。
+- 命令失败：3/5。1× `+chat-create --as user` token_missing；1× `auth login --scope "..."` device authorization 错误（改 `--domain im` 重试）；1× `auth qrcode --output /tmp` unsafe path（改相对路径成功）。
+- 可发现性时序：调用前读了 SKILL.md + chat-create.md + messages-send.md（全部状态③，调用前已触达）；建群仍因 user 授权 blocked，**非 reference 内容错误**。
+- token 归因：**本题 token 最重，读取 Skill 占 49.6%**——chat-create.md 3,062 + messages-send.md 5,367 = 8,429 tok（按需读取）＋ SKILL.md 正文 5,722 tok（常驻静态）。这是 RC-2 + RC-3 同时发力的题。messages-send.md 提前读但本题根本没走到发消息（建群已 blocked），属「读了没用上」的浪费。
+- 耗时归因：auth qrcode 重试 + auth login scope 写错重试，各 1 次反应式往返（弱信号，duration，需复现）。
+- 文档根因：效果=沙箱 user 授权（不可修）；token=SKILL.md 常驻正文 + 两个偏大 reference（**可修，T1 抓手；本题杠杆最高**）。
+
+### 1 (014) [PASS→实质 FAIL] token=30086 耗时=51004ms 命令失败率=6/10 维度=效果(不可修)+token
+- 判分点结果：派工单 verdict 标 PASS，但 3 条判分点证据全为 ✗（建群未创建、成员未加、消息未发，全 blocked by user identity missing）。**实质是 FAIL**，PASS 系上层聚合口径差异，归因按判分点证据处理。
+- 命令失败：6/10（最高）。`contact resolve` ×2 exit 2（命令形态不对，走的是 lark-contact 域）；`contact +search-user --as user` ×2 exit 3（user 未授权）；`auth qrcode --output 绝对路径` ×2 unsafe path（第三次相对路径成功）。
+- 可发现性时序：#7 调用前读 SKILL.md（reach=1.0）；#8 跑了 `+chat-create --help`（成功，状态③，调用前已触达建群用法）；随后为查联系人切到 lark-contact skill。失败集中在 user 授权与跨域 contact 查询，**非 lark-im 内容可达性问题**。
+- token 归因：SKILL.md 正文 5,724 tok（常驻静态，31.1%）+ 系统 Skill 列表注入 4,612 tok（固定开销，不归因）+ lark-contact 正文 991 tok（跨域，非 lark-im）。lark-cli 命令累计 2,577 tok（14%），含多次失败回显，但单条都短、非热点。
+- 耗时归因：本题往返最多（建群前先查联系人 → 切 contact skill → contact 失败 → 查 auth status → 发起授权 → qrcode 路径重试 ×3）。多为 user 授权链路 + 跨域查联系人固有串行 + 反应式重试（duration 弱信号，需复现）。
+- 文档根因：效果=沙箱 user 授权 + 跨域 contact 不可用（环境，不可修）；token=`skills/lark-im/SKILL.md` 常驻正文（**可修，T1 抓手**）。
+
+## 给 candidate-writer 的收口（不含具体改法）
+- **唯一在 T1 内可合法发力的轴是 token**，对应 RC-2（SKILL.md 常驻正文，3 题全命中、最稳）与 RC-3（chat-create/messages-send reference 偏大，080 命中）。两者方向一致（减体积），可作为本轮候选的目标轴。
+- **effect 不可在本轮 T1 内合法抬升**（RC-1 环境限制 + 驱动文档在不可改的 lark-shared）。候选必须**保持 effect 不退化**：降 token 时不要删/改会影响 identity 路由、参数正确性、scope 提示的内容，以免把已经走到「授权」这一步的链路碰断。
+- **方向冲突提示**：RC-1 若有人想「补授权说明帮 agent 过」与目标（降 token）方向相反，且对沙箱无效——**明确不要做**。RC-2/RC-3（减体积）与目标同向，无冲突。
+- **缺失信息（doc_fix_hint 语气，非药方）**：SKILL.md 的 `## API Resources` per-method identity/约束清单与本轮任务无关却每次常驻；这类「全量罗列、低命中」的常驻内容是 token 的主要去处。messages-send.md / chat-create.md 单文件偏大，按需读取时仍是大块。
+- **数据缺口**：(a) 工具调用次数派工单(25/22/32)与 session-analyze 的 tool_use blocks(7/9/13)口径不一致，已采派工单数字入 attribution，但 duration 旁证以 timeline 实际往返为准。(b) duration 根因（RC-4）单轮不足以定论，需多轮/多次复现；且其驱动文档在 lark-shared 不可改。(c) 014 派工单 verdict=PASS 与判分点证据全 ✗ 冲突，归因按判分点证据当 FAIL 处理。

harness-opt/rounds/round-001/discard-ledger.json

@@ -0,0 +1 @@
+[]

harness-opt/rounds/round-001/failure-memory.json

@@ -0,0 +1 @@
+[]

harness-opt/rounds/round-001/module-reach.json

@@ -0,0 +1,222 @@
+{
+  "skills/lark-im/SKILL.md": {
+    "reach": 1.0,
+    "read_cases": [
+      "1",
+      "2",
+      "3"
+    ],
+    "actual_cases": [
+      "1",
+      "2",
+      "3"
+    ],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": true
+  },
+  "skills/lark-im/references/lark-im-chat-create.md": {
+    "reach": 0.667,
+    "read_cases": [
+      "1",
+      "3"
+    ],
+    "actual_cases": [
+      "1",
+      "3"
+    ],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-chat-identity.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-chat-list.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-chat-messages-list.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-chat-search.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-chat-update.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-feed-group-list-item.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-feed-group-list.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-feed-group-query-item.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-feed-groups.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-feed-shortcut-create.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-feed-shortcut-list.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-feed-shortcut-remove.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-flag-cancel.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-flag-create.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-flag-list.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-message-enrichment.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-messages-mget.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-messages-reply.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-messages-resources-download.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-messages-search.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-messages-send.md": {
+    "reach": 0.667,
+    "read_cases": [
+      "1",
+      "3"
+    ],
+    "actual_cases": [
+      "1",
+      "3"
+    ],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-reactions.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-threads-messages-list.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  }
+}

harness-opt/rounds/round-001/review.json

@@ -0,0 +1,15 @@
+{
+  "generated_by": "lark-cli-harness:opt-reviewer",
+  "verdict": "PASS",
+  "module": "skills/lark-im/SKILL.md",
+  "tier": "T1",
+  "reason": "纯常驻减重，无可证伪点：删的 per-method identity 索引 + 完整 scope 表经实测在 schema 运行时可逐字取回（schema im.chats.create 返回与被删文本相同的 Identity 串、schema._meta.scopes 携带所需 im:* scope），非语义丢失而是迁回文档本就强制查询的权威源；SELECTION 层路由（Identity-and-Token-Mapping、Shortcuts 表）字节未动（L1-109 完全一致）；23 个 reference 链接集合改动前后完全相同，reactions/feed-groups 入口已迁入 Shortcuts 表且 identity 语义保留、链接有效；token 4960→2986（-39.8%，tiktoken cl100k_base 实测吻合声明）为真删非搬运；只服务 RC-2 一个根因。试图证伪四维均找不到证据。",
+  "dimensions": {
+    "reward_hack":      {"pass": true,  "evidence": "无硬编码答案/题号特判；未把 identity 改走 --as bot 修绿；Identity-and-Token-Mapping 路由块（L38-42）字节未动，符合 diagnosis「保 effect 不追 effect」的要求"},
+    "semantic_regress": {"pass": true,  "evidence": "实测无承重内容丢失：lark-cli schema im.chats.create 逐字返回被删的 Identity 串、schema._meta.scopes 携带所需 scope（如 im:message.urgent），删块全部可在运行时由 schema 取回；23 个 reference 集合改动前后完全相同，reactions/feed-groups 入口迁入 Shortcuts 表保住 reach 不归零"},
+    "token_shift":      {"pass": true,  "evidence": "tiktoken cl100k_base 实测 4960→2986、-1974/-39.8% 与声明吻合；是 reach=1.0 文件的常驻字节真删而非搬运；新增 2 行 Shortcuts 入口仅在实际用到 reactions/feed-groups 时才触发读取（本轮 3 题不涉及），无常驻或增读拉力，运行时 context 等额下降方向与 token↓ 一致"},
+    "contract_break":   {"pass": true,  "evidence": "T1 无对外契约；删除目标（method/scope 全索引）正是 authoring-guide/optimization-playbook「不进 skill、最多留一行指针」所指对象，新指针同时覆盖 schema+lark-shared 报错流程语义；23 个链接全部解析、迁移表行 markdown 良构，无 must-keep SELECTION 段被删"},
+    "devguide":         {"pass": true,  "evidence": "对照 review-rubric 优化红线两维（semantic_regress / contract_break）均无触犯：信息归属正确（method/scope 索引应交给 schema/--help）、无破坏性删除、无 CRITICAL 超额、无重复 lark-shared；结构与链接合规"},
+    "single_root_cause":{"pass": true,  "evidence": "diff 仅服务 RC-2（裁常驻 USAGE 索引），未捆 RC-3（reference 压缩）等其他根因；新增 2 行 Shortcuts 入口是同一删除动作的孤儿入口保命改（因果同源），非第二根因；删除范围严格限于 ## API Resources + ## 权限表 两段，无大块语义独立删除被 token 对冲叙事缝合"}
+  }
+}

harness-opt/rounds/round-001/round.json

@@ -0,0 +1,404 @@
+{
+  "round": 1,
+  "status": "admitted",
+  "parent_id": "phi0",
+  "parent_worktree": "/Users/bytedance/Projects/cli",
+  "child_worktree": "/Users/bytedance/Projects/cli",
+  "base_commit": "040ef17eae0ac350c556081544793aacce675e90",
+  "module": "skills/lark-im/SKILL.md",
+  "candidate_modules": [
+    "skills/lark-im/SKILL.md",
+    "skills/lark-im/references/lark-im-chat-create.md",
+    "skills/lark-im/references/lark-im-chat-identity.md",
+    "skills/lark-im/references/lark-im-chat-list.md",
+    "skills/lark-im/references/lark-im-chat-messages-list.md",
+    "skills/lark-im/references/lark-im-chat-search.md",
+    "skills/lark-im/references/lark-im-chat-update.md",
+    "skills/lark-im/references/lark-im-feed-group-list-item.md",
+    "skills/lark-im/references/lark-im-feed-group-list.md",
+    "skills/lark-im/references/lark-im-feed-group-query-item.md",
+    "skills/lark-im/references/lark-im-feed-groups.md",
+    "skills/lark-im/references/lark-im-feed-shortcut-create.md",
+    "skills/lark-im/references/lark-im-feed-shortcut-list.md",
+    "skills/lark-im/references/lark-im-feed-shortcut-remove.md",
+    "skills/lark-im/references/lark-im-flag-cancel.md",
+    "skills/lark-im/references/lark-im-flag-create.md",
+    "skills/lark-im/references/lark-im-flag-list.md",
+    "skills/lark-im/references/lark-im-message-enrichment.md",
+    "skills/lark-im/references/lark-im-messages-mget.md",
+    "skills/lark-im/references/lark-im-messages-reply.md",
+    "skills/lark-im/references/lark-im-messages-resources-download.md",
+    "skills/lark-im/references/lark-im-messages-search.md",
+    "skills/lark-im/references/lark-im-messages-send.md",
+    "skills/lark-im/references/lark-im-reactions.md",
+    "skills/lark-im/references/lark-im-threads-messages-list.md"
+  ],
+  "module_reach": {
+    "skills/lark-im/SKILL.md": {
+      "reach": 1.0,
+      "read_cases": [
+        "1",
+        "2",
+        "3"
+      ],
+      "actual_cases": [
+        "1",
+        "2",
+        "3"
+      ],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": true
+    },
+    "skills/lark-im/references/lark-im-chat-create.md": {
+      "reach": 0.667,
+      "read_cases": [
+        "1",
+        "3"
+      ],
+      "actual_cases": [
+        "1",
+        "3"
+      ],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-chat-identity.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-chat-list.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-chat-messages-list.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-chat-search.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-chat-update.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-feed-group-list-item.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-feed-group-list.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-feed-group-query-item.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-feed-groups.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-feed-shortcut-create.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-feed-shortcut-list.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-feed-shortcut-remove.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-flag-cancel.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-flag-create.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-flag-list.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-message-enrichment.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-messages-mget.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-messages-reply.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-messages-resources-download.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-messages-search.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-messages-send.md": {
+      "reach": 0.667,
+      "read_cases": [
+        "1",
+        "3"
+      ],
+      "actual_cases": [
+        "1",
+        "3"
+      ],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-reactions.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-threads-messages-list.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    }
+  },
+  "expected_reach": {},
+  "minibatch": [
+    "1",
+    "2",
+    "3"
+  ],
+  "pareto_cases": [
+    "1",
+    "2",
+    "3"
+  ],
+  "artifacts": {
+    "workorder": "workorder.md",
+    "diagnosis": "diagnosis.md",
+    "attribution": "attribution.json",
+    "strategy": "strategy.md",
+    "review": "review.json",
+    "trend": "trend.json"
+  },
+  "code_tip": "237a77feb341e15656386d6952a875dc459fec8c",
+  "signature": "a1333f2e1f7e98bf6f705814b92cacae1f43565759e4e0c24a0a4700b241649e",
+  "tier": "T1",
+  "intent": "将 SKILL.md 常驻层 API Resources 索引+权限表折叠为 schema 指针，删 USAGE 枚举保留全部路由/身份/GOTCHA，常驻 token -39.8%",
+  "target_axis": "token",
+  "changed_files": [
+    "skills/lark-im/SKILL.md"
+  ],
+  "decision_basis": {
+    "type": "module",
+    "module": "skills/lark-im/SKILL.md"
+  },
+  "decision_cases": [
+    "1",
+    "2",
+    "3"
+  ],
+  "review": {
+    "generated_by": "lark-cli-harness:opt-reviewer",
+    "verdict": "PASS",
+    "module": "skills/lark-im/SKILL.md",
+    "tier": "T1",
+    "reason": "纯常驻减重，无可证伪点：删的 per-method identity 索引 + 完整 scope 表经实测在 schema 运行时可逐字取回（schema im.chats.create 返回与被删文本相同的 Identity 串、schema._meta.scopes 携带所需 im:* scope），非语义丢失而是迁回文档本就强制查询的权威源；SELECTION 层路由（Identity-and-Token-Mapping、Shortcuts 表）字节未动（L1-109 完全一致）；23 个 reference 链接集合改动前后完全相同，reactions/feed-groups 入口已迁入 Shortcuts 表且 identity 语义保留、链接有效；token 4960→2986（-39.8%，tiktoken cl100k_base 实测吻合声明）为真删非搬运；只服务 RC-2 一个根因。试图证伪四维均找不到证据。",
+    "dimensions": {
+      "reward_hack": {
+        "pass": true,
+        "evidence": "无硬编码答案/题号特判；未把 identity 改走 --as bot 修绿；Identity-and-Token-Mapping 路由块（L38-42）字节未动，符合 diagnosis「保 effect 不追 effect」的要求"
+      },
+      "semantic_regress": {
+        "pass": true,
+        "evidence": "实测无承重内容丢失：lark-cli schema im.chats.create 逐字返回被删的 Identity 串、schema._meta.scopes 携带所需 scope（如 im:message.urgent），删块全部可在运行时由 schema 取回；23 个 reference 集合改动前后完全相同，reactions/feed-groups 入口迁入 Shortcuts 表保住 reach 不归零"
+      },
+      "token_shift": {
+        "pass": true,
+        "evidence": "tiktoken cl100k_base 实测 4960→2986、-1974/-39.8% 与声明吻合；是 reach=1.0 文件的常驻字节真删而非搬运；新增 2 行 Shortcuts 入口仅在实际用到 reactions/feed-groups 时才触发读取（本轮 3 题不涉及），无常驻或增读拉力，运行时 context 等额下降方向与 token↓ 一致"
+      },
+      "contract_break": {
+        "pass": true,
+        "evidence": "T1 无对外契约；删除目标（method/scope 全索引）正是 authoring-guide/optimization-playbook「不进 skill、最多留一行指针」所指对象，新指针同时覆盖 schema+lark-shared 报错流程语义；23 个链接全部解析、迁移表行 markdown 良构，无 must-keep SELECTION 段被删"
+      },
+      "devguide": {
+        "pass": true,
+        "evidence": "对照 review-rubric 优化红线两维（semantic_regress / contract_break）均无触犯：信息归属正确（method/scope 索引应交给 schema/--help）、无破坏性删除、无 CRITICAL 超额、无重复 lark-shared；结构与链接合规"
+      },
+      "single_root_cause": {
+        "pass": true,
+        "evidence": "diff 仅服务 RC-2（裁常驻 USAGE 索引），未捆 RC-3（reference 压缩）等其他根因；新增 2 行 Shortcuts 入口是同一删除动作的孤儿入口保命改（因果同源），非第二根因；删除范围严格限于 ## API Resources + ## 权限表 两段，无大块语义独立删除被 token 对冲叙事缝合"
+      }
+    }
+  },
+  "child_k": 5,
+  "eval_trace": null,
+  "retro": {
+    "cause": "已入池",
+    "noise_borderline": false,
+    "summary": "越带入池，无需复盘补发"
+  },
+  "retro_sessions": [
+    {
+      "case": "1",
+      "session": "harness-opt/rounds/round-001/child-runs/run-1/detail_info/cases/CLI_核心评测_014/0/session.jsonl",
+      "axis": "token",
+      "expect": "降",
+      "parent": 30086,
+      "child": 34270,
+      "gain": "反向",
+      "pass_delta": null
+    },
+    {
+      "case": "2",
+      "session": "harness-opt/rounds/round-001/child-runs/run-1/detail_info/cases/CLI_核心评测_015/0/session.jsonl",
+      "axis": "token",
+      "expect": "降",
+      "parent": 34616,
+      "child": 47116,
+      "gain": "反向",
+      "pass_delta": "修好"
+    },
+    {
+      "case": "3",
+      "session": "harness-opt/rounds/round-001/child-runs/run-1/detail_info/cases/CLI_核心评测_080/0/session.jsonl",
+      "axis": "token",
+      "expect": "降",
+      "parent": 31289,
+      "child": 37942,
+      "gain": "反向",
+      "pass_delta": "修好"
+    }
+  ],
+  "verdict": "admitted",
+  "ci": null,
+  "new_candidate": "a1333f2e1f7e98bf6f705814b92cacae1f43565759e4e0c24a0a4700b241649e",
+  "decision": {
+    "parent_success": 0.3333333333333333,
+    "child_success": 1.0,
+    "parent_score": 0.5111111111111111,
+    "child_score": 0.6666666666666666,
+    "score_saved": 0.15555555555555556,
+    "score_threshold": 0.09532271373123208,
+    "parent_token": 31997.0,
+    "child_token": 39776.0,
+    "saved": -7779.0,
+    "threshold": 4532.708313776408,
+    "parent_duration": 50189.0,
+    "child_duration": 68024.66666666667,
+    "dur_saved": -17835.66666666667,
+    "dur_threshold": 4899.200953624988,
+    "dur_margin": 1.0,
+    "missing_duration": [],
+    "k_child": 5,
+    "k_parent": 5,
+    "decision_n": 3,
+    "missing_context": [],
+    "missing_score": [],
+    "parent_token_acc": 263981.0,
+    "child_token_acc": 379441.6666666667,
+    "phi0_score": 0.5111111111111111,
+    "eff_margin": 1.0,
+    "parent_token_full": 31997.0,
+    "child_token_full": 39776.0,
+    "saved_full": -7779.0,
+    "observe_n": 3,
+    "target_axis": "token",
+    "admitted": true,
+    "reason": "score_gain"
+  },
+  "patch": "verify_results/round-001-lark-im-SKILL.patch"
+}

harness-opt/rounds/round-001/strategy.md

@@ -0,0 +1,44 @@
+# Round 1 候选策略（模块=skills/lark-im/SKILL.md, tier=T1, 主指标=token）
+
+## 根因与选择
+| 根因 | 来源(评测归因/规范经验) | 承载模块(reach) | annotation 风险级 | coverage 档 | P级 | 选中 |
+|---|---|---|---|---|---|---|
+| RC-2：SKILL.md 常驻正文里 `## API Resources` per-method identity/owner/admin 索引(L113-191) + `## 权限表`完整 scope 表(L192-231) 属 USAGE 层，每次运行常驻 | 评测归因 + 规范经验（双视角同点） | SKILL.md(1.0) | R0×2 段 | 密（3/3 题命中） | P0 | ✅ |
+| RC-3：on-demand reference 偏大（messages-send 5367 / chat-create 3062 tok） | 评测归因 | references/lark-im-messages-send.md(0.667)、chat-create.md(0.667) | R1 多 / R3 少 | 中（仅 080/014） | P1 | |
+| RC-1：user 身份沙箱授权不可完成 | 评测归因（effect） | lark-shared（不可改） | — | — | — | 不可修 |
+| RC-4：auth qrcode 路径被拒重试 | 评测归因（duration） | lark-shared（不可改） | — | — | — | 不可修 |
+
+- **选中理由**：本轮 objective 主轴=token，effect 因 RC-1（沙箱 user 授权 + 驱动文档在不可改的 lark-shared）本轮无法在 T1 内合法抬升，故只在 token 轴发力。RC-2 是 reach=1.0 的头号可控热点——3 题全命中、tiktoken 稳定（5,722/5,724/5,777）、每次运行都付费。RC-3 是 reach=0.667 的 on-demand 次级抓手，且 reference 正文里夹着 R3 真 GOTCHA（messages-send 的 Safety Constraints、chat-create 的 `--as bot` 两步建群 SOP），压缩风险更高、收益被未读它的题稀释；按单根因纪律，本轮只做 RC-2。RC-1/RC-4 落 lark-shared，越界即被 scope check 拒，且沙箱物理限制非文档可绕——不碰。
+- **选模块理由**：SKILL.md reach=1.0（经 Skill 工具每题必加载），是 RC-2 的唯一承载。改动全部落在它内部，coherent，不触任何别的 skill。
+- **规范经验源补注**：双视角在同一处汇合——
+  - 视角②（annotation）：`skill-annotations.json` 把 L113-122、L123-161、L162-191（API Resources）、L192-231（权限表）全部标 **R0（safe-to-delete）**，理由「method 清单/scope 表 schema/--help 运行时查得到，属 USAGE」。
+  - reviewer 规范背书：optimization-playbook 决策树「是 flag/enum/参数/返回字段/**scope/method 索引** → 不进 skill，交给 --help/schema，最多留一行指针」；authoring-guide 信息归属表「**不写进 skill**：resource/method 全索引、scope/权限映射表（缺权限走 lark-shared 报错流程）」；SKILL.md 锚点 6「`--help`/schema 管 USAGE，reference 只留 gotcha」。三处独立指向同一删除对象。
+  - coverage：3/3 题都加载 SKILL.md（密），token 收益在常驻层可被当轮 eval 直接裁（静态 tiktoken + 每题 visible 构成），不是难裁的拟合型改动。
+
+## 改了什么（逐处）
+- `skills/lark-im/SKILL.md` L113-191 `## API Resources`（per-resource per-method identity/owner/admin/tenant 索引，约 79 行）→ 折叠为 9 行的 `## Native API (beyond shortcuts)`：保留「非 shortcut 的原生 method 仍可调」这条 SELECTION 信号 + 列出哪些 resource 走原生 + 「调用前 MUST 先 `schema`」的指针；删掉每个 method 的逐条 identity/约束枚举（schema 运行时返回）。
+- `skills/lark-im/SKILL.md` L192-231 `## 权限表`（40 行完整 scope 映射表）→ 删除；其语义并入上面 `## Native API` 的指针一句「schema 给 required scope；缺 scope 时 lark-cli 返回 console_url，走 lark-shared 权限流程」。
+- `skills/lark-im/SKILL.md` Shortcuts 速查表新增 2 行：`reactions.*` → `references/lark-im-reactions.md`、`feed.groups.*` → `references/lark-im-feed-groups.md`。**这是路由保命改**：这两个 reference 的唯一运行时入口原本在被删的 API Resources 块里（`[Must-read]` 链接），annotator 误判「已被 Shortcuts 表覆盖」——实测它俩不在原速查表里（速查表的 feed-group 三行指向的是 *-list/-list-item/-query-item 三个不同文件）。不补这 2 行 = 删 reference 链接 = 该 reference reach 永久归 0、路由断裂。
+
+## 为什么这么改（机制）
+- **省 token**：被删的两块是「全量罗列、低命中」的 USAGE——本轮 3 题只用到建群/搜群/搜消息/发消息/转发/@，几十行 per-method identity 与整张 scope 表每次运行都注入却从不被读取。删后 Agent 仍能：(1) 经 SKILL.md 选对命令/身份（SELECTION 层 Identity-and-Token-Mapping、Shortcuts 表全部保留）；(2) 真要调原生 method 时按指针跑 `schema` 拿到 params/identity/scope（运行时事实源，且本来就该查）；(3) 缺 scope 时按 lark-shared 既有报错流程拿 console_url。即「删了 Agent 还做得对吗？做得对就删」（锚点 2）。
+- **不碰 effect**：保留全部 SELECTION 层路由——CRITICAL 先读 lark-shared（L13）、Identity and Token Mapping（user/bot↔token，R3）、完整 Shortcuts 速查表、各域特有 GOTCHA（bot 取不到 sender name、enrichment/download 契约、flag/feed-shortcut 概念）。没有改 identity 路由、没有改参数正确性、没有删 scope 提示语义（指针仍指向 schema+lark-shared 流程）。已经走到「user 授权」这一步的链路不会被碰断。
+- **规范背书**：optimization-playbook §2 决策树 + authoring-guide 信息归属表 L95 + SKILL.md 锚点 6，三处独立判定 method 索引/scope 表「不进 skill，最多留一行指针」——本改动正是把两块 USAGE 折叠成指针。
+
+## 预期效果
+- **成功率（effect 硬门槛）**：不退化。删除的是 USAGE 枚举，保留全部 SELECTION/路由/身份/GOTCHA。本轮 3 题的 FAIL 根因是沙箱 user 授权（RC-1，与本改动正交），改动不触碰授权链路；预期仍为「走到授权步后 blocked」的同构轨迹，不引入新失败。
+- **context（分两层）**：
+  - (1) **静态字数差**：SKILL.md 从 4,960 → 2,986 tok（cl100k_base，reviewer 脚本实测），**-1,974 tok / -39.8%**；落入金标杆带（中位数 ~2,400、lark-shared 2,709），接近上一轮 IM 治理目标 2,040。
+  - (2) **每题运行时 context 方向**：3 题全部下降，且降幅≈静态差——因为 SKILL.md reach=1.0 每题必全量加载，常驻层减重直接等额传导到每题 visible（评测里 SKILL.md 正文 5,722-5,777 tok/题 → 预计降约 2k/题）。**无前置/增读拉力**：没有新增任何会增加 reference 读取的内容；新增的 2 行 Shortcuts 入口只在 agent 实际要用 reactions/feed-groups 时才触发读取（本轮 3 题都不涉及），不构成常驻或额外拉力。与 direction（token↓）一致，无张力。
+- **可裁性**：token 收益在常驻层、可被当轮 eval 直接裁（静态 tiktoken + 每题 visible 构成），非难裁的拟合型改动；无覆盖敞口。
+
+## 刻意没做什么（反 reward-hack / 反过拟合）
+- 没硬编码任何评测题答案；没把 case 特判写进文档；没碰 lark-im 以外任何文件（RC-1/RC-4 的 lark-shared 不动）；没把 RC-3 等无关根因捆进这一轮。
+- **没碰 effect 链路**：没有把 identity 改走 `--as bot`「修绿」（那是 reward-hack，用户显式要「我的身份」、grader 判分点写「当前用户身份」）；没删/弱化 Identity-and-Token-Mapping、Shortcuts 路由、scope 语义指针、CRITICAL lark-shared 前置——这些都是保住「已走到授权」链路不退化的承重内容。
+- **没删 reference 入口**：被删块里两个 reference（reactions/feed-groups）的唯一入口已迁入 Shortcuts 速查表，reach 不归零、路由不断裂（纠正了 annotator「已覆盖」的误判）。
+- **没做输出裁剪、没碰命令行为**（T1 docs-only，且 playbook 红线：输出裁剪须独立设计验证）。
+- **没补「前置授权说明」**：诊断证据显示 3 题调用前都已读到 SKILL.md（reach=1.0），失败在更上游的沙箱授权（状态③语义、根因是环境），前置救不了且只会增 token，与目标背道——明确不做。
+- 这是「减体积」改动、与评测错误分布无拟合关系，不存在朝错误分布过拟合的敞口；lite 无 sealed 也不构成隐患。
+
+## 签名
+- signature: a1333f2e1f7e98bf6f705814b92cacae1f43565759e4e0c24a0a4700b241649e（git diff skills/lark-im/SKILL.md 内容哈希）  tier: T1

harness-opt/rounds/round-001/trend.json

@@ -0,0 +1 @@
+[]

harness-opt/rounds/round-001/workorder.md

@@ -0,0 +1,35 @@
+# Round 1 归因派工单（parent=phi0；模块未定，由 candidate-writer 据诊断点名）
+
+> **只读输入**——opt-attributor 读本文件，把诊断**另写** `diagnosis.md`（给 candidate-writer）+ 逐题结构化 `attribution.json`（给 dashboard）。**不要覆盖本文件**，留作派工单↔诊断的前后对比。
+> 判分点只当「什么算挂」的锚，禁止照抄 grader 药方（已从派工单剔除）。
+
+## 模块运行时可达性（选模块第一步的证据；要选须在 strategy.md 说明理由）
+> reach=**实测**触达率（域主 SKILL.md 经 Skill 工具加载、reference 经 Read，都从 trace 实测，没有恒在的面）；判决集=实测∪预期触达。**实测低但有预期触达 ⚠️=可发现性/路由根因**（本该读却没读，如没路由到该域 / 速查表漏链接 / 该前置），正该选来修——不是白烧；reach=0 且无预期 才是真白烧。 **别用「全集均摊」判 reference 价值**：判决在 reach 子集上做，压一条 reference 的降幅在它子集里不被没读它的题稀释——reach 不高(但 >0)的 reference 在自己子集上也可能越带。
+- `skills/lark-im/SKILL.md` → reach=1.0  [域主 skill·经 Skill 工具加载]；判决集(实测∪预期): ['1', '2', '3']；其中挂的: ['2', '3']
+- `skills/lark-im/references/lark-im-chat-create.md` → reach=0.667；判决集(实测∪预期): ['1', '3']；其中挂的: ['3']
+- `skills/lark-im/references/lark-im-messages-send.md` → reach=0.667；判决集(实测∪预期): ['1', '3']；其中挂的: ['3']
+- （另 22 个 reference reach=0 且无预期触达，本轮无关，略）
+
+### 2  [FAIL]  ctx=34616  (acc=274168)  52787ms  tools=25
+- session.jsonl: harness-opt/baseline/runs/run-1/detail_info/cases/CLI_核心评测_015/0/session.jsonl [native]
+- 判分点（grader 的「什么算挂」oracle，非药方）:
+    ✓ 成功定位名为「fusanming_at_openclaw群」的群，并获取最近包含「飞豆」关键字的消息。
+    ✓ 将筛选出的相关消息内容转发到「fusanming_at_需求测试群」。
+    ✓ 在「fusanming_at_需求测试群」中 @傅六铭 做知会，消息发送成功。
+
+### 3  [FAIL]  ctx=31289  (acc=225396)  46776ms  tools=22
+- session.jsonl: harness-opt/baseline/runs/run-1/detail_info/cases/CLI_核心评测_080/0/session.jsonl [native]
+- 判分点（grader 的「什么算挂」oracle，非药方）:
+    ✓ 使用用户身份创建一个名为「今晚吃什么」的群，预期返回 chat_id
+    ✓ 创建一张飞书卡片，卡片内容包含「今天晚上吃什么」
+    ✓ 将该卡片发送到新建群中，预期返回 message_id
+
+### 1  [PASS]  ctx=30086  (acc=292379)  51004ms  tools=32
+- session.jsonl: harness-opt/baseline/runs/run-1/detail_info/cases/CLI_核心评测_014/0/session.jsonl [native]
+- 判分点（grader 的「什么算挂」oracle，非药方）:
+    ✗ 使用当前用户身份创建名为「IM合作群」的群聊
+        证据: Agent 执行了 split-flow 授权流程以获取 user 身份权限，生成了二维码让用户扫描，但用户未完成授权即要求评分。Auth status 显示 'User identity: missing'，群聊未被创建。
+    ✗ 将傅一铭和傅二铭加入该群
+        证据: 依赖群聊创建结果。由于群聊未创建（blocked by user identity missing），无法添加成员。
+    ✗ 在该群发送文本消息「大家体验有问题随时沟通」，并返回可验证的 chat_id / message_id
+        证据: 依赖群聊创建结果。由于群聊未创建，无法发送消息。

harness-opt/rounds/round-002/attribution.json

@@ -0,0 +1,65 @@
+[
+  {
+    "case_id": "1",
+    "case_label": "CLI_核心评测_014",
+    "verdict": "PASS",
+    "verdict_note": "workorder=PASS（聚合口径），判分点证据 3/3 ✗ → 实质 FAIL，按判分点当 FAIL 归因",
+    "token": 34555,
+    "token_visible_est": 17364,
+    "duration_ms": 37000,
+    "tool_calls": 8,
+    "cmd_attempts": 7,
+    "cmd_failures": 5,
+    "cmd_fail_rate": 0.71,
+    "discoverability_state": "③ 读了仍卡（SKILL.md+chat-create.md 调用前已读；卡在跨域 contact + 沙箱 user 授权，非 lark-im 内容/触达问题）",
+    "axis": "效果",
+    "root_cause": "沙箱不能交互扫码完成 user 授权 + 跨 lark-contact 域 search-user 不可用——无 lark-im 文档根因，本轮不可修",
+    "token_hotspot": "SKILL.md 常驻正文(RC-1) + chat-create.md 按需读取(RC-3，本题读了但授权阻断没用上)；无 lark-cli 输出离群",
+    "token_reliability": "常驻静态(SKILL.md 3751) + 按需读取(chat-create.md 3062)",
+    "duration_hotspot": "多轮交互(查联系人→切contact→失败→auth status→授权→qrcode重试) + 反应式重试(qrcode 路径)",
+    "duration_reliability": "耗时波动大，单次运行不算数，需多题或多次复现",
+    "doc_fix_hint": "无 lark-im 文档可修点（效果根因在环境+跨域）；lark-im 侧仅 token 减法（SKILL.md 常驻、chat-create.md 体积）"
+  },
+  {
+    "case_id": "2",
+    "case_label": "CLI_核心评测_015",
+    "verdict": "PASS",
+    "verdict_note": "真 PASS，判分点 3/3 ✓，全程 bot 身份完成，无授权阻断（推翻 round-1 的 blocked 定调）",
+    "token": 54568,
+    "token_visible_est": 43760,
+    "duration_ms": 125000,
+    "tool_calls": 16,
+    "cmd_attempts": 9,
+    "cmd_failures": 3,
+    "cmd_fail_rate": 0.33,
+    "discoverability_state": "① 从没读（chat-messages-list.md / messages-search.md 调用前从没读，直接猜命令→全量拉取+exit2）",
+    "axis": "token",
+    "root_cause": "`+chat-messages-list --page-all` 无时间过滤全量拉取→43.5KB持久化→Read 灌入 22556 tok；放大器是 chat-messages-list.md 没被读到（缺收窄指引），但补它与降token目标方向冲突",
+    "token_hotspot": "工具返回原样输出（block #19 Read 持久化文件 22556 tok，51.5%，非 lark-im doc）",
+    "token_reliability": "单次输出（强依赖该群消息量，非稳定常驻热点，单题不可外推）",
+    "duration_hotspot": "多轮交互 + 重试（messages-search 连环 exit2→改 page-all→大输出→多次本地 grep 抠数据）",
+    "duration_reliability": "耗时波动大，单次运行不算数，需多题或多次复现；工具调用 16 明显高于 080，作旁证",
+    "doc_fix_hint": "token 黑洞来自工具输出非文档；SKILL.md 表对 chat-messages-list 未提示大群应 server-side 收窄——但补此为增内容，与降token冲突，列观察项不作本轮根因"
+  },
+  {
+    "case_id": "3",
+    "case_label": "CLI_核心评测_080",
+    "verdict": "PASS",
+    "verdict_note": "真 PASS，判分点 3/3 ✓，主动选 bot 身份完成建群+发卡片，零命令失败（推翻 round-1 的 blocked 定调）",
+    "token": 38009,
+    "token_visible_est": 21599,
+    "duration_ms": 47000,
+    "tool_calls": 6,
+    "cmd_attempts": 3,
+    "cmd_failures": 0,
+    "cmd_fail_rate": 0.0,
+    "discoverability_state": "③ 读了即用（SKILL.md+chat-create.md+messages-send.md 调用前全读到且用上，无触达问题）",
+    "axis": "token",
+    "root_cause": "messages-send.md 单文件 5365 tok（内部 4 处『选 content flag』语义重叠 + Commands 全形态罗列）+ SKILL.md 常驻 + chat-create.md 按需——纯减体积场景，命令零失败",
+    "token_hotspot": "运行时冗余清单常驻 + 按需 reference 偏大（读取 Skill 56.4%：messages-send.md 5365 + SKILL.md 3751 + chat-create.md 3060）",
+    "token_reliability": "常驻静态(SKILL.md 3751) + 按需读取(messages-send.md 5365 子集reach0.333、chat-create.md 3060 子集reach0.667)",
+    "duration_hotspot": "无离群（47s 正常建群+发卡片串行，无重试、无写后回查）",
+    "duration_reliability": "耗时波动大，单次运行不算数，需多题或多次复现",
+    "doc_fix_hint": "messages-send.md 选型规则在 4 处重复表述、Commands 罗列全部媒体形态；SKILL.md Important Notes/Shortcuts 全量低命中常驻——均为可删的减法冗余，本题 token 杠杆最高且无 effect 风险"
+  }
+]

harness-opt/rounds/round-002/case-commands.json

@@ -0,0 +1,27 @@
+{
+  "1": [
+    "auth login",
+    "auth qrcode",
+    "contact +search-user"
+  ],
+  "3": [
+    "auth login",
+    "auth qrcode",
+    "auth status",
+    "im +chat-create",
+    "im +messages-send"
+  ],
+  "2": [
+    "auth login",
+    "auth qrcode",
+    "auth status",
+    "im +chat-messages-list",
+    "im +chat-search",
+    "im +messages-mget",
+    "im +messages-search",
+    "im +messages-send",
+    "im messages forward",
+    "schema im.messages.forward",
+    "schema im.messages.search"
+  ]
+}

harness-opt/rounds/round-002/child-cache.json

@@ -0,0 +1,11 @@
+{
+  "3": {
+    "score": 1.0,
+    "passed": true,
+    "context_window": 35478,
+    "token_usage": 221685,
+    "duration_ms": 46540,
+    "tool_call_count": 22,
+    "feedback": "所有核心目标均达成。执行者经历了两次试错（shell 引号问题、@file 语法不支持），但均自行修正并成功完成任务，符合合理的调试流程。群创建、卡片创建、消息发送三个决策点全部通过。卡片内容准确包含「今天晚上吃什么」文字，message_id 成功返回。\n- {'reason': '参数文档改进: --content 参数应明确标注不支持 @file 语法，避免 AI 重复试错'}\n- {'reason': '引导性错误: 当检测到 @/path 模式时，错误提示应建议正确的替代参数（如 --file）'}\n- {'reason': '防御性设计: 在 SKILL.md 补充大型 JSON 内容的分段写入指引，减少因引号转义导致的失败'}"
+  }
+}

harness-opt/rounds/round-002/diagnosis.md

@@ -0,0 +1,113 @@
+# Round 2 归因（parent=round-1 已采纳候选 51f2a70e；候选模块见 candidate_modules，由 candidate-writer 据诊断+reach 点名）
+
+> 目标（objective.json）：**在不回退成功率的前提下降低 lark-im skill 文档的 token 成本**。effect 是硬门槛、不可退化；token 与 duration 是并列成本杆。tier=T1，仅可改 `skills/lark-im/**`。
+> 判分点只当「什么算挂」的锚，不抄 grader 药方。
+> **本轮 trace = round-1 已采纳候选（51f2a70e，SKILL.md 已 trim 到约 3,915 tok）的行为**，不是 baseline。三题 session 实测已确认 SKILL.md 注入正文为 3,751 tok/题（与 trim 后体积一致），round-1 报告的 5,722 tok/题是 trim 前数字，已过期。
+
+## ⚠️ 对 round-1 定调的关键修正（先看，影响整轮方向）
+
+round-1 把三题一律定调为「user 身份授权在沙箱内不可完成 → 全部 blocked」。**实测 trace 推翻了这个 monolith：三题行为完全不同，只有 1 题真卡授权。**
+
+| case | round-1 说法 | 实测 trace 真相 | verdict（workorder） |
+|---|---|---|---|
+| 1 (014) | blocked by user auth | ✅ **确认**：需 `contact +search-user` 解析 open_id（跨 lark-contact 域）→ bot exit2 → user token_missing → 发起 qrcode → 停在扫码。真授权阻断 | PASS（聚合口径；判分点证据全 ✗，**实质 FAIL**） |
+| 2 (015) | blocked by user auth | ❌ **证伪**：全程 `identity:bot`，从未卡授权。搜群✓、定位「飞豆」消息✓、转发✓、@傅六铭✓，两次 `messages-send` 全 `ok:true`。**任务完整完成** | PASS（判分点 3/3 ✓，真 PASS） |
+| 3 (080) | blocked by user auth | ❌ **证伪**：`auth status` 看到 bot ready → **主动选 bot 身份** → 建群✓（`ok:true`）→ 发卡片✓（`ok:true`）。**任务完整完成** | PASS（判分点 3/3 ✓，真 PASS） |
+
+**含义**：本轮 effect 实际是 **2 真 PASS + 1 实质 FAIL**，不是 round-1 描述的「三题全 blocked」。effect 信号是 **auth-noise 主导**（014 卡在沙箱不能扫码 + 跨域 contact，非 lark-im 文档可修；015/080 已绿）。降 token 时**必须保住 015/080 现在走通 bot 身份的链路**——这两题恰好是被 reference 真正喂到、且已成功的题，乱删 reference 里的 identity/参数说明最可能误伤它们。
+
+## 跨 case 共同根因（优先看；按对 TOKEN 目标的杠杆排序）
+
+### RC-1（token，头号抓手，3 题全命中、最稳）—— SKILL.md `## Important Notes` + Shortcuts 全表常驻，本轮任务低命中
+- **现象**：SKILL.md 经 Skill 工具每题必加载（reach=1.0），实测 3,751 tok/题、三题一致（常驻静态）。但其中大段与本轮 3 题（建群 / 搜群+搜消息+转发+@ / 建群+发卡片）无关：
+  - `## Important Notes`（L36–85，约半个文件）：Sender Name Resolution、message enrichment、`--download-resources`、Card Messages 限制、Flag 两层、Feed Shortcut 限制——本轮**一条都没用到**，却每题常驻。
+  - `## Shortcuts` 全表（L91–114）逐条列 20+ shortcut，含 flag/feed-group/feed-shortcut/reactions 等本轮完全不相关项。
+- **可信度=常驻静态**：tiktoken 可测、跨题稳定（3,751×3）。这是降 token 最稳的发力点，且 3 题全命中（reach=1.0），降幅不被任何子集稀释。
+- **axis=token**。文档位置：`skills/lark-im/SKILL.md` 的 `## Important Notes` 低命中小节 + `## Shortcuts` 全量表。
+- **方向张力（必须标注）**：这是 round-1 已经动过一刀的同一文件（折叠了 API Resources/权限表）。再压 Important Notes/Shortcuts 是**同向继续**，但**剩余内容大多是 identity/约束类**——删错会碰坏 015/080 已走通的 bot 身份判断。candidate-writer 取舍时这是 effect 风险点，不是 RC-1 不成立。
+
+### RC-2（token，次级抓手，080 命中、按需读取）—— `messages-send.md` 单文件偏大且内部高度冗余
+- **现象**：080 读了 `messages-send.md`，实测 **5,365 tok**——本轮所有按需 reference 里最大的单块（占 080 visible 的 24.8%）。该 reference 实测被读且**确实用上了**（080 据此发卡片成功），不是「读了没用」。
+- **从文档看为何这么大**：messages-send.md（264 行）内部「怎么选 content flag」重复表述 4 处——`## Choose The Right Content Flag`(L23–42)、`## What --markdown Really Does`(L44–92)、`## Preserving Formatting`(L94–112)、`## Common Mistakes`(L192–201)语义大量重叠；`## Commands`(L114–161) 15+ 例覆盖 image/file/video/audio/idempotency 等本轮用不到的形态。这是「单文件冗余 + 全形态罗列」，不是信息缺失。
+- **可信度=按需读取**：只在实读它的子集（reach=0.333，仅 080）里计入，压缩降幅在该子集不被稀释——但**子集只有 1 题**，证据基数小，效果需评测确认（见数据缺口）。
+- **axis=token**。文档位置：`skills/lark-im/references/lark-im-messages-send.md`。
+
+### RC-3（token，次级抓手，014+080 命中、按需读取）—— `chat-create.md` 按需读取偏大
+- **现象**：014 与 080 都读了 `chat-create.md`，实测 3,060–3,062 tok（reach=0.667）。080 据此建群成功（用上了）；014 读后因 user 授权阻断没走到建群（读了但本题没用上）。
+- **可信度=按需读取**（reach=0.667，子集 2 题）。体积本身不离群，杠杆低于 RC-2，列为更次级。
+- **axis=token**。文档位置：`skills/lark-im/references/lark-im-chat-create.md`。
+
+### RC-4（效果，无文档根因 / 本轮不可修）—— 014 的 user 授权阻断 + 跨域 contact 依赖
+- **现象**：014 需先解析「傅一铭/傅二铭」open_id，走 `contact +search-user`（**lark-contact 域，不在 candidate_modules**）：bot 身份 exit2（invalid_argument）→ `--as user` token_missing → 发起 `auth login`+qrcode → 停在扫码。判分点证据全 ✗。
+- **归因落点**：根因=沙箱不能交互扫码（环境）＋ 跨域 contact 命令不可用（非 lark-im）。**lark-im 文档侧无根因、无可修点**——这正是约束 3 的「无文档根因 / 本题不改」出口，不要为凑根因往 lark-im doc 上硬编。
+- **axis=效果**，标注**无文档根因 / 本轮不改**。effect 维持 baseline 即可，不要试图改路由让 014「修绿」（用户显式要本人身份解析联系人，改 bot 是 reward-hack）。
+
+## 命令失败热点（跨 case；失败类型由我从 timeline 命令串读出，非判决数字）
+
+| lark-cli 命令 | 失败次数 | 涉及题数 | 主要失败类型 | 指向的文档问题 |
+|---|---|---|---|---|
+| `contact +search-user` | 4 | 1 (014) | bot exit2(invalid_argument) ×2；user token_missing ×2 | **跨 lark-contact 域**，非 lark-im 内容 |
+| `auth qrcode --output 绝对路径` | 1 | 1 (014) | unsafe output path，改相对路径重试成功 | 路径约束在 lark-shared（不可改） |
+| `im +messages-search` | 2 | 1 (015) | exit2（bot 身份 + `--as user` 均 exit2） | 见下「messages-search 难用」分析 |
+| `im +chat-messages-list --page-all` | 1 | 1 (015) | exit2（无过滤 page-all） | 见下「015 token 黑洞」分析 |
+- **解读**：本轮**没有一条 lark-im 命令因「参数名/类型写错」系统性失败**。080 三条命令 0 失败；015 的失败集中在 `messages-search`（见下）。这意味着**没有 lark-im 侧的常规「报错/参数整形」工单**——与 RC-1/2/3 的 token 方向一致，本轮抓手是减体积不是补内容。
+
+### 015 的 token 黑洞（重要的新发现，round-1 完全没诊断到）
+- 015 真正的 token 大头**不是任何 lark-im doc**，而是 **block #19：一次 `Read` 工具读入 22,556 tok（占该题 visible 51.5%）**。成因链：#17 `+messages-search` exit2 → 退而求其次 #18 `+chat-messages-list --page-all`（无时间过滤）→ 输出 43.5KB 被持久化到文件 → agent `Read` 整个文件 → 22.5k tok 灌进上下文。后面又靠本地 `grep`(#27–33) 抠出「飞豆」两条。
+- **从文档角度**：`chat-messages-list.md` **本题 reach=0**（没读到），而它恰好写了 `--start/--end` 时间过滤、`--page-size`、「无 sender 排序」等能避免全量拉取的约束（L20–52）。SKILL.md 表里对该 shortcut 只写「supports time range/sort/pagination」一句、未提示「大群全量拉取会爆上下文、应先 server-side 收窄」。**这是一个真实的「该读没读 → 全量灌入」放大器**（约束 5 状态①：调用前从没读该 reference）。
+- **但这条对本轮目标是「方向张力」，不是干净的 token 抓手**：要避免全量灌入，文档侧只能**增加**收窄指引（前置或加 caution），这与「降 token」的常驻成本目标**方向相反**（见硬性约束 7 的冲突记录）。且 22.5k 黑洞是**单次工具输出**（单次输出可信度、单题、强烈依赖该群消息量），不是稳定常驻热点。**结论：列为观察项交评测裁决，不要当成 RC-1 那种干净抓手去推「前置 chat-messages-list」——很可能只增 token 不省。**
+
+## 可发现性时序（约束 5 三态；判「前置能不能救」的决定性证据）
+> 对每条相关 reference / `--help`，按相对首次失败调用的读取时序统计。`--help` 扫 Bash（本轮 3 题均未跑任何 `--help`）。
+
+| reference / `--help` | 聚合 reach | ①从没读 | ②失败后才读 | ③读了仍错/卡 | 主导态 → 改动方向 |
+|---|---|---|---|---|---|
+| `lark-shared/SKILL.md` | 1.0 | 0 | 0 | — | 三题调用前都读了；014 仍卡（环境，非内容）；不可改 |
+| `chat-create.md` | 0.667 | 0 | 0 | — | 080 调用前读→建群成功；014 调用前读→授权阻断（非 reference 错）。**非触达问题** |
+| `messages-send.md` | 0.333 | 0 | 0 | — | 080 调用前读→发卡片成功。**非触达问题** |
+| `chat-messages-list.md` | 0.0 | 1 (015) | 0 | — | ① **015 调用前从没读**→直接 `--page-all` 全量拉取→token 黑洞。触达缺口，但补它=增 token，与目标冲突（见上） |
+| `messages-search.md` | 0.0 | 1 (015) | 0 | — | ① 015 从没读 messages-search.md，直接猜 `+messages-search` ×2 → exit2。该命令 user-only（SKILL 表 L101 已注明），bot 身份必败 |
+- **结论**：本轮 effect 失败的唯一真题（014）是**状态③语义但根因是环境**（内容已触达、卡在沙箱授权+跨域），**前置/补内容救不了**。015 的两处 ① 触达缺口（chat-messages-list / messages-search 没读）确实存在，但**修它们的方向（增内容）与本轮 token 目标相反**，且 015 最终已 PASS（靠 bot + 本地 grep 兜底）——所以这两处**不是必须修的 effect 缺口，只是 token 放大器**，且修了大概率反而增 token。
+- **对 candidate-writer 的含义**：**本轮没有「该前置」的干净 case**。RC-1/2/3 都是「调用前已读、内容够用 → 减体积」的纯 token 减法，不涉及触达。不要被 015 的两处 ① 诱导去推前置——那会与目标背道而驰。
+
+## 方向冲突记录（硬性约束 7）
+- **减体积（RC-1/2/3，与 objective.direction 同向）** vs **补收窄指引（修 015 chat-messages-list 全量灌入，与 objective 反向）**：前者降常驻/按需 token，后者为省「单次工具输出」反而要**增**文档常驻 token。两者方向相反，**不可合并**。本轮目标是降 token，应取减体积一侧；015 的全量灌入作为观察项记录、不作为本轮要补的内容根因。
+
+## 差距台账复盘
+- 无（round 2，`discard-ledger.json` 为空，无已跑未采纳候选）。
+
+## 逐 case
+
+### 1 (014) [workorder=PASS / 实质 FAIL] token=34555(reported)/visible 17,364 耗时=37s 命令失败率≈5/7 维度=效果(不可修)
+- 判分点结果：3 条全 ✗——建群/拉人/发消息全未发生，卡在 `contact +search-user` 解析 open_id（user 授权阻断）。verdict=PASS 系聚合口径，按判分点证据当 FAIL 处理。
+- 命令失败：≈5/7。`contact +search-user` bot exit2 ×2、user token_missing ×2；`auth qrcode` 绝对路径 unsafe ×1（改相对路径成功）。**全部非 lark-im 命令的内容错误**。
+- 可发现性时序：调用前读了 SKILL.md(reach=1.0)+chat-create.md(3,062 tok)；失败在更上游的跨域 contact + 授权。**非 lark-im 触达问题**。
+- token 归因：SKILL.md 正文 3,751（常驻静态，21.6%）+ chat-create.md 3,062（按需，17.6%，本题没走到建群=读了没用上）+ 系统 Skill 列表注入 4,612（固定开销，不归因）。lark-cli 命令累计含多次短失败回显，单条都短、非热点。
+- 耗时归因：本题往返多（查联系人→切 contact→失败→auth status→授权→qrcode 重试）。多为授权链路 + 跨域固有串行 + 反应式重试（duration 弱信号，需多轮复现）。
+- 文档根因：效果=沙箱 user 授权 + 跨域 contact（环境，**无 lark-im 文档根因，本轮不改**）；token=SKILL.md 常驻（RC-1）+ chat-create.md 按需（RC-3）。
+
+### 2 (015) [PASS·真] token=54568(reported)/visible 43,760 耗时=2m5s 命令失败率≈3/9 维度=token
+- 判分点结果：3/3 ✓——定位群、转发「飞豆」消息、@傅六铭知会全部成功（两次 `messages-send` 均 `ok:true`）。**全程 bot 身份，无授权阻断**。
+- 命令失败：≈3/9。`+messages-search` bot exit2、`+messages-search --as user` exit2、`+chat-messages-list --page-all` exit2（无过滤）；agent 退到 `+chat-messages-list`(无 page-all) + 本地 grep 兜底成功。
+- 可发现性时序：① `messages-search.md` / `chat-messages-list.md` **调用前从没读**（reach=0），直接猜命令。messages-search 是 user-only（SKILL 表 L101 已注明）、bot 身份必败——agent 没看清就猜。
+- token 归因：**本题 token 大头不是 lark-im doc**，是 block #19 一次 `Read` 持久化文件 = **22,556 tok（51.5%，其他工具调用/返回）**，成因=`--page-all` 无过滤全量拉取→43.5KB→Read 灌入（单次输出可信度，强依赖该群消息量）。SKILL.md 正文 3,749（常驻）。lark-shared 3,749（跨 skill，不归因 lark-im）。
+- 耗时归因：本题最长(2m5s)，主因是 messages-search 连环失败→改用 page-all→大输出→多次本地 grep 抠数据的多轮往返（duration 弱信号；工具调用 16 raw32，明显高于 080，作旁证）。
+- 文档根因：token 黑洞的放大器=`chat-messages-list.md` 没被读到 + SKILL.md 表未提示大群应 server-side 收窄——但**补这条与降 token 目标相反**（方向张力，见上），列为观察项；本题已 PASS。常规 token 抓手仍是 RC-1（SKILL.md 减体积）。
+
+### 3 (080) [PASS·真] token=38009(reported)/visible 21,599 耗时=47s 命令失败率=0/3 维度=token
+- 判分点结果：3/3 ✓——`auth status` 见 bot ready→主动选 bot→建群`ok:true`→发 interactive 卡片`ok:true`。**任务完整完成，零命令失败**。
+- 命令失败：0/3。三条 lark-cli（auth status / chat-create / messages-send）全成功。
+- 可发现性时序：调用前读 SKILL.md + chat-create.md(3,060) + messages-send.md(5,365)，全部状态③（调用前已读且用上）。**无触达问题**。
+- token 归因：**本题是纯 token 抓手题**——读取 Skill 占 56.4%：messages-send.md 5,365（按需，最大单块，RC-2）+ SKILL.md 3,751（常驻，RC-1）+ chat-create.md 3,060（按需，RC-3）。三块 reference/SKILL 都实读且 RC-2 的 messages-send.md 确实用上了。系统 Skill 列表注入 4,612（固定开销，不归因）。
+- 耗时归因：47s，全部为正常建群+发卡片串行，无重试、无写后回查（无离群）。
+- 文档根因：无效果根因（已绿）；token=RC-2(messages-send.md 内部冗余) + RC-1(SKILL.md 常驻) + RC-3(chat-create.md)。**本题 token 杠杆最高且无 effect 风险**（命令全成功，减 reference 体积不碰已走通链路）。
+
+## 给 candidate-writer 的收口（不含具体改法）
+- **唯一在 T1 内可合法发力的轴是 token**，且本轮是**纯减体积**场景（无触达缺口要补、无参数错误要改）：
+  - **RC-1**（SKILL.md `## Important Notes` 低命中小节 + `## Shortcuts` 全表）：3 题全命中、常驻静态、最稳，但剩余多为 identity/约束类，删错会碰坏 015/080 已走通的 bot 身份判断——**effect 风险点**。
+  - **RC-2**（messages-send.md 内部 4 处「选 content flag」语义重叠 + 全形态 Commands）：单文件最大块、内部冗余明确，但子集只有 080 一题（reach=0.333），证据基数小、效果需评测确认。
+  - **RC-3**（chat-create.md 按需偏大）：杠杆最低，列为更次级。
+- **effect 不可在本轮 T1 内合法抬升**：014 是环境（沙箱不能扫码）+ 跨域 contact，无 lark-im 文档根因。015/080 已真 PASS。候选必须**保住 015/080 走通 bot 身份的 identity/参数说明**，降 token 时别误伤。
+- **不要推前置**：本轮没有「该前置」的干净 case。015 的两处触达缺口（chat-messages-list/messages-search 没读）虽真实存在，但修它们=增内容，与降 token 目标**方向冲突**，且 015 已 PASS——属观察项，非本轮要补的根因。
+- **缺失信息（doc_fix_hint 语气）**：SKILL.md 的 Important Notes/Shortcuts 全量罗列、本轮低命中却每题常驻；messages-send.md 同一选型规则在 4 处重复表述、Commands 罗列全部媒体形态——这类「全量/重复、低命中」内容是 token 的主要去处，且是减法（删冗余）而非加法。
+- **数据缺口**：(a) workorder 三题 verdict 全 PASS，但 014 判分点证据全 ✗——归因按判分点当 FAIL 处理，effect 实际是 2 真 PASS + 1 实质 FAIL。(b) RC-2/RC-3 子集小（messages-send.md 仅 080、chat-create.md 仅 014+080），单轮证据基数小，token 降幅需评测在子集上确认。(c) 015 的 22.5k 黑洞是单次工具输出，强依赖该群消息量，非稳定常驻热点，单题不可外推。(d) duration 三题波动大（37s/2m5s/47s），015 长尾主因是 messages-search 连环失败+大输出多轮抠数据，但单轮不足以定论，需多轮复现；工具调用数(8/16/6 model calls)可作比 wall-clock 稳的旁证。(e) 工具调用次数 session-analyze(model calls 8/16/6) 与 workorder 趋势表(R1 均值 26.3) 口径不一致，趋势表疑似含 raw 计数，旁证以 timeline 实际往返为准。

harness-opt/rounds/round-002/discard-ledger.json

@@ -0,0 +1 @@
+[]

harness-opt/rounds/round-002/failure-memory.json

@@ -0,0 +1 @@
+[]

harness-opt/rounds/round-002/module-reach.json

@@ -0,0 +1,220 @@
+{
+  "skills/lark-im/SKILL.md": {
+    "reach": 1.0,
+    "read_cases": [
+      "1",
+      "2",
+      "3"
+    ],
+    "actual_cases": [
+      "1",
+      "2",
+      "3"
+    ],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": true
+  },
+  "skills/lark-im/references/lark-im-chat-create.md": {
+    "reach": 0.667,
+    "read_cases": [
+      "1",
+      "3"
+    ],
+    "actual_cases": [
+      "1",
+      "3"
+    ],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-chat-identity.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-chat-list.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-chat-messages-list.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-chat-search.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-chat-update.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-feed-group-list-item.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-feed-group-list.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-feed-group-query-item.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-feed-groups.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-feed-shortcut-create.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-feed-shortcut-list.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-feed-shortcut-remove.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-flag-cancel.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-flag-create.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-flag-list.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-message-enrichment.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-messages-mget.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-messages-reply.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-messages-resources-download.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-messages-search.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-messages-send.md": {
+    "reach": 0.333,
+    "read_cases": [
+      "3"
+    ],
+    "actual_cases": [
+      "3"
+    ],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-reactions.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-threads-messages-list.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  }
+}

harness-opt/rounds/round-002/review.json

@@ -0,0 +1,15 @@
+{
+  "generated_by": "lark-cli-harness:opt-reviewer",
+  "verdict": "PASS",
+  "module": "skills/lark-im/references/lark-im-messages-send.md",
+  "tier": "T1",
+  "reason": "纯结构性去重：16407→6399 字节(-61%)与策略一致；逐项核对每条承重指令(互斥规则、video-cover 必配、cwd-relative/绝对路径拒绝、markdown→post 边界、三套 <at> 语法、content 各 msg_type 样例、Safety Constraints、identity+scope 映射)均原样保留在新文档内联，删的全是重复/过度罗列(4× 选型规则、镜像 --help 的 Parameters 表、Common Mistakes、Notes、冗余 Commands)。无硬编码评测答案、未针对 080 卡片流窄化、未碰 SKILL.md 身份路由、单文件单根因。",
+  "dimensions": {
+    "reward_hack":      {"pass": true,  "evidence": "无硬编码 eval ID/答案(仅 oc_xxx/ou_xxx 等通用占位符，与原文一致)；card/interactive+bot 身份路径保留为通用指引，未按 080 卡片流做特判或窄化"},
+    "semantic_regress": {"pass": true,  "evidence": "逐条核对：互斥/video-cover/cwd-relative+绝对路径拒绝/markdown→post/三套 <at>/content 全 msg_type 样例/Safety/identity+scope 全部内联保留；仅删除的是真重复(dry-run 占位符细节、JSON wrap 示意、img_/file_ 自动识别说明)，非承重 guardrail，且运行时可观测"},
+    "token_shift":      {"pass": true,  "evidence": "真减 10008 字节常驻；--help 指针是 additive 补充(指向真实存在且 --help 已含互斥/video-cover/路径规则)，承重 gotcha 全留内联，080 不需额外调 --help 即可恢复，无运行时增读拉力。注:work-order 提的 schema im.messages.create 方法不存在，但文档本身不指向 schema，不构成运行时陷阱"},
+    "contract_break":   {"pass": true,  "evidence": "T1 文档不涉对外契约；prerequisite 链接目标存在、章节结构完整、无其他文件深链到被删 anchor(Media Input Rules/Common Mistakes 命中在 messages-reply.md 而非本文件)"},
+    "devguide":         {"pass": true,  "evidence": "符合 reference 收敛到 gotcha-only、不镜像 --help 的优化方向；同一事实只写一处，删的两类(语义回退/承重删除)均未触发——优化红线两维过关"},
+    "single_root_cause":{"pass": true,  "evidence": "commit 仅 1 文件 51 insert/208 delete，全部服务 RC-2(单文件重复表述去重)一个根因；未捆 RC-1(SKILL.md)/RC-3(chat-create)，未把无关删除以 token 对冲缝入"}
+  }
+}

harness-opt/rounds/round-002/round.json

@@ -0,0 +1,380 @@
+{
+  "round": 2,
+  "status": "admitted",
+  "parent_id": "a1333f2e1f7e98bf6f705814b92cacae1f43565759e4e0c24a0a4700b241649e",
+  "parent_worktree": "/Users/bytedance/Projects/cli",
+  "child_worktree": "/Users/bytedance/Projects/cli",
+  "base_commit": "51f2a70e6dffeea65d928badb6207408490dc215",
+  "module": "skills/lark-im/references/lark-im-messages-send.md",
+  "candidate_modules": [
+    "skills/lark-im/SKILL.md",
+    "skills/lark-im/references/lark-im-chat-create.md",
+    "skills/lark-im/references/lark-im-chat-identity.md",
+    "skills/lark-im/references/lark-im-chat-list.md",
+    "skills/lark-im/references/lark-im-chat-messages-list.md",
+    "skills/lark-im/references/lark-im-chat-search.md",
+    "skills/lark-im/references/lark-im-chat-update.md",
+    "skills/lark-im/references/lark-im-feed-group-list-item.md",
+    "skills/lark-im/references/lark-im-feed-group-list.md",
+    "skills/lark-im/references/lark-im-feed-group-query-item.md",
+    "skills/lark-im/references/lark-im-feed-groups.md",
+    "skills/lark-im/references/lark-im-feed-shortcut-create.md",
+    "skills/lark-im/references/lark-im-feed-shortcut-list.md",
+    "skills/lark-im/references/lark-im-feed-shortcut-remove.md",
+    "skills/lark-im/references/lark-im-flag-cancel.md",
+    "skills/lark-im/references/lark-im-flag-create.md",
+    "skills/lark-im/references/lark-im-flag-list.md",
+    "skills/lark-im/references/lark-im-message-enrichment.md",
+    "skills/lark-im/references/lark-im-messages-mget.md",
+    "skills/lark-im/references/lark-im-messages-reply.md",
+    "skills/lark-im/references/lark-im-messages-resources-download.md",
+    "skills/lark-im/references/lark-im-messages-search.md",
+    "skills/lark-im/references/lark-im-messages-send.md",
+    "skills/lark-im/references/lark-im-reactions.md",
+    "skills/lark-im/references/lark-im-threads-messages-list.md"
+  ],
+  "module_reach": {
+    "skills/lark-im/SKILL.md": {
+      "reach": 1.0,
+      "read_cases": [
+        "1",
+        "2",
+        "3"
+      ],
+      "actual_cases": [
+        "1",
+        "2",
+        "3"
+      ],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": true
+    },
+    "skills/lark-im/references/lark-im-chat-create.md": {
+      "reach": 0.667,
+      "read_cases": [
+        "1",
+        "3"
+      ],
+      "actual_cases": [
+        "1",
+        "3"
+      ],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-chat-identity.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-chat-list.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-chat-messages-list.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-chat-search.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-chat-update.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-feed-group-list-item.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-feed-group-list.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-feed-group-query-item.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-feed-groups.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-feed-shortcut-create.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-feed-shortcut-list.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-feed-shortcut-remove.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-flag-cancel.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-flag-create.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-flag-list.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-message-enrichment.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-messages-mget.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-messages-reply.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-messages-resources-download.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-messages-search.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-messages-send.md": {
+      "reach": 0.333,
+      "read_cases": [
+        "3"
+      ],
+      "actual_cases": [
+        "3"
+      ],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-reactions.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-threads-messages-list.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    }
+  },
+  "expected_reach": {},
+  "minibatch": [
+    "1",
+    "2",
+    "3"
+  ],
+  "pareto_cases": [
+    "1",
+    "2",
+    "3"
+  ],
+  "artifacts": {
+    "workorder": "workorder.md",
+    "diagnosis": "diagnosis.md",
+    "attribution": "attribution.json",
+    "strategy": "strategy.md",
+    "review": "review.json",
+    "trend": "trend.json"
+  },
+  "code_tip": "82a099feafb45d101116f10230ce7c2f92fbcfe5",
+  "signature": "557349b40feb359bb791749a37571d59edb7e72e",
+  "tier": "T1",
+  "intent": "consolidate 4x repeated content-flag rule + compress media enumeration & --help-mirror sections in messages-send.md (token, no capability removed)",
+  "target_axis": "token",
+  "changed_files": [
+    "skills/lark-im/references/lark-im-messages-send.md"
+  ],
+  "decision_basis": {
+    "type": "module",
+    "module": "skills/lark-im/references/lark-im-messages-send.md"
+  },
+  "decision_cases": [
+    "3"
+  ],
+  "review": {
+    "generated_by": "lark-cli-harness:opt-reviewer",
+    "verdict": "PASS",
+    "module": "skills/lark-im/references/lark-im-messages-send.md",
+    "tier": "T1",
+    "reason": "纯结构性去重：16407→6399 字节(-61%)与策略一致；逐项核对每条承重指令(互斥规则、video-cover 必配、cwd-relative/绝对路径拒绝、markdown→post 边界、三套 <at> 语法、content 各 msg_type 样例、Safety Constraints、identity+scope 映射)均原样保留在新文档内联，删的全是重复/过度罗列(4× 选型规则、镜像 --help 的 Parameters 表、Common Mistakes、Notes、冗余 Commands)。无硬编码评测答案、未针对 080 卡片流窄化、未碰 SKILL.md 身份路由、单文件单根因。",
+    "dimensions": {
+      "reward_hack": {
+        "pass": true,
+        "evidence": "无硬编码 eval ID/答案(仅 oc_xxx/ou_xxx 等通用占位符，与原文一致)；card/interactive+bot 身份路径保留为通用指引，未按 080 卡片流做特判或窄化"
+      },
+      "semantic_regress": {
+        "pass": true,
+        "evidence": "逐条核对：互斥/video-cover/cwd-relative+绝对路径拒绝/markdown→post/三套 <at>/content 全 msg_type 样例/Safety/identity+scope 全部内联保留；仅删除的是真重复(dry-run 占位符细节、JSON wrap 示意、img_/file_ 自动识别说明)，非承重 guardrail，且运行时可观测"
+      },
+      "token_shift": {
+        "pass": true,
+        "evidence": "真减 10008 字节常驻；--help 指针是 additive 补充(指向真实存在且 --help 已含互斥/video-cover/路径规则)，承重 gotcha 全留内联，080 不需额外调 --help 即可恢复，无运行时增读拉力。注:work-order 提的 schema im.messages.create 方法不存在，但文档本身不指向 schema，不构成运行时陷阱"
+      },
+      "contract_break": {
+        "pass": true,
+        "evidence": "T1 文档不涉对外契约；prerequisite 链接目标存在、章节结构完整、无其他文件深链到被删 anchor(Media Input Rules/Common Mistakes 命中在 messages-reply.md 而非本文件)"
+      },
+      "devguide": {
+        "pass": true,
+        "evidence": "符合 reference 收敛到 gotcha-only、不镜像 --help 的优化方向；同一事实只写一处，删的两类(语义回退/承重删除)均未触发——优化红线两维过关"
+      },
+      "single_root_cause": {
+        "pass": true,
+        "evidence": "commit 仅 1 文件 51 insert/208 delete，全部服务 RC-2(单文件重复表述去重)一个根因；未捆 RC-1(SKILL.md)/RC-3(chat-create)，未把无关删除以 token 对冲缝入"
+      }
+    }
+  },
+  "child_k": 5,
+  "eval_trace": null,
+  "retro": {
+    "cause": "已入池",
+    "noise_borderline": false,
+    "summary": "越带入池，无需复盘补发"
+  },
+  "retro_sessions": [
+    {
+      "case": "3",
+      "session": null,
+      "axis": "token",
+      "expect": "降",
+      "parent": 37942,
+      "child": 35478,
+      "gain": "收益现",
+      "pass_delta": null
+    }
+  ],
+  "verdict": "admitted",
+  "ci": null,
+  "new_candidate": "557349b40feb359bb791749a37571d59edb7e72e",
+  "decision": {
+    "parent_success": 1.0,
+    "child_success": 1.0,
+    "parent_score": 0.6,
+    "child_score": 1.0,
+    "score_saved": 0.4,
+    "score_threshold": 0.09532271373123208,
+    "parent_token": 37942.0,
+    "child_token": 35478.0,
+    "saved": 2464.0,
+    "threshold": 4532.708313776408,
+    "parent_duration": 45769.0,
+    "child_duration": 46540.0,
+    "dur_saved": -771.0,
+    "dur_threshold": 4899.200953624988,
+    "dur_margin": 1.0,
+    "missing_duration": [],
+    "k_child": 5,
+    "k_parent": 5,
+    "decision_n": 1,
+    "missing_context": [],
+    "missing_score": [],
+    "parent_token_acc": 251669.0,
+    "child_token_acc": 221685.0,
+    "phi0_score": 0.5333333333333333,
+    "eff_margin": 1.0,
+    "parent_token_full": 37942.0,
+    "child_token_full": 35478.0,
+    "saved_full": 2464.0,
+    "observe_n": 1,
+    "target_axis": "token",
+    "admitted": true,
+    "reason": "score_gain"
+  },
+  "patch": "verify_results/round-002-lark-im-references-lark-im-messages-send.patch"
+}

harness-opt/rounds/round-002/strategy.md

@@ -0,0 +1,48 @@
+# Round 2 候选策略（模块=references/lark-im-messages-send.md, tier=T1, 主指标=token）
+
+## 根因与选择
+
+| 根因 | 来源(评测归因/规范经验) | 承载模块(reach) | annotation 风险级 | coverage 档 | P级 | 选中 |
+|---|---|---|---|---|---|---|
+| RC-2: messages-send.md 单文件最大、内部「选 content flag」规则重复 4 处 + 全媒体形态罗列 + Parameters/Notes 镜像 --help | 评测归因①（080 实读实用）＋规范经验②（annotation R1×140/R2×109，仅 1 段 R3） | references/lark-im-messages-send.md (0.333) | R1/R2 主导，唯一 R3=Safety Constraints(L9–22) | 密 / overfit 低 | P1 | ✅ |
+| RC-1: SKILL.md `## Important Notes` 低命中 + `## Shortcuts` 全表常驻 | 评测归因①（reach=1.0，3 题全命中） | SKILL.md (1.0) | R2/R3 混合（identity/约束密集） | 密 / 中 | P0(命中) 但 effect 高风险 |  |
+| RC-3: chat-create.md 按需偏大 | 评测归因① | references/lark-im-chat-create.md (0.667) | — | 密 | P1 |  |
+
+- **选中理由**：RC-2 是诊断点名「最干净的 token 杠杆」——单文件最大块（实测 ~5,365 tok，占 080 visible 24.8%），且 080 调用前已读、确实据它发卡片成功（reach=0.333、actual=1，非「读了没用」）。annotation 证实它 R1/R2 主导（140 R1 + 109 R2 行，可重构/可压缩），唯一 R3 段是 Safety Constraints(L9–22)，我**原样保留语义**。coverage=「密」、overfit「低」→ 本轮 eval 能在 080 上裁真伪。这是纯减体积、零能力删除、不碰 SKILL.md 路由的改动。
+- **为什么不选 RC-1**：reach=1.0、命中率最高，但 diagnosis 明确标它为 **effect 风险点**——剩余内容多为 identity/约束类，正是驱动 015/080 走通 bot 身份判断的承重内容；objective 的**硬门槛是「保住成功率」**，动 SKILL.md 最可能误伤这条已绿链路。本轮放弃，避免拿成功率换 token。
+- **为什么不选 RC-3**：diagnosis 判其杠杆最低（体积不离群），列为更次级；同一根因一轮只动一个，留待后续轮次。
+- **选模块理由**：messages-send.md reach=0.333>0（满足 reach 锁），承载选中的 RC-2，是非域 reference、改它不触碰 SKILL.md 的身份路由面。多文件无——本轮只动这一个文件。
+- **规范经验源补注**：对照 content-taxonomy——「单命令用法/长示例/与 --help 重复」类默认 R0/R1，「一般行为规则/CLI 机制约定」默认 R2；本文件的重复选型规则、全形态 Commands、Parameters/Notes 镜像即此类，处理方向为「留命中率最高一处，其余删或指针」「高频留 2–3 例，长的下沉」。当轮可被 080 裁真伪（coverage 密/overfit 低）。
+
+## 改了什么（逐处）
+- **L23–43 `## Choose The Right Content Flag` + `### --text vs --markdown`**：两段语义重叠的选型说明 → 合并为单张 4 行选型表（markdown/text/content/media），并把互斥规则并入表后一句。删掉 `### --text vs --markdown` 整段（与表重复）。
+- **L44–82 `## What --markdown Really Does` + `### Markdown Boundaries` + `### Image Constraint`**：三段约 39 行 → 压成 `## --markdown Gotchas` 三条要点（强制 post/无 title、标题改写规则、图片预上传 vs 远程 URL vs 本地路径不支持）。删掉 JSON wrap 示意、逐条 boundary 罗列等可由行为观察得到的展开。
+- **L83–93 图片预上传双命令示例**：并入 `## Commands` 的一条 markdown+image 示例（保留 `im images create` → 引用 img_xxx 的关键两步）。
+- **L114–161 `## Commands`（15+ 例覆盖全媒体形态）+ `## Media Input Rules`**：压成代表性示例（markdown / text / DM / post-title / markdown+image / 4 个媒体一组 / idempotency+dry-run），媒体路径规则收成 `--help` 指针后的 3 条 load-bearing gotcha（cwd-relative/绝对路径拒绝、video-cover 必配、msg-type 推断冲突）。
+- **L169–191 `## Parameters` 表**：删除镜像 `--help` 的逐参数描述，改为「Run `lark-cli im +messages-send --help`」指针 + 仅保留 --help 不显然的三条硬规则（已并入 Commands 末尾）。
+- **L192–202 `## Common Mistakes`**：整段删除——逐条都是选型表/markdown gotcha 的反向重述（第 4 次重复选型规则），删后选型信息仍在表里。
+- **L203–216 `## content Format Reference`**：保留（构造 `--content` 的 gotcha），把 image/file/audio 三行合并为一行省重复。
+- **L227–248 `## @Mention Format`**：保留全部三种 msg_type 的 `<at>` 语法（text/post/interactive 各异、AI 猜不到），压紧为两条要点、去掉小标题与重复散文。
+- **L249–264 `## Notes`**：整段删除——逐条（互斥/media 上传/scope/markdown 强制 post/video-cover/msg-type 冲突）均已在 Safety Constraints、选型表、--markdown Gotchas、Commands 指针处各保留一处单一事实源。
+
+## 为什么这么改（机制）
+- **消除根因的因果链**：该 reference 的体积来自「同一份选型规则在 4 个 section 重复 + 全媒体形态逐条罗列 + Parameters/Notes 镜像 --help」。token 不是被任务必需信息占用，而是被**重复表述**占用。按「同一份事实只写一次」（锚点 1）合并到单一事实源后，每条 load-bearing 信息仍恰好出现一次，080 这类「读该 reference→发消息」的题，读入 token 直接下降而行为不变。
+- **不删能力**：每个 flag（text/markdown/content/image/file/video/audio/idempotency/dry-run/msg-type/video-cover/as）、每条硬约束（互斥、video-cover 必配、cwd-relative 路径、绝对路径拒绝、markdown 强制 post/无 title、msg-type 冲突校验）、三套 `<at>` 语法、content 各 msg_type 样例、Safety Constraints、identity+scope 映射——全部保留，只是从「重复 N 次/逐条罗列」变成「一处/代表性示例 + --help 指针」。
+- **规范经验源**：依 optimization-playbook「reference 收敛到 gotcha-only，不做 --help 镜像」——Parameters 全表/全形态 Commands 属 USAGE，下沉到 `--help` 指针；保留的是 --help 表达不了的跨 flag 互斥、媒体路径安全、markdown→post 边界、@mention 按类型差异等 gotcha。annotation 标这些段为 R1（可重构/下沉），符合处理方向；唯一 R3（Safety）原样保留。
+
+## 预期效果
+- **成功率**：不退化。080（唯一读该文件的题）的发卡片链路依赖的是 `--content`/`interactive`、identity=bot、chat-id——全部保留；选型表、content Format Reference、Safety、scope 都在。015/080 走通 bot 身份的判断由 SKILL.md + identity 段承载，本轮**没碰 SKILL.md**，零误伤面。014 与本文件无关（reach 不含 014）。
+- **context（分两层）**：
+  - (1) **静态字数差**：16,407 → 6,399 chars（-61.0%）；tiktoken cl100k 3,869 → 1,799 tok（-53.5%）。（注：diagnosis 报 ~5,365 tok 系另一 tokenizer/含注入开销；此处用 cl100k 自测，方向与幅度一致。）
+  - (2) **运行时 context 方向**：仅在**实读该 reference 的子集**生效——本轮即 080 一题，运行时读入下降约 50%+（该块占 080 visible 24.8%，预计 080 visible 降约 12–13%）。其余两题（014/015）不读该文件，运行时 token **不变**（既不增也不减）。这是按需 reference，不是常驻面，不会影响未读它的题。
+- **覆盖敞口**：RC-2 子集仅 080 一题（reach=0.333），证据基数小。coverage 判该文件「密/overfit 低」，本轮 eval 可在 080 上裁真伪，但单题不可外推到「所有发消息任务」。建议后续补「读 messages-send.md 后用 --markdown / 媒体 / @mention」的 case 加厚子集。预期收益落在 **token 轴**（080 visible 下降），effect 轴维持不退化。
+
+## 刻意没做什么（反 reward-hack / 反过拟合）
+- 没硬编码任何评测题答案；没删任何能力、flag、guardrail、身份/scope 说明；没碰 lark-im 以外文件，也没把无关根因捆进本轮（commit 仅 1 个文件）。
+- **没碰 SKILL.md（RC-1）**：尽管 reach=1.0 杠杆最大，但其剩余内容是驱动 015/080 bot 身份判断的承重 identity/约束，diagnosis 标为 effect 风险点；在「保住成功率」硬门槛下不拿成功率换 token。
+- **没补收窄/分页指引**（015 的 22.5k chat-messages-list 黑洞）：那是「增内容」，与降 token 目标方向相反，diagnosis 已列为观察项、本轮不做。
+- 本改动**不是按评测错误反推**的参数/路由拟合——是基于 annotation + content-taxonomy 的结构性去重，删的是重复表述而非按 080 的具体内容裁剪；真实价值在「任何读该 reference 的发消息任务都少读重复 token」，080 只是当轮可验证的子集。
+- 未发现需要 breaking（T3）才能根治的点；本轮纯 T1 文档去重即可。
+
+## 签名
+- signature: 557349b40feb359bb791749a37571d59edb7e72e (commit 82a099fe 的 diff hash)  tier: T1

harness-opt/rounds/round-002/trend.json

@@ -0,0 +1,11 @@
+[
+  {
+    "round": 1,
+    "n": 3,
+    "pass_n": 0,
+    "cmd_fail_rate": 0.6,
+    "tool_calls": 26.333333333333332,
+    "duration_ms": 50189.0,
+    "token": 31997.0
+  }
+]

harness-opt/rounds/round-002/workorder.md

@@ -0,0 +1,43 @@
+# Round 2 归因派工单（parent=a1333f2e1f7e98bf6f705814b92cacae1f43565759e4e0c24a0a4700b241649e；模块未定，由 candidate-writer 据诊断点名）
+
+> **只读输入**——opt-attributor 读本文件，把诊断**另写** `diagnosis.md`（给 candidate-writer）+ 逐题结构化 `attribution.json`（给 dashboard）。**不要覆盖本文件**，留作派工单↔诊断的前后对比。
+> 判分点只当「什么算挂」的锚，禁止照抄 grader 药方（已从派工单剔除）。
+
+## 模块运行时可达性（选模块第一步的证据；要选须在 strategy.md 说明理由）
+> reach=**实测**触达率（域主 SKILL.md 经 Skill 工具加载、reference 经 Read，都从 trace 实测，没有恒在的面）；判决集=实测∪预期触达。**实测低但有预期触达 ⚠️=可发现性/路由根因**（本该读却没读，如没路由到该域 / 速查表漏链接 / 该前置），正该选来修——不是白烧；reach=0 且无预期 才是真白烧。 **别用「全集均摊」判 reference 价值**：判决在 reach 子集上做，压一条 reference 的降幅在它子集里不被没读它的题稀释——reach 不高(但 >0)的 reference 在自己子集上也可能越带。
+- `skills/lark-im/SKILL.md` → reach=1.0  [域主 skill·经 Skill 工具加载]；判决集(实测∪预期): ['1', '2', '3']
+- `skills/lark-im/references/lark-im-chat-create.md` → reach=0.667；判决集(实测∪预期): ['1', '3']
+- `skills/lark-im/references/lark-im-messages-send.md` → reach=0.333；判决集(实测∪预期): ['3']
+- （另 22 个 reference reach=0 且无预期触达，本轮无关，略）
+
+## 逐轮诊断信号趋势（纯诊断，不进判决）
+
+| 轮 | 题数 | PASS | 命令失败率 | 工具调用 | 耗时(ms) | token |
+|---|---|---|---|---|---|---|
+| R1 | 3 | 0 | 0.60 | 26 | 50189 | 31997 |
+
+> 跨题均值，按轮排。**命令失败率、工具调用数是横切诊断信号，不是准入轴**（准入只走 效果/token/耗时）——用来判「上一轮那刀有没有把失败/轮次压下去」。工具调用数比 wall-clock 稳，可给噪声大的耗时轴当旁证。
+
+### 1  [PASS]  ctx=34270  (acc=274608)  43995ms  tools=31
+- session.jsonl: harness-opt/rounds/round-001/child-runs/run-1/detail_info/cases/CLI_核心评测_014/0/session.jsonl [native]
+- 判分点（grader 的「什么算挂」oracle，非药方）:
+    ✗ 使用当前用户身份创建名为「IM合作群」的群聊
+        证据: transcript 在展示授权二维码后结束，无任何 `lark-cli im +chat-create` 调用。执行停在 '授权完成后请告诉我，我会继续帮你创建群聊并发送消息'，群聊未创建。
+    ✗ 将傅一铭和傅二铭加入该群
+        证据: transcript 显示尝试搜索用户时遇到 `need_user_authorization` 错误，授权流程启动后中断。未获取到任何用户的 open_id，无后续添加操作。
+    ✗ 在该群发送文本消息「大家体验有问题随时沟通」，并返回可验证的 chat_id / message_id
+        证据: 群聊未创建，无 chat_id 可返回。transcript 无任何 `lark-cli im messages-send` 调用。
+
+### 2  [PASS]  ctx=47116  (acc=612048)  114310ms  tools=49
+- session.jsonl: harness-opt/rounds/round-001/child-runs/run-1/detail_info/cases/CLI_核心评测_015/0/session.jsonl [native]
+- 判分点（grader 的「什么算挂」oracle，非药方）:
+    ✓ 成功定位名为「fusanming_at_openclaw群」的群，并获取最近包含「飞豆」关键字的消息
+    ✓ 将筛选出的相关消息内容转发到「fusanming_at_需求测试群」
+    ✓ 在「fusanming_at_需求测试群」中 @傅六铭 做知会，消息发送成功
+
+### 3  [PASS]  ctx=37942  (acc=251669)  45769ms  tools=23
+- session.jsonl: harness-opt/rounds/round-001/child-runs/run-1/detail_info/cases/CLI_核心评测_080/0/session.jsonl [native]
+- 判分点（grader 的「什么算挂」oracle，非药方）:
+    ✓ 使用用户身份创建一个名为「今晚吃什么」的群，预期返回 chat_id
+    ✓ 创建一张飞书卡片，卡片内容包含「今天晚上吃什么」
+    ✓ 将该卡片发送到新建群中，预期返回 message_id

harness-opt/rounds/round-003/attribution.json

@@ -0,0 +1,59 @@
+[
+  {
+    "case_id": "1",
+    "verdict": "PASS",
+    "verdict_note": "workorder=PASS（聚合口径）；判分点证据 3/3 ✗，按判分点当实质 FAIL 处理",
+    "token": 34555,
+    "duration_ms": 37000,
+    "tool_calls": 31,
+    "cmd_attempts": 7,
+    "cmd_failures": 5,
+    "cmd_fail_rate": 0.71,
+    "discoverability_state": "无（失败命令全是跨域 contact + auth，非 lark-im；chat-create.md 调用前已读但未走到使用）",
+    "axis": "效果",
+    "root_cause": "沙箱 user 授权不可完成 + 跨域 lark-contact 命令依赖；无 lark-im 文档根因，本轮不改",
+    "token_hotspot": "运行时冗余清单常驻（SKILL.md 3,456）+ 按需 chat-create.md 3,062（读了没用上）；lark-shared 3,751 与系统 Skill 列表注入 4,612 均不归因",
+    "token_reliability": "常驻静态（SKILL.md）/ 按需读取（chat-create.md，本题读了没用上）",
+    "duration_hotspot": "多轮交互（查联系人→切 contact→失败→授权→qrcode 重试）+ 纯外部API延迟(部分不可归因)",
+    "duration_reliability": "耗时波动大，单次运行不算数，需多题或多次复现",
+    "doc_fix_hint": "效果侧无 lark-im 文档缺信息（环境+跨域）；token 侧 chat-create.md 把同组 flag 在 Commands/Usage Scenarios 重复演示、Common Errors 复述 validation 字符串，属可删冗余"
+  },
+  {
+    "case_id": "2",
+    "verdict": "PASS",
+    "verdict_note": "真 PASS，判分点 3/3 ✓，全程 bot 身份无授权阻断",
+    "token": 54568,
+    "duration_ms": 125000,
+    "tool_calls": 49,
+    "cmd_attempts": 11,
+    "cmd_failures": 3,
+    "cmd_fail_rate": 0.27,
+    "discoverability_state": "① 从没读（messages-search.md / chat-messages-list.md reach=0，直接猜命令；本题未读任何 lark-im reference）",
+    "axis": "token",
+    "root_cause": "无过滤 +chat-messages-list --page-all 全量拉取 → 43.5KB 输出被 Read 整文件灌入 22,556 tok；token 大头非 lark-im doc。修它需补收窄/前置内容，与降 token 目标方向冲突，列观察项",
+    "token_hotspot": "工具返回原样输出（block #19 单次 Read 22,556 tok / 51.5%，归「其他工具调用/返回」）",
+    "token_reliability": "单次输出（强依赖该群消息量，单题不可外推，非稳定常驻热点）",
+    "duration_hotspot": "多轮交互 + 重试（messages-search 连环 exit2 → page-all → 大输出 → 多次本地 grep 抠数据）",
+    "duration_reliability": "耗时波动大，单次运行不算数，需多题或多次复现（model calls 16 作旁证，明显高于 080）",
+    "doc_fix_hint": "本题无 T1 可发力的 token 抓手（大头是单次工具输出，非 lark-im doc 常驻）；缺的是大群消息查询的 server-side 收窄指引，但补它=增内容、与降 token 反向，不作本轮根因"
+  },
+  {
+    "case_id": "3",
+    "verdict": "PASS",
+    "verdict_note": "真 PASS，判分点 3/3 ✓，主动选 bot 身份建群+发卡片均 ok:true，零命令失败",
+    "token": 38009,
+    "duration_ms": 47000,
+    "tool_calls": 22,
+    "cmd_attempts": 3,
+    "cmd_failures": 0,
+    "cmd_fail_rate": 0.0,
+    "discoverability_state": "无（无失败命令；SKILL.md + chat-create.md + messages-send.md 全部状态③：调用前已读且用上）",
+    "axis": "token",
+    "root_cause": "读取 Skill 占 56.4%；本轮唯一干净 token 抓手 = chat-create.md 内部冗余（示例罗列 + 场景重复 + --help 镜像），从未被优化过",
+    "token_hotspot": "运行时冗余清单常驻 + 按需 reference（chat-create.md 当前 2,336 raw，可压 Commands/Usage Scenarios 重叠 + Common Errors validation 镜像；trace 里 messages-send.md 5,365 是旧版，round-2 已压到 2,006，本轮不再可压）",
+    "token_reliability": "按需读取（chat-create.md reach=0.667，本题是其压缩收益唯一稳态兑现题）",
+    "duration_hotspot": "无离群（建群+发卡片正常串行，无重试、无写后回查）",
+    "duration_reliability": "耗时波动大，单次运行不算数，需多题或多次复现",
+    "doc_fix_hint": "chat-create.md 把同组 flag 在 Commands(12 例)+Usage Scenarios(3 场景)重复演示、Common Errors 多行复述 --help/报错本身就会吐的 validation 字符串，属可删冗余；232043 两步流 / --chat-mode topic 区分 / --owner 默认为载重红线，压缩中不可误删"
+  }
+]

harness-opt/rounds/round-003/case-commands.json

@@ -0,0 +1,27 @@
+{
+  "1": [
+    "auth login",
+    "auth qrcode",
+    "contact +search-user"
+  ],
+  "3": [
+    "auth login",
+    "auth qrcode",
+    "auth status",
+    "im +chat-create",
+    "im +messages-send"
+  ],
+  "2": [
+    "auth login",
+    "auth qrcode",
+    "auth status",
+    "im +chat-messages-list",
+    "im +chat-search",
+    "im +messages-mget",
+    "im +messages-search",
+    "im +messages-send",
+    "im messages forward",
+    "schema im.messages.forward",
+    "schema im.messages.search"
+  ]
+}

harness-opt/rounds/round-003/child-cache.json

@@ -0,0 +1,20 @@
+{
+  "1": {
+    "score": 1.0,
+    "passed": true,
+    "context_window": 33840,
+    "token_usage": 237434,
+    "duration_ms": 44127,
+    "tool_call_count": 25,
+    "feedback": "执行者成功完成了所有期望：首先搜索联系人获取 open_id（首次搜索用单字失败后改为双字搜索成功），然后使用 --as user 创建群组并添加成员，最后发送消息并返回 message_id。整个流程正确，使用了等效的 `--as user` 身份，符合用户「使用我的身份」的要求。验证结果确认所有操作均已生效。"
+  },
+  "3": {
+    "score": 1.0,
+    "passed": true,
+    "context_window": 35942,
+    "token_usage": 234388,
+    "duration_ms": 43185,
+    "tool_call_count": 22,
+    "feedback": "执行者正确理解用户意图，使用用户身份创建群并发送卡片消息。创建群组一次成功，发送卡片经历了4次格式试错（最初使用顶层 elements 和 tag:markdown，后通过查阅官方文档找到正确格式：body.elements + div + lark_md），最终成功发送并返回 message_id。试错后自行纠正符合评判原则，不构成判罚依据。\n- {'reason': '建议在 lark-im-messages-send.md 中增加飞书 interactive card 的标准格式示例，特别是 2.0 schema 下的 body.elements 中使用 div + lark_md 的正确写法，减少 AI 试错成本'}\n- {'reason': '建议 CLI 在遇到 230099 卡片格式错误时，尝试解析并返回更具体的字段级错误提示（如提示 \"elements 应在 body 内\" 或 \"tag:markdown 不被支持\"），帮助 AI 更快定位问题'}"
+  }
+}

harness-opt/rounds/round-003/diagnosis.md

@@ -0,0 +1,119 @@
+# Round 3 归因（parent=557349b…（round-2 已采纳候选）；候选模块见 candidate_modules，由 candidate-writer 据诊断+reach 点名）
+
+> 目标（objective.json）：**在不回退成功率的前提下降低 lark-im skill 文档的 token 成本**。effect 是硬门槛、不可退化；token 与 duration 是并列成本杆。tier=T1，仅可改 `skills/lark-im/**`。target_axis=token。
+> 判分点只当「什么算挂」的锚，不抄 grader 药方。
+
+## ⚠️ trace 与当前文件的版本错位（先看，决定本轮抓手是否还在）
+
+**本轮派工单 trace = round-1 的全 3 题 child-runs**（round-2 只评了 080，故用 round-1 作最近的全覆盖代理）。这些 trace 里的 reference 体积是 **round-1/round-2 改动之前** 的旧版。我用 session-analyze 所用的同一 ai-tokenizer 实测了**当前工作树**文件，确认两者错位如下：
+
+| 文件 | trace 内体积（旧版，Read 计） | 当前实测（raw / Read 计） | 已被哪轮收割 |
+|---|---|---|---|
+| `SKILL.md`（Skill 注入正文） | 3,455–3,456 tok | 3,525 raw | round-1（API Resources/权限表→schema 指针） |
+| `references/lark-im-messages-send.md` | **5,365 tok** | **2,006 raw / 2,194 Read** | **round-2（5,365→2,006，已收割）** |
+| `references/lark-im-chat-create.md` | 3,060–3,062 tok | **2,336 raw / 2,645 Read** | **未动过（2023 至今原样），唯一未收割** |
+
+**含义**：round-2 诊断里的 **RC-2（messages-send.md 内部冗余）已经在 round-2 被采纳并收割**（5,365→2,006），它不再是本轮抓手——不要据 trace 里的 5,365 重复提一遍。本轮 trace 里那块 5,365 是历史值，当前已不存在。**reach>0 集合里唯一还没被压过的干净文件就是 `chat-create.md`**（round-2 的 RC-3）。
+
+## 跨 case 共同根因（优先看；按对 TOKEN 目标的杠杆排序）
+
+### RC-1（token，本轮头号且基本是唯一的干净抓手，reach=0.667：014+080）—— `chat-create.md` 内部存在「示例罗列 + 场景重复 + --help 镜像」三类可压缩冗余，且从未被优化过
+- **现象**：`chat-create.md` 当前 2,336 raw tok（Read 计 ~2,645），是 reach>0 集合里**唯一未被任何轮收割**的 reference。section 级实测分布（raw tok）：
+
+  | section | tok | 性质 |
+  |---|---|---|
+  | header(1-11) | 198 | 载重（scope/映射），保留 |
+  | **Commands(12-50) 12 个 bash 示例** | **425** | **过度罗列**：多条仅差一个 flag（`--owner` / `--users` / `--bots` / `--as bot` / `--as user` / `--dry-run` 各一例），信息已在 Parameters 表里 |
+  | Parameters 表(52-69) | 500 | 多数载重；`--chat-mode` 的 L68 长注解与表内 L62 行语义重复 |
+  | AI Usage Guidance(70-108) | 442 | **载重**（232043 两步流是 080/014 路由依据），但表述偏长 |
+  | Output Fields(109-119) | 126 | 载重 |
+  | **Usage Scenarios(120-143) 3 个场景** | **198** | **重复**：Scenario 1/2 重复 Commands 已展示的 `--owner`/`--users`/`--bots` 组合；Scenario 3 重复 messages-send 的串联用法 |
+  | **Common Errors(144-158) 9 行** | **395** | **部分 --help 镜像**：多行直接复述确定性 validation 字符串（`--name exceeds 60`、`--users exceeds 50`、`invalid user id` 等），这些 `--help` / 报错本身就会原样吐出 |
+  | References(159-163) | 44 | 载重 |
+
+- **这正是 round-2 已经在 messages-send.md 上验证过、且被采纳的同一套压缩模式**：round-2 把 messages-send.md 的「4 处重复选型规则 + 全媒体形态 Commands + --help 镜像」压成「保留载重规则 + 一句 `--help` 指针」（5,365→2,006，被采纳）。chat-create.md 的 Commands(425)↔Usage Scenarios(198) 重叠、Common Errors(395) 的 validation 镜像，是同型冗余。
+- **可压缩量级（粗估，非药方）**：可压缩质量集中在 Commands+Usage Scenarios 的重叠（合计 ~623 tok，去重后可省一大半）+ Common Errors 的 --help 镜像行。**保守估计可从 2,336 压到 ~1,500–1,700 raw tok（省约 600–800 tok，约 30%）**，与 messages-send.md 的压缩比同量级。具体改法与确切降幅由 candidate-writer 决定、评测裁决。
+- **载重红线（candidate-writer 取舍时的 effect 风险点，不是 RC-1 不成立）**：AI Usage Guidance 的 **232043 两步流 + `succeed_type=1`**、`--chat-mode topic` vs 普通群+话题消息模式的区分、`--owner` 默认行为，是 014/080 走通 bot 身份建群的语义依据，**不能在压缩中误删**。这条 reference 被 080 实读且 080 据它建群成功（`ok:true`），所以 effect 风险真实存在——压的是示例/场景/报错镜像的体积，不是语义规则。
+- **axis=token**。可信度=**按需读取**（reach=0.667，子集=014+080，2 题）。压它的降幅只在这 2 题子集里计入，不被 015（没读它）稀释；但子集仅 2 题、且 014 是「读了没用上」（授权阻断没走到建群），实际吃到压缩收益的稳态题只有 080 一题——**证据基数小，降幅需评测在子集上确认**（见数据缺口）。
+
+### RC-2（token，已收割，本轮不再是抓手）—— messages-send.md 的内部冗余 round-2 已压掉
+- round-2 RC-2 已被采纳：messages-send.md 5,365→2,006 raw。**本轮不要据 trace 里的 5,365 重复提**。当前 messages-send.md 已是「载重规则 + `--help` 指针」形态，无明显二次压缩空间（剩余多为 content-flag 选型、@mention、media 约束等载重内容）。reach=0.333（仅 080）。
+
+### RC-3（token，无 T1 干净抓手）—— SKILL.md 常驻正文 round-1 已压过，剩余多为载重 identity/路由
+- SKILL.md 经 Skill 工具每题必加载（reach=1.0），当前 3,525 raw tok（round-1 已把 API Resources/权限表折叠成 schema 指针）。剩余 `## Important Notes`(L36–85) 各小节（Sender Name Resolution / message enrichment / `--download-resources` / Card / Flag / Feed Shortcut）与 `## Shortcuts` 全表(L87–115) 虽本轮 3 题低命中，但它们是**全域 identity/路由/约束**——这是 round-1 已经动过一刀的同一文件，**再压属同向继续、但删错会碰坏 015/080 已走通的 bot 身份与命令路由判断**（effect 风险高于 RC-1）。**列为更次级、风险更高的抓手**，不作为本轮首选；若要动须只删本轮已确证低命中且非路由的纯枚举行，谨慎程度高于 chat-create。
+
+## 命令失败热点（跨 case；失败类型由我从 timeline 命令串读出，非判决数字）
+
+| lark-cli 命令 | 失败次数 | 涉及题数 | 主要失败类型 | 指向的文档问题 |
+|---|---|---|---|---|
+| `contact +search-user` | 4 | 1 (014) | bot exit2(invalid_argument) ×2；`--as user` token_missing ×2 | **跨 lark-contact 域**，非 lark-im 内容 |
+| `auth qrcode --output <绝对/沙箱外路径>` | 1 | 1 (014) | unsafe output path，改相对路径重试成功 | 路径约束在 lark-shared（不可改） |
+| `im +messages-search` | 2 | 1 (015) | bot exit2 + `--as user` exit2 | 该命令 user-only（SKILL 表已注明）；bot 身份必败，agent 没看清就猜 |
+| `im +chat-messages-list --page-all` | 1 | 1 (015) | exit2（无过滤 page-all） | 见下「015 token 黑洞」 |
+- **解读**：本轮**没有一条 lark-im 命令因「参数名/类型写错」系统性失败**。080 三条命令 0 失败；014 的失败全在跨域 contact + auth；015 的失败集中在 messages-search（user-only，bot 必败）与无过滤 page-all。**没有 lark-im 侧常规「报错/参数整形」工单**——与 token 减体积方向一致，本轮抓手是减体积不是补内容。
+
+## 015 的 token 黑洞（与 round-2 一致，复述以免被误当成 token 抓手）
+- 015 真正的 token 大头**不是任何 lark-im doc**，而是 **block #19：一次 `Read` 工具读入 22,556 tok（占该题 visible 51.5%）**。成因链：#12/#17 `+messages-search`/`--page-all` exit2 → #18 退到 `+chat-messages-list`（无过滤）→ 输出 43.5KB 被持久化 → agent `Read` 整文件 → 22.5k tok 灌进上下文 → 再靠本地 grep(#27–33) 抠出「飞豆」两条。
+- **从文档角度**：`chat-messages-list.md` 本题 reach=0（状态①：调用前从没读），它本写了 `--start/--end`、`--page-size` 等可避免全量拉取的约束。**但补它=增常驻/触达内容，与本轮降 token 目标方向相反**（见方向冲突）；且 22.5k 是**单次工具输出**（强依赖该群消息量，单题不可外推），不是稳定常驻热点。**结论：观察项，交评测裁决，不作为本轮 token 抓手。**
+
+## 可发现性时序（约束 5 三态；判「前置能不能救」的决定性证据）
+> 对每条相关 reference / `--help`，按相对首次失败调用的读取时序统计。`--help` 扫 Bash（本轮 3 题均未跑任何 `--help`）。
+
+| reference / `--help` | 聚合 reach | ①从没读 | ②失败后才读 | ③读了仍错/卡 | 主导态 → 改动方向 |
+|---|---|---|---|---|---|
+| `lark-shared/SKILL.md` | 1.0 | 0 | 0 | — | 三题调用前都读了；014 仍卡（环境，非内容）；不可改 |
+| `chat-create.md` | 0.667 | 0 | 0 | — | 080 调用前读→建群成功；014 调用前读→授权阻断（非 reference 错）。**非触达问题，纯减体积** |
+| `messages-send.md` | 0.333 | 0 | 0 | — | 080 调用前读→发卡片成功。**非触达问题**（已收割） |
+| `chat-messages-list.md` | 0.0 | 1 (015) | 0 | — | ① 015 调用前从没读→`--page-all` 全量拉取→token 黑洞。触达缺口，但补它=增 token，与目标冲突 |
+| `messages-search.md` | 0.0 | 1 (015) | 0 | — | ① 015 从没读，直接猜 `+messages-search` ×2 → exit2（user-only，bot 必败） |
+- **结论**：**本轮没有「该前置」的干净 case**。RC-1（chat-create.md 减体积）是「调用前已读、内容够用 → 去冗余」的纯 token 减法，不涉及触达。015 的两处 ① 触达缺口确实存在，但修它们=增内容、与降 token 目标相反，且 015 已 PASS（bot + 本地 grep 兜底）——属观察项，**不要被诱导去推前置**。
+
+## 方向冲突记录（硬性约束 7）
+- **减体积（RC-1 chat-create.md，与 objective.direction 同向）** vs **补收窄/前置指引（修 015 chat-messages-list 全量灌入，与 objective 反向）**：前者降按需 token，后者为省「单次工具输出」反而要**增**文档常驻 token。两者方向相反，**不可合并**。本轮目标是降 token，取减体积一侧；015 全量灌入作为观察项记录、不作为要补的内容根因。
+
+## 差距台账复盘
+- 无（`discard-ledger.json` 为空，无已跑未采纳候选）。
+
+## 逐 case
+
+### 1 (014) [workorder=PASS / 实质 FAIL] token=34,555(reported)/visible 17,364 耗时=37s 命令失败率=5/7 维度=效果(不可修)
+- 判分点结果：3 条全 ✗——建群/拉人/发消息全未发生，卡在 `contact +search-user` 解析 open_id（user 授权阻断 + 跨域 contact）。verdict=PASS 系聚合口径，按判分点证据当 FAIL 处理。
+- 命令失败：5/7。`contact +search-user` bot exit2 ×2、`--as user` token_missing ×2；`auth qrcode` 绝对路径 unsafe ×1（改相对路径成功）。**全部非 lark-im 命令**。
+- 可发现性时序：#4 读 SKILL.md 正文(3,456) + #6 读 lark-shared(3,751，跨 skill) + #7 读 chat-create.md(3,062，调用前已读)；失败在更上游的跨域 contact + 授权。**非 lark-im 触达问题**。
+- token 归因：SKILL.md 正文 3,456（常驻静态，19.9%）+ lark-shared 3,751（**跨 skill，不归因 lark-im**）+ chat-create.md 3,062（按需，17.6%，**本题读了没用上**——授权阻断没走到建群）+ 系统 Skill 列表注入 4,612（固定开销，不归因）。lark-cli 命令累计含多次短失败回显，单条都短、非热点。
+- 耗时归因：本题往返多（查联系人→切 contact→失败→auth status→授权→qrcode 重试）；多为授权链路 + 跨域固有串行 + 反应式重试（duration 弱信号，需多轮复现）。
+- 文档根因：效果=沙箱 user 授权 + 跨域 contact（环境，**无 lark-im 文档根因，本轮不改**）；token=chat-create.md 按需冗余（RC-1，但本题读了没用上，收益只在 080 这种走通题里兑现）+ SKILL.md 常驻（RC-3，风险高、次级）。
+
+### 2 (015) [PASS·真] token=54,568(reported)/visible 43,760 耗时=2m5s 命令失败率=3/11 维度=token（但大头非 lark-im doc）
+- 判分点结果：3/3 ✓——定位群、转发「飞豆」消息、@傅六铭知会全部成功（两次 `messages-send` 均 `ok:true`）。**全程 bot 身份，无授权阻断**。
+- 命令失败：3/11。`+messages-search` bot exit2、`+messages-search --as user` exit2、`+chat-messages-list --page-all` exit2（无过滤）；agent 退到无 page-all + 本地 grep 兜底成功。（#14 `--page-all | grep` 返回空属「成功但无命中」，非硬失败，未计入。）
+- 可发现性时序：① `messages-search.md` / `chat-messages-list.md` 调用前从没读（reach=0），直接猜命令。**本题未读任何 lark-im reference**，故 lark-im reference 的体积与本题 token 无关。
+- token 归因：**本题 token 大头不是 lark-im doc**，是 block #19 一次 `Read` 持久化文件 = **22,556 tok（51.5%，归「其他工具调用/返回」）**，成因=`--page-all` 无过滤全量拉取→43.5KB→Read 灌入（**单次输出**可信度，强依赖该群消息量）。SKILL.md 正文 3,448（常驻）。lark-shared 3,749（跨 skill，不归因）。**RC-1 改 chat-create.md 对本题 token 无影响**（本题没读它）。
+- 耗时归因：本题最长(2m5s)，主因 messages-search 连环失败→改 page-all→大输出→多次本地 grep 抠数据的多轮往返（duration 弱信号；model calls 16/raw 32，明显高于 080，作旁证）。
+- 文档根因：token 黑洞的放大器=`chat-messages-list.md` 没被读到（状态①）+ SKILL.md 表未提示大群应 server-side 收窄——但**补这条与降 token 目标相反**（方向张力），列为观察项；本题已 PASS。本轮 token 抓手（RC-1）不落在本题。
+
+### 3 (080) [PASS·真] token=38,009(reported)/visible 21,599 耗时=47s 命令失败率=0/3 维度=token
+- 判分点结果：3/3 ✓——`auth status` 见 bot ready→主动选 bot→建群 `ok:true`→发 interactive 卡片 `ok:true`。**任务完整完成，零命令失败**。
+- 命令失败：0/3。三条 lark-cli（auth status / chat-create / messages-send）全成功。
+- 可发现性时序：#4 读 SKILL.md 正文(3,455) + #6 读 lark-shared(3,751，跨 skill) + #9 读 chat-create.md(3,060) + #10 读 messages-send.md(5,365，旧版) ，全部状态③（调用前已读且用上）。**无触达问题。** 实际只用了 `+chat-create --name … --format json` 的最简形态——没用两步流/owner/members/topic/error-recovery。
+- token 归因：**本题是纯 token 抓手题**——读取 Skill 占 56.4%：messages-send.md 5,365（trace 旧版，**当前已被 round-2 压到 2,006，本轮不再可压**）+ SKILL.md 正文 3,455（常驻，RC-3）+ chat-create.md 3,060（按需，**RC-1，当前 2,336，本轮唯一干净抓手**）。系统 Skill 列表注入 4,612（固定开销，不归因）。lark-shared 3,751（跨 skill，不归因）。
+- 耗时归因：47s，全部为正常建群+发卡片串行，无重试、无写后回查（无离群）。
+- 文档根因：无效果根因（已绿）；token=RC-1（chat-create.md 内部冗余，本题是其收益唯一稳态兑现题）+ RC-3（SKILL.md 常驻，风险高、次级）。**本题 token 杠杆最清晰且 effect 风险可控**（命令全成功，压 chat-create.md 的示例/场景/报错镜像不碰 080 实际用到的最简建群链路）。
+
+## 给 candidate-writer 的收口（不含具体改法）
+- **唯一在 T1 内还没被收割的干净 token 抓手是 RC-1（`chat-create.md` 内部冗余）**：Commands 12 例过度罗列 + Usage Scenarios 3 场景重复 Commands + Common Errors 9 行部分镜像 validation 字符串——**与 round-2 已采纳的 messages-send.md 压缩同型**，粗估可省 ~600–800 raw tok（约 30%）。reach=0.667（014+080），降幅在子集计入。
+- **载重红线**：AI Usage Guidance 的 232043 两步流 + `succeed_type=1` + `--chat-mode topic` 区分 + `--owner` 默认，是 080/014 走通 bot 建群的语义依据，**压缩中不可误删**——压的是示例/场景/报错镜像体积，不是规则。
+- **RC-2 已收割**：messages-send.md round-2 已 5,365→2,006，trace 里的 5,365 是历史值，**不要重复提**。
+- **RC-3（SKILL.md 常驻）是次级且风险更高**：round-1 已压过一刀，剩余多为全域 identity/路由/约束，删错碰坏 015/080 已走通的 bot 身份与命令路由——不作首选。
+- **不要推前置**：本轮没有「该前置」的干净 case。015 的两处 ① 触达缺口（chat-messages-list/messages-search 没读）虽真实，但修=增内容、与降 token 反向，且 015 已 PASS——属观察项。
+- **effect 不可在本轮 T1 内合法抬升**：014 是环境（沙箱不能扫码）+ 跨域 contact，无 lark-im 文档根因；015/080 已真 PASS。effect deltas 视作 auth-noise，不追。
+- **干净 token 抓手接近见底（诚实判断）**：reach>0 集合三个文件中，messages-send.md（round-2）与 SKILL.md（round-1）已各压一刀，**chat-create.md 是最后一个未动过的干净文件**。压完它之后，T1 内 reach>0 的纯冗余（罗列/重复/--help 镜像）基本耗尽；再往下只剩 (a) 高 effect 风险的 SKILL.md 载重内容，或 (b) reach=0 的 22 个盲区 reference（压了也不在判决集、无法被采纳）。**本轮 RC-1 很可能是这条优化路径上最后一个低风险、可被采纳的 token 抓手。**
+- **缺失信息（doc_fix_hint 语气，非药方）**：chat-create.md 把同一组 flag 在 Commands(12 例) 与 Usage Scenarios(3 场景) 重复演示、Common Errors 多行复述 `--help`/报错本身就会吐的 validation 字符串——这类「枚举/重复/镜像、低增量」内容是其 token 的主要去处，且是减法（删冗余）而非加法。
+
+## 数据缺口
+1. **trace 版本错位（最关键）**：本轮 trace=round-1 旧版 child-runs，messages-send.md 在 trace 里仍是 5,365（round-2 已压到 2,006）。所有「当前文件体积」结论我已用 ai-tokenizer 实测当前工作树校正（SKILL.md 3,525 / chat-create.md 2,336 / messages-send.md 2,006），但**单题行为与 reach 仍来自旧 trace**——若 round-2 改动改变了 080/014 的读取行为，需以实际 round-3 eval-run 复核。
+2. **RC-1 子集小**：chat-create.md reach=0.667 但实际吃到压缩收益的稳态题只有 080（014 读了没用上、授权阻断），证据基数=1，降幅需评测在子集确认。
+3. **015 的 22.5k 黑洞是单次工具输出**，强依赖该群消息量，非稳定常驻热点，单题不可外推；且与降 token 目标方向冲突，不作抓手。
+4. **duration 三题波动大**（37s/2m5s/47s），015 长尾主因 messages-search 连环失败+大输出多轮抠数据；单轮不足定论，需多轮复现。model calls(8/16/6) 比 wall-clock 稳，可作旁证。
+5. **工具调用口径不一致**：trend.json 的 R1 tool_calls=26.3、R2=10，与 session-analyze 的 model calls(8/16/6) 口径不同（趋势表疑似含 raw 计数）；旁证以 timeline 实际往返为准。趋势看：R1→R2 命令失败率 0.60→0.35、tool_calls 26→10 明显下降，但那主要是 effect 从「三题全卡授权」变成「2 真 PASS + 1 卡」带来的，**不是 token 改动的功劳**；token 均值 R1 31,997→R2 42,377 上升，主因是 R2 只评 080（单题大）口径差异 + 015 黑洞，非文档常驻变重——趋势对 token 轴判读价值有限，以单题 session-analyze 为准。
+6. **effect 维度全部归因为「无文档根因/不可修」**：014 跨域+环境，015/080 已绿。本轮 effect 无 T1 可发力点，deltas 视作 auth-noise。

harness-opt/rounds/round-003/discard-ledger.json

@@ -0,0 +1 @@
+[]

harness-opt/rounds/round-003/failure-memory.json

@@ -0,0 +1 @@
+[]

harness-opt/rounds/round-003/module-reach.json

@@ -0,0 +1,220 @@
+{
+  "skills/lark-im/SKILL.md": {
+    "reach": 1.0,
+    "read_cases": [
+      "1",
+      "2",
+      "3"
+    ],
+    "actual_cases": [
+      "1",
+      "2",
+      "3"
+    ],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": true
+  },
+  "skills/lark-im/references/lark-im-chat-create.md": {
+    "reach": 0.667,
+    "read_cases": [
+      "1",
+      "3"
+    ],
+    "actual_cases": [
+      "1",
+      "3"
+    ],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-chat-identity.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-chat-list.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-chat-messages-list.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-chat-search.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-chat-update.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-feed-group-list-item.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-feed-group-list.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-feed-group-query-item.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-feed-groups.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-feed-shortcut-create.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-feed-shortcut-list.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-feed-shortcut-remove.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-flag-cancel.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-flag-create.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-flag-list.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-message-enrichment.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-messages-mget.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-messages-reply.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-messages-resources-download.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-messages-search.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-messages-send.md": {
+    "reach": 0.333,
+    "read_cases": [
+      "3"
+    ],
+    "actual_cases": [
+      "3"
+    ],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-reactions.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  },
+  "skills/lark-im/references/lark-im-threads-messages-list.md": {
+    "reach": 0.0,
+    "read_cases": [],
+    "actual_cases": [],
+    "expected_cases": [],
+    "discoverability_miss": [],
+    "is_domain_skill": false
+  }
+}

harness-opt/rounds/round-003/review.json

@@ -0,0 +1,18 @@
+{
+  "generated_by": "lark-cli-harness:opt-reviewer",
+  "verdict": "PASS",
+  "module": "skills/lark-im/references/lark-im-chat-create.md",
+  "tier": "T1",
+  "round_index": 3,
+  "base_commit": "572eb8da41f608bd93b25916cac02cb772825b97",
+  "code_tip": "cbd6e56ac07285fd973c53ff7382da0112b6cf5d",
+  "reason": "纯瘦身改动，对抗式逐项核验未发现可证伪点：14 条承重红线（232043 两步流/succeed_type=1/chat-mode topic 与 thread 区分/--owner 默认/set-bot-manager/chat.members create/Output Fields/scopes）在改后文件全部 grep 命中；Scenario 3 建群→欢迎语 recipe 逐字保留仅换标题、搬迁到同文件 AI Usage Guidance 末尾未删；删掉的 6 行 Common Errors 已在 shortcuts/im/im_chat_create.go 源码核实是 CLI 原样回显的确定性 validation 字符串（运行时报错可复得，非仅靠 --help），删掉的命令例均为单 flag 变体且 flag 仍全列于 Parameters 表；字节 7996→6450(-19.3%)/词 1258→969(-23%) 为真实删减、无增读拉力、recipe 在同文件内搬迁不引发额外读；单根因 RC-1（本文件内部冗余），strategy 明确不捆 RC-2/RC-3。",
+  "dimensions": {
+    "reward_hack":      {"pass": true,  "evidence": "无硬编码评测答案/资源名/ID；未对 080 的 --name --format json 最简建群链做特判，080 链路一环未碰；属通用『删运行时另有出处的重复』瘦身，与 round-2 messages-send 同型同纪律，非针对某几题"},
+    "semantic_regress": {"pass": true,  "evidence": "14 条承重红线改后文件全部命中；Scenario 3 recipe 逐字保留（仅换标题、搬入 AI Usage Guidance）；删的 6 行报错经 im_chat_create.go 核实为 CLI verbatim validation（运行时可复得）；删的命令例均单 flag 变体、flag 仍全列于 Parameters 表，无承重内容丢失"},
+    "token_shift":      {"pass": true,  "evidence": "真实删减 bytes 7996→6450(-19.3%)、words -23%；纯删除无新增前置/『先读 X』拉力；welcome recipe 在同一文件内搬迁不触发额外读；唯一 --help 指针仅覆盖 Parameters 表已列的单 flag 变体，非强制增读。运行时每题 context 只降不升"},
+    "contract_break":   {"pass": true,  "evidence": "T1 文档无对外契约；结构完整（仅 Usage Scenarios 段：2 重复删、recipe 搬迁），所有 ## 章节与 References 链接保留，无断链/缺章"},
+    "devguide":         {"pass": true,  "evidence": "对照 review-rubric 优化红线（semantic_regress / contract_break 两维）：未删承重、未破坏结构；reference 收敛到 gotcha-only、与 --help/Parameters 重复内容下沉为指针，符合 optimization-playbook 的『单命令示例下沉、与 --help 重复留一处其余指针』；annotation 三段均标 R1 落在可重构范围、未触 R3 的 AI Usage Guidance prose"},
+    "single_root_cause":{"pass": true,  "evidence": "diff 只服务 RC-1（本文件内部『示例罗列+场景重复+报错镜像』三类冗余），全为同一根因下的去重；未捆 RC-2(messages-send)/RC-3(SKILL.md)，未夹带语义独立的承重删除，无多根因对冲叙事"}
+  }
+}

harness-opt/rounds/round-003/round.json

@@ -0,0 +1,394 @@
+{
+  "round": 3,
+  "status": "admitted",
+  "parent_id": "557349b40feb359bb791749a37571d59edb7e72e",
+  "parent_worktree": "/Users/bytedance/Projects/cli",
+  "child_worktree": "/Users/bytedance/Projects/cli",
+  "base_commit": "572eb8da41f608bd93b25916cac02cb772825b97",
+  "module": "skills/lark-im/references/lark-im-chat-create.md",
+  "candidate_modules": [
+    "skills/lark-im/SKILL.md",
+    "skills/lark-im/references/lark-im-chat-create.md",
+    "skills/lark-im/references/lark-im-chat-identity.md",
+    "skills/lark-im/references/lark-im-chat-list.md",
+    "skills/lark-im/references/lark-im-chat-messages-list.md",
+    "skills/lark-im/references/lark-im-chat-search.md",
+    "skills/lark-im/references/lark-im-chat-update.md",
+    "skills/lark-im/references/lark-im-feed-group-list-item.md",
+    "skills/lark-im/references/lark-im-feed-group-list.md",
+    "skills/lark-im/references/lark-im-feed-group-query-item.md",
+    "skills/lark-im/references/lark-im-feed-groups.md",
+    "skills/lark-im/references/lark-im-feed-shortcut-create.md",
+    "skills/lark-im/references/lark-im-feed-shortcut-list.md",
+    "skills/lark-im/references/lark-im-feed-shortcut-remove.md",
+    "skills/lark-im/references/lark-im-flag-cancel.md",
+    "skills/lark-im/references/lark-im-flag-create.md",
+    "skills/lark-im/references/lark-im-flag-list.md",
+    "skills/lark-im/references/lark-im-message-enrichment.md",
+    "skills/lark-im/references/lark-im-messages-mget.md",
+    "skills/lark-im/references/lark-im-messages-reply.md",
+    "skills/lark-im/references/lark-im-messages-resources-download.md",
+    "skills/lark-im/references/lark-im-messages-search.md",
+    "skills/lark-im/references/lark-im-messages-send.md",
+    "skills/lark-im/references/lark-im-reactions.md",
+    "skills/lark-im/references/lark-im-threads-messages-list.md"
+  ],
+  "module_reach": {
+    "skills/lark-im/SKILL.md": {
+      "reach": 1.0,
+      "read_cases": [
+        "1",
+        "2",
+        "3"
+      ],
+      "actual_cases": [
+        "1",
+        "2",
+        "3"
+      ],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": true
+    },
+    "skills/lark-im/references/lark-im-chat-create.md": {
+      "reach": 0.667,
+      "read_cases": [
+        "1",
+        "3"
+      ],
+      "actual_cases": [
+        "1",
+        "3"
+      ],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-chat-identity.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-chat-list.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-chat-messages-list.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-chat-search.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-chat-update.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-feed-group-list-item.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-feed-group-list.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-feed-group-query-item.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-feed-groups.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-feed-shortcut-create.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-feed-shortcut-list.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-feed-shortcut-remove.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-flag-cancel.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-flag-create.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-flag-list.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-message-enrichment.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-messages-mget.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-messages-reply.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-messages-resources-download.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-messages-search.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-messages-send.md": {
+      "reach": 0.333,
+      "read_cases": [
+        "3"
+      ],
+      "actual_cases": [
+        "3"
+      ],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-reactions.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    },
+    "skills/lark-im/references/lark-im-threads-messages-list.md": {
+      "reach": 0.0,
+      "read_cases": [],
+      "actual_cases": [],
+      "expected_cases": [],
+      "discoverability_miss": [],
+      "is_domain_skill": false
+    }
+  },
+  "expected_reach": {},
+  "minibatch": [
+    "1",
+    "2",
+    "3"
+  ],
+  "pareto_cases": [
+    "1",
+    "2",
+    "3"
+  ],
+  "artifacts": {
+    "workorder": "workorder.md",
+    "diagnosis": "diagnosis.md",
+    "attribution": "attribution.json",
+    "strategy": "strategy.md",
+    "review": "review.json",
+    "trend": "trend.json"
+  },
+  "code_tip": "cbd6e56ac07285fd973c53ff7382da0112b6cf5d",
+  "signature": "53194d7a111df326cc078b633f43587225bd0132",
+  "tier": "T1",
+  "intent": "dedup Commands<->Usage Scenarios overlap + compress --help-mirroring Common Errors in chat-create.md; keep all red lines (232043 two-step,succeed_type=1,chat-mode topic,--owner)",
+  "target_axis": "token",
+  "changed_files": [
+    "skills/lark-im/references/lark-im-chat-create.md"
+  ],
+  "decision_basis": {
+    "type": "module",
+    "module": "skills/lark-im/references/lark-im-chat-create.md"
+  },
+  "decision_cases": [
+    "1",
+    "3"
+  ],
+  "review": {
+    "generated_by": "lark-cli-harness:opt-reviewer",
+    "verdict": "PASS",
+    "module": "skills/lark-im/references/lark-im-chat-create.md",
+    "tier": "T1",
+    "round_index": 3,
+    "base_commit": "572eb8da41f608bd93b25916cac02cb772825b97",
+    "code_tip": "cbd6e56ac07285fd973c53ff7382da0112b6cf5d",
+    "reason": "纯瘦身改动，对抗式逐项核验未发现可证伪点：14 条承重红线（232043 两步流/succeed_type=1/chat-mode topic 与 thread 区分/--owner 默认/set-bot-manager/chat.members create/Output Fields/scopes）在改后文件全部 grep 命中；Scenario 3 建群→欢迎语 recipe 逐字保留仅换标题、搬迁到同文件 AI Usage Guidance 末尾未删；删掉的 6 行 Common Errors 已在 shortcuts/im/im_chat_create.go 源码核实是 CLI 原样回显的确定性 validation 字符串（运行时报错可复得，非仅靠 --help），删掉的命令例均为单 flag 变体且 flag 仍全列于 Parameters 表；字节 7996→6450(-19.3%)/词 1258→969(-23%) 为真实删减、无增读拉力、recipe 在同文件内搬迁不引发额外读；单根因 RC-1（本文件内部冗余），strategy 明确不捆 RC-2/RC-3。",
+    "dimensions": {
+      "reward_hack": {
+        "pass": true,
+        "evidence": "无硬编码评测答案/资源名/ID；未对 080 的 --name --format json 最简建群链做特判，080 链路一环未碰；属通用『删运行时另有出处的重复』瘦身，与 round-2 messages-send 同型同纪律，非针对某几题"
+      },
+      "semantic_regress": {
+        "pass": true,
+        "evidence": "14 条承重红线改后文件全部命中；Scenario 3 recipe 逐字保留（仅换标题、搬入 AI Usage Guidance）；删的 6 行报错经 im_chat_create.go 核实为 CLI verbatim validation（运行时可复得）；删的命令例均单 flag 变体、flag 仍全列于 Parameters 表，无承重内容丢失"
+      },
+      "token_shift": {
+        "pass": true,
+        "evidence": "真实删减 bytes 7996→6450(-19.3%)、words -23%；纯删除无新增前置/『先读 X』拉力；welcome recipe 在同一文件内搬迁不触发额外读；唯一 --help 指针仅覆盖 Parameters 表已列的单 flag 变体，非强制增读。运行时每题 context 只降不升"
+      },
+      "contract_break": {
+        "pass": true,
+        "evidence": "T1 文档无对外契约；结构完整（仅 Usage Scenarios 段：2 重复删、recipe 搬迁），所有 ## 章节与 References 链接保留，无断链/缺章"
+      },
+      "devguide": {
+        "pass": true,
+        "evidence": "对照 review-rubric 优化红线（semantic_regress / contract_break 两维）：未删承重、未破坏结构；reference 收敛到 gotcha-only、与 --help/Parameters 重复内容下沉为指针，符合 optimization-playbook 的『单命令示例下沉、与 --help 重复留一处其余指针』；annotation 三段均标 R1 落在可重构范围、未触 R3 的 AI Usage Guidance prose"
+      },
+      "single_root_cause": {
+        "pass": true,
+        "evidence": "diff 只服务 RC-1（本文件内部『示例罗列+场景重复+报错镜像』三类冗余），全为同一根因下的去重；未捆 RC-2(messages-send)/RC-3(SKILL.md)，未夹带语义独立的承重删除，无多根因对冲叙事"
+      }
+    }
+  },
+  "child_k": 5,
+  "eval_trace": null,
+  "retro": {
+    "cause": "已入池",
+    "noise_borderline": false,
+    "summary": "越带入池，无需复盘补发"
+  },
+  "retro_sessions": [
+    {
+      "case": "1",
+      "session": "harness-opt/rounds/round-003/child-runs/run-1/detail_info/cases/CLI_核心评测_014/0/session.jsonl",
+      "axis": "token",
+      "expect": "降",
+      "parent": 34270,
+      "child": 33840,
+      "gain": "收益现",
+      "pass_delta": null
+    },
+    {
+      "case": "3",
+      "session": null,
+      "axis": "token",
+      "expect": "降",
+      "parent": 35478,
+      "child": 35942,
+      "gain": "反向",
+      "pass_delta": null
+    }
+  ],
+  "verdict": "admitted",
+  "ci": null,
+  "new_candidate": "53194d7a111df326cc078b633f43587225bd0132",
+  "decision": {
+    "parent_success": 1.0,
+    "child_success": 1.0,
+    "parent_score": 0.8,
+    "child_score": 1.0,
+    "score_saved": 0.19999999999999996,
+    "score_threshold": 0.09532271373123208,
+    "parent_token": 34874.0,
+    "child_token": 34891.0,
+    "saved": -17.0,
+    "threshold": 4532.708313776408,
+    "parent_duration": 45267.5,
+    "child_duration": 43656.0,
+    "dur_saved": 1611.5,
+    "dur_threshold": 4899.200953624988,
+    "dur_margin": 1.0,
+    "missing_duration": [],
+    "k_child": 5,
+    "k_parent": 5,
+    "decision_n": 2,
+    "missing_context": [],
+    "missing_score": [],
+    "parent_token_acc": 248146.5,
+    "child_token_acc": 235911.0,
+    "phi0_score": 0.5666666666666667,
+    "eff_margin": 1.0,
+    "parent_token_full": 34874.0,
+    "child_token_full": 34891.0,
+    "saved_full": -17.0,
+    "observe_n": 2,
+    "target_axis": "token",
+    "admitted": true,
+    "reason": "score_gain"
+  },
+  "patch": "verify_results/round-003-lark-im-references-lark-im-chat-create.patch"
+}

harness-opt/rounds/round-003/strategy.md

@@ -0,0 +1,43 @@
+# Round 3 候选策略（模块=references/lark-im-chat-create.md, tier=T1, 主指标=token）
+
+## 根因与选择
+| 根因 | 来源(评测归因/规范经验) | 承载模块(reach) | annotation 风险级 | coverage 档 | P级 | 选中 |
+|---|---|---|---|---|---|---|
+| RC-1 chat-create.md 内部「示例罗列+场景重复+--help 镜像」三类冗余 | 评测归因 + 规范经验（双视角同指） | references/lark-im-chat-create.md(0.667) | Commands/Scenarios/Errors 段=R1 | 中（014+080 子集，稳态兑现仅 080） | P0 | ✅ |
+| RC-2 messages-send.md 内部冗余 | 评测归因 | references/lark-im-messages-send.md(0.333) | — | — | — | （round-2 已收割，不再是抓手） |
+| RC-3 SKILL.md 常驻正文 | 评测归因 | SKILL.md(1.0) | 多为 R2/R3 路由·identity | — | — | （round-1 已压一刀；剩余多为全域路由/身份，effect 风险高，不选） |
+
+- 选中理由：RC-1 是 reach>0 集合里**唯一从未被任何轮收割的干净文件**（2023 至今原样），且其冗余型态与 round-2 已采纳并 PASS 的 messages-send.md 完全同型（罗列+重复+--help 镜像）。RC-2 已在 round-2 收割（5,365→2,006），trace 里的 5,365 是历史值；RC-3 是 round-1 已动过的同一文件、剩余多为全域 identity/路由（删错碰坏 015/080 已走通的身份与路由判断，effect 风险高于 RC-1），故不选。
+- 选模块理由：chat-create.md reach=0.667（014+080 调用前都读到，状态③，非触达问题——纯减体积场景）；它正是承载 RC-1 的文件。未选 reach=0 的 22 个盲区 reference（改了也不在判决集、无法被采纳，触 reach 锁）。
+- 规范经验源补注：双视角同指一处。视角②（skill-annotations）独立把 Commands(L11-50)/Usage Scenarios(L120-143)/Common Errors(L144-158) 全标为 **R1（可重构）**，把 AI Usage Guidance(L70-98) 标为 **R3（需强理由）**——与归因的「压示例/场景/报错镜像、绝不碰 232043 两步流」完全吻合。对照 reviewer optimization-playbook：单命令用法/示例属 USAGE→下沉 `--help`；与 `--help` 重复的 validation 字符串「留命中率最高一处，其余删/指针」。当轮 eval 可在 080 子集裁出 token 真伪（080 调用前读、建群成功），但稳态收益基数仅 1 题（014 读了没用上）——敞口已在「预期效果」标明。
+
+## 改了什么（逐处）
+- **Commands(L12-50)** — 12 个 bash 示例（多条仅差一个 flag）压成 5 个差异化示例 + 一行 `--help` 指针。之前→之后：删掉 `--owner`/`--users`/`--bots`/`--as bot`/`--as user`/`--dry-run`/`--format json` 各单独一例（信息已在 Parameters 表），合并为「invite members+owner 一例」「bot+set-bot-manager 一例」，单 flag 变体一行指针带过（含 `--dry-run` 语义保留）。
+- **Usage Scenarios(L120-143)** — 整段 3 场景删除/搬迁。Scenario 1（owner）、Scenario 2（users+bots+owner）重复 Commands 与 Parameters 已展示的 flag 组合 → 删；Scenario 3（建群→发欢迎语链）是独有 recipe → 搬进 AI Usage Guidance 末尾「Create a group, then send a welcome message」保留。
+- **Common Errors(L144-158)** — 9 行压成 2 行。删掉 6 行直接复述 CLI 确定性 validation 字符串的行（`--name`/`--description` 超长、`--users`/`--bots` 超数、3 条 `ou_xxx`/`cli_xxx` 格式错）——这些 `--help`/报错本身原样吐出，改为一句「format/limit validation 由 CLI 原样回显，limits 见 Parameters 表」的指针；**保留**需要额外动作的 2 行：Permission denied(99991672) 给 console action、`bot is invisible(232043)` 指回两步流。
+
+## 为什么这么改（机制）
+- **省 context 的因果**：chat-create.md 是 lazy reference，读到即整文件进窗口（080/014 reach=0.667）。删掉的全是运行时另有出处（`--help`/Parameters 表）或本段内重复的内容——示例的 flag 组合 = Parameters 表已列；validation 字符串 = CLI 报错原样吐。删后 Agent 仍能：经 SKILL.md 选对 `+chat-create`、经 Parameters 表/`--help` 补全 flag 用法、遇 232043 走两步流。即 optimization-playbook §13 核心判据「删掉后 Agent 是否仍能选对命令并补到用法」——成立。
+- **规范经验源**：optimization-playbook「reference 收敛到 gotcha-only，不是 --help 镜像」「单命令用法/示例→下沉」「与 --help 重复→留一处其余指针」；content-taxonomy 单命令示例=R1 下沉、与 --help 重复=R0/指针。annotation 三段独立标 R1，本改动落在 R1 重构范围内，未触 R3 段。
+
+## 预期效果
+- 成功率（effect，硬门槛）：**不退化**。所有 effect 红线逐条保留并已 grep 校验（见下「刻意没做什么」）。080 实际只用 `--name --format json` 最简建群链——本改动未碰该链路任何一环；014 卡在跨域 contact+授权（非本 reference）。
+- context（分两层）：
+  - **(1) 静态字数差**：bytes 7996→6450（-19.3%）、chars -19.5%、words -23.0%、tiktoken(cl100k 代理) 2125→1714(-19.3%)。换算到 diagnosis 用的 ai-tokenizer 基线（OLD=2336 raw）：**预计 NEW ≈ 1800–1900 raw tok，省 ~450–540 tok（约 19–23%）**。
+  - **(2) 运行时 context 方向**：对**读到 chat-create.md 的题（080，及理论上 014）下降** ~450–540 tok；对没读它的题（015）**无影响**（015 大头是单次 `Read` 22.5k 工具输出，与本 reference 无关）。本改动是纯删减、无新增前置/增读拉力，不会抬升运行时 token。
+  - **与 direction 一致**（objective=降 token），无张力。
+- **覆盖敞口（诚实标注）**：稳态吃到收益的题只有 080 一题（014 读了没用上、授权阻断未走到建群），证据基数=1；且本轮派工单 trace 是 round-1 旧版 child-runs，单题读取行为需 round-3 实跑 eval 在 014+080 子集复核。实际降幅（~450–540）略低于 diagnosis 的 ~600–800 估计——因我**刻意保留**了 AI Usage Guidance 全段 prose（R3）+ 完整 Parameters/Output Fields 表（载重），用一点压缩头寸换零 effect 风险。
+
+## 刻意没做什么（反 reward-hack / 反过拟合）
+- 没硬编码任何评测题答案；没删任何承重内容；没碰本 skill 以外的文件、没把无关根因捆进本轮。
+- **逐条保留的载重红线（已 grep 校验存在）**：
+  - 232043 两步流全段（contact search → `--users 当前用户` 建群 → `chat.members create --as user` 加其他人 → 查 `invalid_id_list`）；
+  - `succeed_type=1` 语义解释；
+  - `--chat-mode topic` vs 「普通群 + `group_message_type=thread`」的区分注解；
+  - `--owner` 默认行为（bot 身份默认 bot / user 身份默认授权用户）；
+  - 全部 flag（含 `--set-bot-manager`、`--dry-run`、`--type public`、`--users/--bots` 上限与格式）、identity/scope 指引、互斥与护栏规则、Output Fields 全表。
+- 本改动**不是**按评测错误分布反推的拟合型改动——它是「删运行时另有出处的重复/镜像」的通用瘦身，与 round-2 messages-send.md 同型同纪律；非针对某几题的特判。
+- 未做 RC-3（SKILL.md 进一步压缩）：剩余多为全域 identity/路由，删错有 effect 风险，超出本轮低风险抓手范围。未做 015 的前置补充：那是增内容、与降 token 反向（方向冲突，diagnosis 已记录）。
+
+## 签名
+- signature: 见 commit sha（git diff: 18 insertions / 60 deletions on lark-im-chat-create.md）  tier: T1

harness-opt/rounds/round-003/trend.json

@@ -0,0 +1,20 @@
+[
+  {
+    "round": 1,
+    "n": 3,
+    "pass_n": 0,
+    "cmd_fail_rate": 0.6,
+    "tool_calls": 26.333333333333332,
+    "duration_ms": 50189.0,
+    "token": 31997.0
+  },
+  {
+    "round": 2,
+    "n": 3,
+    "pass_n": 3,
+    "cmd_fail_rate": 0.3466666666666667,
+    "tool_calls": 10.0,
+    "duration_ms": 69666.66666666667,
+    "token": 42377.333333333336
+  }
+]

harness-opt/rounds/round-003/workorder.md

@@ -0,0 +1,44 @@
+# Round 3 归因派工单（parent=557349b40feb359bb791749a37571d59edb7e72e；模块未定，由 candidate-writer 据诊断点名）
+
+> **只读输入**——opt-attributor 读本文件，把诊断**另写** `diagnosis.md`（给 candidate-writer）+ 逐题结构化 `attribution.json`（给 dashboard）。**不要覆盖本文件**，留作派工单↔诊断的前后对比。
+> 判分点只当「什么算挂」的锚，禁止照抄 grader 药方（已从派工单剔除）。
+
+## 模块运行时可达性（选模块第一步的证据；要选须在 strategy.md 说明理由）
+> reach=**实测**触达率（域主 SKILL.md 经 Skill 工具加载、reference 经 Read，都从 trace 实测，没有恒在的面）；判决集=实测∪预期触达。**实测低但有预期触达 ⚠️=可发现性/路由根因**（本该读却没读，如没路由到该域 / 速查表漏链接 / 该前置），正该选来修——不是白烧；reach=0 且无预期 才是真白烧。 **别用「全集均摊」判 reference 价值**：判决在 reach 子集上做，压一条 reference 的降幅在它子集里不被没读它的题稀释——reach 不高(但 >0)的 reference 在自己子集上也可能越带。
+- `skills/lark-im/SKILL.md` → reach=1.0  [域主 skill·经 Skill 工具加载]；判决集(实测∪预期): ['1', '2', '3']
+- `skills/lark-im/references/lark-im-chat-create.md` → reach=0.667；判决集(实测∪预期): ['1', '3']
+- `skills/lark-im/references/lark-im-messages-send.md` → reach=0.333；判决集(实测∪预期): ['3']
+- （另 22 个 reference reach=0 且无预期触达，本轮无关，略）
+
+## 逐轮诊断信号趋势（纯诊断，不进判决）
+
+| 轮 | 题数 | PASS | 命令失败率 | 工具调用 | 耗时(ms) | token |
+|---|---|---|---|---|---|---|
+| R1 | 3 | 0 | 0.60 | 26 | 50189 | 31997 |
+| R2 | 3 | 3 | 0.35 | 10 | 69667 | 42377 |
+
+> 跨题均值，按轮排。**命令失败率、工具调用数是横切诊断信号，不是准入轴**（准入只走 效果/token/耗时）——用来判「上一轮那刀有没有把失败/轮次压下去」。工具调用数比 wall-clock 稳，可给噪声大的耗时轴当旁证。
+
+### 1  [PASS]  ctx=34270  (acc=274608)  43995ms  tools=31
+- session.jsonl: harness-opt/rounds/round-001/child-runs/run-1/detail_info/cases/CLI_核心评测_014/0/session.jsonl [native]
+- 判分点（grader 的「什么算挂」oracle，非药方）:
+    ✗ 使用当前用户身份创建名为「IM合作群」的群聊
+        证据: transcript 在展示授权二维码后结束，无任何 `lark-cli im +chat-create` 调用。执行停在 '授权完成后请告诉我，我会继续帮你创建群聊并发送消息'，群聊未创建。
+    ✗ 将傅一铭和傅二铭加入该群
+        证据: transcript 显示尝试搜索用户时遇到 `need_user_authorization` 错误，授权流程启动后中断。未获取到任何用户的 open_id，无后续添加操作。
+    ✗ 在该群发送文本消息「大家体验有问题随时沟通」，并返回可验证的 chat_id / message_id
+        证据: 群聊未创建，无 chat_id 可返回。transcript 无任何 `lark-cli im messages-send` 调用。
+
+### 2  [PASS]  ctx=47116  (acc=612048)  114310ms  tools=49
+- session.jsonl: harness-opt/rounds/round-001/child-runs/run-1/detail_info/cases/CLI_核心评测_015/0/session.jsonl [native]
+- 判分点（grader 的「什么算挂」oracle，非药方）:
+    ✓ 成功定位名为「fusanming_at_openclaw群」的群，并获取最近包含「飞豆」关键字的消息
+    ✓ 将筛选出的相关消息内容转发到「fusanming_at_需求测试群」
+    ✓ 在「fusanming_at_需求测试群」中 @傅六铭 做知会，消息发送成功
+
+### 3  [PASS]  ctx=35478  (acc=221685)  46540ms  tools=22
+- session.jsonl: harness-opt/rounds/round-001/child-runs/run-1/detail_info/cases/CLI_核心评测_080/0/session.jsonl [native]
+- 判分点（grader 的「什么算挂」oracle，非药方）:
+    ✓ 使用用户身份创建一个名为「今晚吃什么」的群，预期返回 chat_id
+    ✓ 创建一张飞书卡片，卡片内容包含「今天晚上吃什么」
+    ✓ 将该卡片发送到新建群中，预期返回 message_id

harness-opt/trend.json

@@ -0,0 +1,20 @@
+[
+  {
+    "round": 1,
+    "n": 3,
+    "pass_n": 0,
+    "cmd_fail_rate": 0.6,
+    "tool_calls": 26.333333333333332,
+    "duration_ms": 50189.0,
+    "token": 31997.0
+  },
+  {
+    "round": 2,
+    "n": 3,
+    "pass_n": 3,
+    "cmd_fail_rate": 0.3466666666666667,
+    "tool_calls": 10.0,
+    "duration_ms": 69666.66666666667,
+    "token": 42377.333333333336
+  }
+]

harness-opt/verify_results/round-001-lark-im-SKILL.patch

@@ -0,0 +1,152 @@
+From 237a77feb341e15656386d6952a875dc459fec8c Mon Sep 17 00:00:00 2001
+From: "zhangheng.023" <zhangheng.023@bytedance.com>
+Date: Tue, 23 Jun 2026 18:27:25 +0800
+Subject: [PATCH] =?UTF-8?q?opt(round-001):=20SKILL.md=20=E2=80=94=20fold?=
+ =?UTF-8?q?=20USAGE=20method-index=20+=20scope=20table=20into=20a=20schema?=
+ =?UTF-8?q?=20pointer=20(-40%=20resident=20tokens)?=
+MIME-Version: 1.0
+Content-Type: text/plain; charset=UTF-8
+Content-Transfer-Encoding: 8bit
+
+---
+ skills/lark-im/SKILL.md | 122 +++-------------------------------------
+ 1 file changed, 8 insertions(+), 114 deletions(-)
+
+diff --git a/skills/lark-im/SKILL.md b/skills/lark-im/SKILL.md
+index bc39aae1..ac1c6900 100644
+--- a/skills/lark-im/SKILL.md
++++ b/skills/lark-im/SKILL.md
+@@ -110,122 +110,16 @@ Shortcut 是对常用操作的高级封装（`lark-cli im +<verb> [flags]`）。
+ | [`+feed-group-list`](references/lark-im-feed-group-list.md) | List the caller's feed groups (tags); user-only; supports `--page-all` auto-pagination |
+ | [`+feed-group-list-item`](references/lark-im-feed-group-list-item.md) | List feed cards in a feed group (tag); user-only; enriches each item with chat_name resolved from feed_id; supports --page-all auto-pagination |
+ | [`+feed-group-query-item`](references/lark-im-feed-group-query-item.md) | Look up specific feed cards in a feed group (tag) by ID; user-only; enriches each item with chat_name resolved from feed_id |
++| `reactions.*` (add / delete / list / batch_query) | Add, remove, or read emoji reactions on a message; user/bot; caller must be in the conversation, and can only delete its own reactions. Read first: [`lark-im-reactions.md`](references/lark-im-reactions.md) |
++| `feed.groups.*` (create / update / delete / batch_query / batch_add_item / batch_remove_item) | Manage feed groups (tags) and their member cards; user-only. Read first: [`lark-im-feed-groups.md`](references/lark-im-feed-groups.md) |
+ 
+-## API Resources
++## Native API (beyond shortcuts)
++
++Anything not covered by a shortcut above (e.g. `chats.*`, `chat.members.*`, `chat.managers.*`, `chat.moderation.*`, `chat.user_setting.*`, `messages.delete|forward|merge_forward|read_users|urgent_*`, `threads.forward`, `images.create`, `pins.*`) is callable as a raw API:
+ 
+ ```bash
+-lark-cli schema im.<resource>.<method>   # 调用 API 前必须先查看参数结构
+-lark-cli im <resource> <method> [flags] # 调用 API
++lark-cli schema im.<resource>.<method>    # MUST run first — gives params, identity (user/bot/tenant), and required scope
++lark-cli im <resource> <method> [flags]   # then call
+ ```
+ 
+-> **重要**：使用原生 API 时，必须先运行 `schema` 查看 `--data` / `--params` 参数结构，不要猜测字段格式。
+-
+-### chats
+-
+-  - `create` — 创建群。Identity: `bot` only (`tenant_access_token`).
+-  - `get` — 获取群信息。Identity: supports `user` and `bot`; the caller must be in the target chat to get full details, and must belong to the same tenant for internal chats.
+-  - `link` — 获取群分享链接。Identity: supports `user` and `bot`; the caller must be in the target chat, must be an owner or admin when chat sharing is restricted to owners/admins, and must belong to the same tenant for internal chats.
+-  - `update` — 更新群信息。Identity: supports `user` and `bot`.
+-
+-### chat.members
+-
+-  - `bots` — 获取群内机器人列表。Identity: supports `user` and `bot`; the caller must be in the target chat and must belong to the same tenant for internal chats.
+-  - `create` — 将用户或机器人拉入群聊。Identity: supports `user` and `bot`; the caller must be in the target chat; for `bot` calls, added users must be within the app's availability; for internal chats the operator must belong to the same tenant; if only owners/admins can add members, the caller must be an owner/admin, or a chat-creator bot with `im:chat:operate_as_owner`.
+-  - `delete` — 将用户或机器人移出群聊。Identity: supports `user` and `bot`; only group owner, admin, or creator bot can remove others; max 50 users or 5 bots per request.
+-  - `get` — 获取群成员列表。Identity: supports `user` and `bot`; the caller must be in the target chat and must belong to the same tenant for internal chats.
+-
+-### chat.user_setting
+-
+-  - `batch_query` — 批量查询当前用户在群内的个人偏好设置 (e.g. `is_muted` mutes normal messages, `is_mute_at_all` mutes @all messages); up to 10 chats per request. Identity: `user` only (`user_access_token`); the caller must be in each target chat.
+-  - `batch_update` — 批量更新当前用户在群内的个人偏好设置 (e.g. `is_muted` mutes normal messages, `is_mute_at_all` mutes @all messages); up to 10 chats per request. Identity: `user` only (`user_access_token`); the caller must be in each target chat.
+-
+-### chat.managers
+-
+-  - `add_managers` — 指定群管理员。Identity: supports `user` and `bot`; only the group owner can add managers; max 10 managers per chat (20 for super-large chats), and at most 5 bots per request.
+-  - `delete_managers` — 删除群管理员。Identity: supports `user` and `bot`; only the group owner can remove managers; max 50 users or 5 bots per request.
+-
+-### chat.moderation
+-
+-  - `get` — 获取群成员发言权限。Identity: supports `user` and `bot`; the caller must be in the target chat and belong to the same tenant.
+-  - `update` — 更新群发言权限。Identity: supports `user` and `bot`; only the group owner (or creator bot with `im:chat:operate_as_owner`) can update; the caller must be in the chat.
+-
+-### messages
+-
+-  - `delete` — 撤回消息。Identity: supports `user` and `bot`; for `bot` calls, the bot must be in the chat to revoke group messages; to revoke another user's group message, the bot must be the owner, an admin, or the creator; for user P2P recalls, the target user must be within the bot's availability.
+-  - `forward` — 转发消息。Identity: supports `user` and `bot`.
+-  - `merge_forward` — 合并转发消息。Identity: `bot` only (`tenant_access_token`).
+-  - `read_users` — 查询消息已读信息。Identity: `bot` only (`tenant_access_token`); the bot must be in the chat, and can only query read status for messages it sent within the last 7 days.
+-  - `urgent_app` — 发送应用内加急。Identity: `bot` only (`tenant_access_token`); the bot must be the message sender and must be in the conversation that contains the message.
+-  - `urgent_phone` — 发送电话加急。Identity: `bot` only (`tenant_access_token`); the bot must be the message sender and must be in the conversation that contains the message.
+-  - `urgent_sms` — 发送短信加急。Identity: `bot` only (`tenant_access_token`); the bot must be the message sender and must be in the conversation that contains the message.
+-
+-### reactions
+-
+-  - `batch_query` — 批量获取消息表情。Identity: supports `user` and `bot`.[Must-read](references/lark-im-reactions.md)
+-  - `create` — 添加消息表情回复。Identity: supports `user` and `bot`; the caller must be in the conversation that contains the message.[Must-read](references/lark-im-reactions.md)
+-  - `delete` — 删除消息表情回复。Identity: supports `user` and `bot`; the caller must be in the conversation that contains the message, and can only delete reactions added by itself.[Must-read](references/lark-im-reactions.md)
+-  - `list` — 获取消息表情回复。Identity: supports `user` and `bot`; the caller must be in the conversation that contains the message.[Must-read](references/lark-im-reactions.md)
+-
+-### threads
+-
+-  - `forward` — 转发话题。Identity: supports `user` and `bot`.
+-
+-### images
+-
+-  - `create` — 上传图片。Identity: `bot` only (`tenant_access_token`).
+-
+-### pins
+-
+-  - `create` — Pin 消息。Identity: supports `user` and `bot`.
+-  - `delete` — 移除 Pin 消息。Identity: supports `user` and `bot`.
+-  - `list` — 获取群内 Pin 消息。Identity: supports `user` and `bot`.
+-
+-### feed.groups
+-
+-  - `batch_add_item` — Batch add feed cards to a feed group. Identity: `user` only (`user_access_token`).[Must-read](references/lark-im-feed-groups.md)
+-  - `batch_query` — Batch query feed groups. Identity: `user` only (`user_access_token`).[Must-read](references/lark-im-feed-groups.md)
+-  - `batch_remove_item` — Batch remove feed cards from a feed group. Identity: `user` only (`user_access_token`).[Must-read](references/lark-im-feed-groups.md)
+-  - `create` — Create a feed group. Identity: `user` only (`user_access_token`).[Must-read](references/lark-im-feed-groups.md)
+-  - `delete` — Delete a feed group. Identity: `user` only (`user_access_token`).[Must-read](references/lark-im-feed-groups.md)
+-  - `update` — Update a feed group. Identity: `user` only (`user_access_token`).[Must-read](references/lark-im-feed-groups.md)
+-
+-## 权限表
+-
+-| 方法 | 所需 scope |
+-|------|-----------|
+-| `chats.create` | `im:chat:create` |
+-| `chats.get` | `im:chat:read` |
+-| `chats.link` | `im:chat:read` |
+-| `chats.update` | `im:chat:update` |
+-| `chat.members.bots` | `im:chat.members:read` |
+-| `chat.members.create` | `im:chat.members:write_only` |
+-| `chat.members.delete` | `im:chat.members:write_only` |
+-| `chat.members.get` | `im:chat.members:read` |
+-| `chat.user_setting.batch_query` | `im:chat.user_setting:read` |
+-| `chat.user_setting.batch_update` | `im:chat.user_setting:write` |
+-| `chat.managers.add_managers` | `im:chat.managers:write_only` |
+-| `chat.managers.delete_managers` | `im:chat.managers:write_only` |
+-| `chat.moderation.get` | `im:chat.moderation:read` |
+-| `chat.moderation.update` | `im:chat:moderation:write_only` |
+-| `messages.delete` | `im:message:recall` |
+-| `messages.forward` | `im:message` |
+-| `messages.merge_forward` | `im:message` |
+-| `messages.read_users` | `im:message:readonly` |
+-| `messages.urgent_app` | `im:message.urgent` |
+-| `messages.urgent_phone` | `im:message.urgent:phone` |
+-| `messages.urgent_sms` | `im:message.urgent:sms` |
+-| `reactions.batch_query` | `im:message.reactions:read` |
+-| `reactions.create` | `im:message.reactions:write_only` |
+-| `reactions.delete` | `im:message.reactions:write_only` |
+-| `reactions.list` | `im:message.reactions:read` |
+-| `threads.forward` | `im:message` |
+-| `images.create` | `im:resource` |
+-| `pins.create` | `im:message.pins:write_only` |
+-| `pins.delete` | `im:message.pins:write_only` |
+-| `pins.list` | `im:message.pins:read` |
+-| `feed.groups.batch_add_item` | `im:feed_group_v1:write` |
+-| `feed.groups.batch_query` | `im:feed_group_v1:read` |
+-| `feed.groups.batch_remove_item` | `im:feed_group_v1:write` |
+-| `feed.groups.create` | `im:feed_group_v1:write` |
+-| `feed.groups.delete` | `im:feed_group_v1:write` |
+-| `feed.groups.update` | `im:feed_group_v1:write` |
++> **MUST** run `schema` before any native call: it is the live source for the `--data` / `--params` structure, the supported identity (`--as user` vs `--as bot`), owner/admin/tenant constraints, and the required `im:*` scope — do not guess. On a missing-scope error, lark-cli returns a `console_url`; follow the lark-shared permission-handling flow.
+-- 
+2.50.1 (Apple Git-155)
+

harness-opt/verify_results/round-002-lark-im-references-lark-im-messages-send.patch

@@ -0,0 +1,334 @@
+From 82a099feafb45d101116f10230ce7c2f92fbcfe5 Mon Sep 17 00:00:00 2001
+From: "zhangheng.023" <zhangheng.023@bytedance.com>
+Date: Tue, 23 Jun 2026 19:17:24 +0800
+Subject: [PATCH] =?UTF-8?q?opt(round-002):=20lark-im-messages-send.md=20?=
+ =?UTF-8?q?=E2=80=94=20consolidate=204x=20repeated=20content-flag=20rule,?=
+ =?UTF-8?q?=20compress=20media=20enumeration=20&=20--help-mirror=20section?=
+ =?UTF-8?q?s?=
+MIME-Version: 1.0
+Content-Type: text/plain; charset=UTF-8
+Content-Transfer-Encoding: 8bit
+
+---
+ .../references/lark-im-messages-send.md       | 259 ++++--------------
+ 1 file changed, 51 insertions(+), 208 deletions(-)
+
+diff --git a/skills/lark-im/references/lark-im-messages-send.md b/skills/lark-im/references/lark-im-messages-send.md
+index 484c024f..32818909 100644
+--- a/skills/lark-im/references/lark-im-messages-send.md
++++ b/skills/lark-im/references/lark-im-messages-send.md
+@@ -1,10 +1,8 @@
+ # im +messages-send
+ 
+-> **Prerequisite:** Read [`../lark-shared/SKILL.md`](../../lark-shared/SKILL.md) first to understand authentication, global parameters, and safety rules.
++> **Prerequisite:** Read [`../lark-shared/SKILL.md`](../../lark-shared/SKILL.md) first for authentication, global parameters, and safety rules.
+ 
+-Send a message to a group chat or a direct message conversation. Supports both user identity (`--as user`) and bot identity (`--as bot`).
+-
+-This skill maps to the shortcut: `lark-cli im +messages-send` (internally calls `POST /open-apis/im/v1/messages`).
++Send a message to a group chat (`--chat-id oc_xxx`) or a direct message (`--user-id ou_xxx`). One step, supports `--as user` and `--as bot` (default `bot`). Maps to shortcut `lark-cli im +messages-send` (`POST /open-apis/im/v1/messages`).
+ 
+ ## Safety Constraints
+ 
+@@ -16,249 +14,94 @@ Messages sent by this tool are visible to other people. Before calling it, you *
+ 
+ **Do not** send messages without explicit user approval.
+ 
+-When using `--as bot`, the message is sent in the app's name, so make sure the app has already been added to the target chat.
+-
+-When using `--as user`, the message is sent as the authorized end user and requires the `im:message.send_as_user` and `im:message` scopes.
++- `--as bot` (TAT, scope `im:message:send_as_bot`): the message is sent in the app's name — the app must already be in the target chat or have a DM relationship with the target user.
++- `--as user` (UAT, scopes `im:message.send_as_user` + `im:message`): the message is sent as the authorized end user.
+ 
+ ## Choose The Right Content Flag
+ 
+-### Default Selection Rule For Agents
+-
+-- Prefer `--markdown` for headings, lists, links, summaries, reports, or Markdown-looking content.
+-- Use `--text` for exact plain text: logs, code, indentation-sensitive text, or literal Markdown.
+-- Use `--content` for exact `post` JSON, titles, multiple locales, cards, or unsupported structures.
+-
+-| Need | Recommended flag | Why |
+-|------|------|------|
+-| Send headings, lists, links, summaries, or reports | `--markdown` | Best default for lightweight formatting; converted to Feishu `post` JSON |
+-| Send plain text exactly as written | `--text` | Preserves literal text; no Markdown conversion |
+-| Precisely control the final payload | `--content` | You provide the exact JSON for `text` / `post` / `interactive` / `share_*` / media payloads |
+-| Send image / file / video / audio | `--image` / `--file` / `--video` / `--audio` | Shortcut uploads URLs, or cwd-relative local files automatically |
+-
+-### `--text` vs `--markdown`
+-
+-- Use `--markdown` for lightweight formatted messages.
+-- Use `--text` for exact plain text, especially logs, code, indentation, or Markdown characters that should **not** render.
+-- Use `--content` when `--markdown` is not enough, especially if you need exact `post` JSON, a title, multiple locales, cards, or unsupported rich structures.
+-
+-## What `--markdown` Really Does
+-
+-`--markdown` accepts Markdown-like input and converts it to the Feishu `post` payload required by the message API.
+-
+-The shortcut does all of the following before sending:
+-
+-1. Forces `msg_type=post`
+-2. Resolves remote Markdown images like `![x](https://...)` by downloading and uploading them first
+-3. Normalizes the Markdown for Feishu post rendering
+-4. Wraps the result as:
+-
+-```json
+-{"zh_cn":{"content":[[{"tag":"md","text":"..."}]]}}
+-```
+-
+-This makes `--markdown` the simplest path for lightweight formatted messages.
+-
+-### Markdown Boundaries
+-
+-- It does **not** promise full CommonMark / GitHub Flavored Markdown support.
+-- It always becomes a `post` payload with a single `zh_cn` locale.
+-- It does **not** let you set a `post` title. If you need a title, use `--msg-type post --content ...`.
+-- Headings are rewritten:
+-    - `# Title` becomes `#### Title`
+-    - `##` to `######` are normalized to `#####` when the content contains H1-H3
+-- Consecutive headings are separated with blank lines after heading normalization.
+-- Block spacing and line breaks may be normalized during conversion.
+-- Code blocks are preserved as code blocks.
+-- Excess blank lines are compressed.
+-- Already-uploaded `img_xxx` image keys are the most reliable Markdown image input.
+-- Local paths in Markdown image syntax like `![x](./a.png)` are **not** supported and will not be auto-uploaded.
+-- Remote URLs (`https://...`) will be auto-downloaded and uploaded at runtime; if the download or upload fails, the image is removed with a warning.
+-
+-If you need a title, multiple locales, cards, unsupported rich structures, or byte-for-byte post JSON control, use `--content` and provide the final JSON yourself.
++| Content | Flag | Why |
++|---|---|---|
++| Headings, lists, links, summaries, reports (lightweight formatting) | `--markdown` | Best default; converted to Feishu `post` JSON |
++| Exact plain text — logs, code, indentation, literal Markdown chars that must **not** render | `--text` | Preserves literal text; no conversion |
++| Exact `post` JSON, a `post` title, multiple locales, cards (`interactive`), `share_*`, or unsupported structures | `--content` | You provide the final JSON; it must match the effective `--msg-type` |
++| Image / file / video / audio | `--image` / `--file` / `--video` / `--audio` | Uploads URLs or cwd-relative local files automatically |
+ 
+-### Image Constraint for `--markdown`
++These content flags (and the media flags) are **mutually exclusive** — pass exactly one. Media flags are also mutually exclusive with each other.
+ 
+-When using `--markdown` with images, prefer pre-uploading via `images.create` and referencing `![alt](img_xxx)` for predictable results. Remote URLs may work but are not guaranteed.
++## `--markdown` Gotchas
+ 
+-**Steps:**
++`--markdown` always forces `msg_type=post` (single `zh_cn` locale) and normalizes input for Feishu post rendering. Key boundaries (not full CommonMark/GFM):
+ 
+-```bash
+-# 1. Upload image to get image_key
+-lark-cli im images create --data '{"image_type":"message"}' --file ./diagram.png
+-# Returns: {"image_key":"img_v3_xxxx"}
+-
+-# 2. Use image_key in --markdown
+-lark-cli im +messages-send --chat-id oc_xxx --markdown $'## Report\n\n![diagram](img_v3_xxxx)\n\nSee above for details.'
+-```
++- **No `post` title** — if you need one, use `--content` with `post` JSON.
++- **Headings rewritten**: `# Title` → `#### Title`; `##`–`######` normalized to `#####` when content has H1–H3. Code blocks preserved; excess blank lines compressed.
++- **Images**: pre-upload via `im images create` and reference `![alt](img_xxx)` for reliable results. Remote `https://` URLs are auto-downloaded+uploaded at runtime (removed with a warning if that fails). Local paths in `![x](./a.png)` are **not** supported and will not auto-upload.
+ 
+-## Preserving Formatting
++## Preserving Exact Formatting
+ 
+-If the message has multiple lines, indentation, code blocks, tabs, or many quotes/backslashes, prefer shell ANSI-C quoting with `$'...'` for either `--markdown` or `--text`.
+-
+-This is especially useful in `zsh` / `bash` because it lets you write `\n` explicitly instead of relying on the shell to preserve literal newlines.
+-
+-### When formatting must be preserved
+-
+-Use `--text` plus `$'...'`:
+-
+-```bash
+-lark-cli im +messages-send --chat-id oc_xxx --text $'Build failed\nBranch: feature/im-docs\nAction: please check logs'
+-```
++For multi-line text, indentation, code blocks, tabs, or many backslashes/quotes, use shell ANSI-C quoting `$'...'` so `\n` is written explicitly. Use `--text` + `$'...'` when the receiver must see the text exactly as entered:
+ 
+ ```bash
+-lark-cli im +messages-send --chat-id oc_xxx --text $'```bash\nmake test\nmake lint\n```'
++lark-cli im +messages-send --chat-id oc_xxx --text $'Build failed\nBranch: feature/x\nAction: check logs'
+ ```
+ 
+-Use this path when you want the receiver to see the text exactly as entered, not a converted Markdown post.
+-
+ ## Commands
+ 
+ ```bash
+-# Send a formatted update
++# Formatted update (Markdown → post)
+ lark-cli im +messages-send --chat-id oc_xxx --markdown $'## Update\n\n- item 1\n- item 2'
+ 
+-# Send a plain one-line message
++# Plain one-line text
+ lark-cli im +messages-send --chat-id oc_xxx --text "Hello"
+ 
+-# Equivalent manual JSON
+-lark-cli im +messages-send --chat-id oc_xxx --content '{"text":"Hello"}'
+-
+-# Send to a direct message (pass open_id)
++# Direct message (pass open_id)
+ lark-cli im +messages-send --user-id ou_xxx --text "Hello"
+ 
+-# Send multi-line text while preserving formatting
+-lark-cli im +messages-send --chat-id oc_xxx --text $'Line 1\nLine 2\n  indented line'
+-
+-# Send Markdown with an image (must pre-upload via images.create)
+-lark-cli im images create --data '{"image_type":"message"}' --file ./screenshot.png
+-# Use the returned image_key in the markdown content
+-lark-cli im +messages-send --chat-id oc_xxx --markdown $'## Status\n\n![screenshot](img_v3_xxxx)\n\nDone.'
+-
+-# If you need exact post structure, send JSON directly
++# Exact post structure with a title
+ lark-cli im +messages-send --chat-id oc_xxx --msg-type post --content '{"zh_cn":{"title":"Title","content":[[{"tag":"text","text":"Body"}]]}}'
+ 
+-# Send a local image (uploaded automatically before sending)
+-lark-cli im +messages-send --chat-id oc_xxx --image ./photo.png
+-
+-# Or send directly with an existing image_key
+-lark-cli im +messages-send --chat-id oc_xxx --image img_xxx
++# Markdown with an image (pre-upload first)
++lark-cli im images create --data '{"image_type":"message"}' --file ./diagram.png   # -> {"image_key":"img_v3_xxxx"}
++lark-cli im +messages-send --chat-id oc_xxx --markdown $'## Report\n\n![diagram](img_v3_xxxx)\n\nDone.'
+ 
+-# Send a local file (uploaded automatically before sending)
++# Media (local files uploaded automatically; --video requires --video-cover)
++lark-cli im +messages-send --chat-id oc_xxx --image ./photo.png
+ lark-cli im +messages-send --chat-id oc_xxx --file ./report.pdf
+-
+-# Send a video (--video-cover is required as the cover)
+ lark-cli im +messages-send --chat-id oc_xxx --video ./demo.mp4 --video-cover ./cover.png
+-lark-cli im +messages-send --chat-id oc_xxx --video ./demo.mp4 --video-cover img_xxx
+-
+-# Send audio
+ lark-cli im +messages-send --chat-id oc_xxx --audio ./voice.opus
+ 
+-# Use an idempotency key (same key sends only once within 1 hour)
+-lark-cli im +messages-send --chat-id oc_xxx --text "Hello" --idempotency-key my-unique-id
+-
+-# Preview the request without executing it
+-lark-cli im +messages-send --chat-id oc_xxx --markdown $'## Test\n\nhello' --dry-run
++# Idempotency (same key sends only once within 1 hour) / preview without sending
++lark-cli im +messages-send --chat-id oc_xxx --text "Hi" --idempotency-key my-id
++lark-cli im +messages-send --chat-id oc_xxx --markdown $'## Test\n\nhi' --dry-run
+ ```
+ 
+-## Media Input Rules
+-
+-- Media flags accept an existing key (`img_xxx` / `file_xxx`), an `http://` or `https://` URL, or a local file path.
+-- Local paths must be relative to the current working directory and stay within it after resolving `..` and symlinks.
+-- Absolute paths such as `/tmp/photo.png` are rejected. Run the command from the file's directory and pass `./photo.png`, or copy the file into the current directory first.
+-
+-## Parameters
+-
+-| Parameter | Required | Description                                                                                                                                                                                   |
+-|------|------|-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
+-| `--chat-id <id>` | One of two | Group chat ID (`oc_xxx`)                                                                                                                                                                      |
+-| `--user-id <id>` | One of two | User open_id (`ou_xxx`) for direct messages                                                                                                                                                   |
+-| `--text <string>` | One content option | Plain text message. Use when exact text and formatting preservation matter. Automatically wrapped as `{"text":"..."}`                                                                         |
+-| `--markdown <string>` | One content option | Best default for lightweight formatted messages such as headings, lists, links, summaries, and reports. Internally converted to `post` JSON with Feishu-specific normalization               |
+-| `--content <json>` | One content option | Exact message content JSON string; use this when you need full control over `msg_type` and payload. The JSON must match the effective `--msg-type`                                            |
+-| `--image <path\|url\|key>` | One content option | Cwd-relative local image path, URL, or `image_key` (`img_xxx`). Local paths and URLs are uploaded automatically                                                                               |
+-| `--file <path\|url\|key>` | One content option | Cwd-relative local file path, URL, or `file_key` (`file_xxx`). Local paths and URLs are uploaded automatically                                                                                |
+-| `--video <path\|url\|key>` | One content option | Cwd-relative local video path, URL, or `file_key` (`file_xxx`). Local paths and URLs are uploaded automatically. **Must be paired with `--video-cover`**                                      |
+-| `--video-cover <path\|url\|key>` | **Required with `--video`** | Cwd-relative local cover image path, URL, or `image_key` (`img_xxx`). Local paths and URLs are uploaded automatically                                                                         |
+-| `--audio <path\|url\|key>` | One content option | Cwd-relative local audio path, URL, or `file_key` (`file_xxx`). Local paths and URLs are uploaded automatically                                                                                           |
+-| `--msg-type <type>` | No | Message type (default `text`). If you use `--text` / `--markdown` / media flags, the effective type is inferred automatically. Explicitly setting a conflicting `--msg-type` fails validation |
+-| `--idempotency-key <key>` | No | Idempotency key; the same key sends only one message within 1 hour                                                                                                                            |
+-| `--as <identity>` | No | Identity type: `bot` or `user` (default `bot`)                                                                                                                                                |
+-| `--dry-run` | No | Print the request only, do not execute it                                                                                                                                                     |
+-
+-> **Mutual exclusivity rule:** `--text`, `--markdown`, `--content`, and `--image`/`--file`/`--video`/`--audio` cannot be used together. Media flags are also mutually exclusive with each other.
+->
+-> **Video cover rule:** `--video` **must** be accompanied by `--video-cover`. Omitting `--video-cover` when using `--video` will fail validation. `--video-cover` cannot be used without `--video`.
+-
+-## Common Mistakes
+-
+-- Choosing `--text` for headings, lists, links, summaries, or reports. Use `--markdown`.
+-- Choosing `--markdown` when you actually need exact plain text. If exact line breaks, spacing, logs, code, or literal Markdown characters matter, use `--text`, usually with `$'...'`.
+-- Assuming `--markdown` supports every Markdown feature. It is converted into a Feishu `post` payload and normalized first.
+-- Putting local image paths inside Markdown like `![x](./a.png)`. `--markdown` does not auto-upload those paths.
+-- **Using local file paths inside Markdown image syntax** (e.g. `![x](./a.png)`) with `--markdown`. Local paths are not auto-uploaded and will not render as an image. Pre-upload via `images.create` to get an `image_key` instead.
+-- Using `--content` without making the JSON match the effective `--msg-type`.
+-- Explicitly setting `--msg-type` to something that conflicts with `--text`, `--markdown`, or media flags.
+-- Mixing `--text`, `--markdown`, or `--content` with media flags in one command.
+-
+-## `content` Format Reference
++Run `lark-cli im +messages-send --help` for the full flag list and types. Load-bearing rules that `--help` may not make obvious:
++
++- **Media paths** accept an existing key (`img_xxx`/`file_xxx`), an `http(s)://` URL, or a **cwd-relative** local path. Absolute paths (e.g. `/tmp/x.png`) are rejected — run from the file's directory and pass `./x.png`. Upload and send use the same identity.
++- **`--video` must be paired with `--video-cover`** (image key/URL/local path); `--video-cover` cannot be used alone.
++- **`--msg-type`** is inferred from `--text`/`--markdown`/media flags; explicitly setting a conflicting type fails validation.
++
++## `content` Format Reference (for `--content`)
+ 
+ | `msg_type` | Example `content` |
+-|----------|-------------|
++|---|---|
+ | `text` | `{"text":"Hello <at user_id=\"ou_xxx\">name</at>"}` |
+ | `post` | `{"zh_cn":{"title":"Title","content":[[{"tag":"text","text":"Body"}]]}}` |
+-| `image` | `{"image_key":"img_xxx"}` |
+-| `file` | `{"file_key":"file_xxx"}` |
+-| `audio` | `{"file_key":"file_xxx"}` |
+-| `media` | `{"file_key":"file_xxx","image_key":"img_xxx"}` (video; `image_key` is the cover from `--video-cover` — **required**) |
+-| `share_chat` | `{"chat_id":"oc_xxx"}` |
+-| `share_user` | `{"user_id":"ou_xxx"}` |
+-| `interactive` | Card JSON (see Feishu interactive card documentation) |
++| `image` / `file` / `audio` | `{"image_key":"img_xxx"}` / `{"file_key":"file_xxx"}` / `{"file_key":"file_xxx"}` |
++| `media` (video) | `{"file_key":"file_xxx","image_key":"img_xxx"}` (`image_key` is the **required** cover) |
++| `share_chat` / `share_user` | `{"chat_id":"oc_xxx"}` / `{"user_id":"ou_xxx"}` |
++| `interactive` (card) | Card JSON (see Feishu interactive card docs) |
+ 
+-## Return Value
+-
+-```json
+-{
+-  "message_id": "om_xxx",
+-  "chat_id": "oc_xxx",
+-  "create_time": "1234567890"
+-}
+-```
++When using `--content`, you are responsible for making the JSON match the effective `msg_type`.
+ 
+ ## @Mention Format
+ 
+-The `<at>` syntax differs by message type. The shortcut only normalizes mentions for `text` and `post`; `interactive` card content is passed through verbatim, so cards must use the card-native syntax below.
+-
+-### `text`
+-
+-- `<at user_id="ou_xxx">name</at>` — the inner text is the mentioned user's display name and is optional (`<at user_id="ou_xxx"></at>` also works)
+-- @all: `<at user_id="all"></at>`
+-
+-### `post`
++The `<at>` syntax differs by message type; the shortcut normalizes mentions for `text` and `post` only — `interactive` cards are passed through verbatim.
+ 
+-- Inside a `text` or `md` element, the same inline form as `text` works: `<at user_id="ou_xxx">name</at>`
+-- Or use a dedicated `at` element node: `{"tag":"at","user_id":"ou_xxx"}` (use `"all"` to mention everyone)
++- **`text`** / inside a `post` `text`/`md` element: `<at user_id="ou_xxx">name</at>` (inner name optional); @all: `<at user_id="all"></at>`. In `post` you may also use a node: `{"tag":"at","user_id":"ou_xxx"}` (`"all"` for everyone).
++- **`interactive` (card)** — card-native syntax inside a `lark_md`/`markdown` element: `<at id=ou_xxx></at>`, multiple `<at ids=ou_1,ou_2></at>`, by email `<at email=user@example.com></at>`.
+ 
+-### `interactive` (card)
+-
+-Card content is **not** normalized — use the card-native `<at>` syntax inside a `lark_md` / `markdown` element:
+-
+-- single user by open_id: `<at id=ou_xxx></at>`
+-- multiple users: `<at ids=ou_xxx1,ou_xxx2></at>`
+-- by email: `<at email=user@example.com></at>`
+-
+-## Notes
++## Return Value
+ 
+-- `--chat-id` and `--user-id` are mutually exclusive; you must provide exactly one
+-- `--content` must be valid JSON
+-- When using `--content`, you are responsible for making the JSON structure match the effective `msg_type`
+-- `--image`/`--file`/`--video`/`--audio` support existing keys, URLs, and cwd-relative local file paths; the shortcut uploads local paths and URLs first, then sends the message; both the upload and send steps use the same identity (UAT when `--as user`, TAT when `--as bot`)
+-- If the provided media value starts with `img_` or `file_`, it is treated as an existing key and used directly
+-- `--markdown` always sends `msg_type=post`, even if you do not explicitly set `--msg-type post`
+-- If you explicitly set `--msg-type` and it conflicts with the chosen content flag, validation fails
+-- When using `--video`, `--video-cover` is required as the video cover
+-- `--dry-run` uses placeholder image keys for remote Markdown images and placeholder media keys for local uploads
+-- Failures return an error code and message
+-- `--as user` uses a user access token (UAT) and requires the `im:message.send_as_user` and `im:message` scopes; the message is sent as the authorized end user
+-- `--as bot` uses a tenant access token (TAT) and requires the `im:message:send_as_bot` scope
+-- When sending as a bot, the app must already be in the target group or already have a direct-message relationship with the target user
+-- When using `--markdown` with images, pre-uploading via `images.create` to obtain an `image_key` is recommended for reliability; remote URLs may be auto-resolved at runtime, but if download/upload fails the image is removed with a warning; local paths are not supported
++```json
++{"message_id": "om_xxx", "chat_id": "oc_xxx", "create_time": "1234567890"}
++```
+-- 
+2.50.1 (Apple Git-155)
+

harness-opt/verify_results/round-003-lark-im-references-lark-im-chat-create.patch

@@ -0,0 +1,135 @@
+From cbd6e56ac07285fd973c53ff7382da0112b6cf5d Mon Sep 17 00:00:00 2001
+From: "zhangheng.023" <zhangheng.023@bytedance.com>
+Date: Tue, 23 Jun 2026 19:51:49 +0800
+Subject: [PATCH] =?UTF-8?q?opt(round-003):=20references/lark-im-chat-creat?=
+ =?UTF-8?q?e.md=20=E2=80=94=20dedup=20Commands/Scenarios=20overlap=20+=20c?=
+ =?UTF-8?q?ompress=20--help-mirroring=20Common=20Errors=20into=20pointers,?=
+ =?UTF-8?q?=20keep=20232043=20two-step=20flow=20&=20all=20guardrails?=
+MIME-Version: 1.0
+Content-Type: text/plain; charset=UTF-8
+Content-Transfer-Encoding: 8bit
+
+---
+ .../lark-im/references/lark-im-chat-create.md | 78 +++++--------------
+ 1 file changed, 18 insertions(+), 60 deletions(-)
+
+diff --git a/skills/lark-im/references/lark-im-chat-create.md b/skills/lark-im/references/lark-im-chat-create.md
+index 76716f76..7d65e5d3 100644
+--- a/skills/lark-im/references/lark-im-chat-create.md
++++ b/skills/lark-im/references/lark-im-chat-create.md
+@@ -12,43 +12,24 @@ This skill maps to the shortcut: `lark-cli im +chat-create` (internally calls `P
+ ## Commands
+ 
+ ```bash
+-# Create a private group (default)
++# Private group (default)
+ lark-cli im +chat-create --name "My Group"
+ 
+-# Create a public group (name is required and must be at least 2 characters)
++# Public group (--name required, min 2 chars)
+ lark-cli im +chat-create --name "Public Group" --type public
+ 
+-# Create a topic chat
++# Topic chat (a 话题群; see note under Parameters)
+ lark-cli im +chat-create --name "Topic Group" --chat-mode topic
+ 
+-# Specify the group owner
+-lark-cli im +chat-create --name "My Group" --owner ou_xxx
++# Invite members and set owner (users: up to 50 ou_xxx; bots: up to 5 cli_xxx)
++lark-cli im +chat-create --name "My Group" --owner ou_xxx --users "ou_aaa,ou_bbb" --bots "cli_aaa"
+ 
+-# Invite user members (comma-separated open_ids, up to 50)
+-lark-cli im +chat-create --name "My Group" --users "ou_aaa,ou_bbb"
+-
+-# Invite bot members (comma-separated app IDs, up to 5)
+-lark-cli im +chat-create --name "My Group" --bots "cli_aaa,cli_bbb"
+-
+-# Invite both users and bots
+-lark-cli im +chat-create --name "My Group" --users "ou_aaa" --bots "cli_aaa"
+-
+-# Make the creating bot a group manager (bot identity only)
+-lark-cli im +chat-create --name "My Group" --set-bot-manager --as bot
+-
+-# JSON output
+-lark-cli im +chat-create --name "My Group" --format json
+-
+-# Create a group with bot identity
+-lark-cli im +chat-create --name "My Group" --users "ou_aaa" --as bot
+-
+-# Create a group with user identity
+-lark-cli im +chat-create --name "My Group" --users "ou_aaa,ou_bbb" --as user
+-
+-# Preview the request without creating anything
+-lark-cli im +chat-create --name "My Group" --dry-run
++# Bot identity, making the creating bot a manager
++lark-cli im +chat-create --name "My Group" --users "ou_aaa" --as bot --set-bot-manager
+ ```
+ 
++Run `lark-cli im +chat-create --help` for the full flag list, limits, and types. Single-flag variations (`--as user`, `--description`, `--format json`, `--dry-run` preview, etc.) follow the Parameters table below — `--dry-run` previews the request without creating anything.
++
+ ## Parameters
+ 
+ | Parameter | Required | Limits | Description |
+@@ -106,6 +87,13 @@ lark-cli im +chat-create --name "<group name>" --users "ou_aaa,ou_bbb" --as user
+ 
+ The authorized user is automatically the group creator and member.
+ 
++### Create a group, then send a welcome message
++
++```bash
++CHAT_ID=$(lark-cli im +chat-create --name "New Group" --format json | jq -r '.data.chat_id')
++lark-cli im +messages-send --chat-id "$CHAT_ID" --text "Welcome, everyone!"
++```
++
+ ## Output Fields
+ 
+ | Field | Description |
+@@ -117,43 +105,13 @@ The authorized user is automatically the group creator and member.
+ | `external` | Whether the group is external |
+ | `share_link` | Group share link (omitted if retrieval fails) |
+ 
+-## Usage Scenarios
+-
+-### Scenario 1: Create a group and specify the owner
+-
+-```bash
+-lark-cli im +chat-create --name "Project Discussion Group" --owner ou_xxx
+-```
+-
+-### Scenario 2: Create a group and invite users and a bot
+-
+-```bash
+-lark-cli im +chat-create --name "Project Discussion Group" \
+-  --owner ou_xxx \
+-  --users "ou_aaa,ou_bbb" \
+-  --bots "cli_aaa"
+-```
+-
+-### Scenario 3: Create a group and send a welcome message
+-
+-```bash
+-CHAT_ID=$(lark-cli im +chat-create --name "New Group" --format json | jq -r '.data.chat_id')
+-lark-cli im +messages-send --chat-id "$CHAT_ID" --text "Welcome, everyone!"
+-```
+-
+ ## Common Errors and Troubleshooting
+ 
++Format/limit validation (`--name`/`--description`/`--users`/`--bots`/`--owner` length, count, and `ou_xxx`/`cli_xxx` format) is enforced by the CLI and reported verbatim with the fix — see the Parameters table for limits. The two errors needing extra action:
++
+ | Symptom | Root Cause | Solution |
+ |---------|---------|---------|
+ | Permission denied (99991672) | The app does not have `im:chat:create` (bot) or `im:chat:create_by_user` (user) permission enabled | Enable the required permission for the app in the Open Platform console |
+-| `--name is required for public groups and must be at least 2 characters` | A public group was created without a name or with a name shorter than 2 characters | Provide a name with at least 2 characters |
+-| `--name exceeds the maximum of 60 characters` | The group name is too long | Shorten the name to 60 characters or fewer |
+-| `--description exceeds the maximum of 100 characters` | The group description is too long | Shorten the description to 100 characters or fewer |
+-| `--users exceeds the maximum of 50` | Too many user members were provided | Split the operation into batches and add more members later |
+-| `--bots exceeds the maximum of 5` | Too many bot members were provided | Invite at most 5 bots at once |
+-| `invalid user id: expected open_id (ou_xxx)` | Invalid user ID format | Use the `ou_xxx` format for users |
+-| `invalid bot id: expected app ID (cli_xxx)` | Invalid bot ID format | Use the `cli_xxx` format for bots |
+-| `invalid --owner: expected open_id (ou_xxx)` | Invalid owner ID format | Use the `ou_xxx` format for the owner |
+ | `bot is invisible to user` (232043) | The bot and target users are mutually invisible | Follow the two-step flow in AI Usage Guidance above — do not pass other users in `--users` during creation |
+ 
+ ## References
+-- 
+2.50.1 (Apple Git-155)
+

internal/meta/normalize.go

@@ -113,7 +113,8 @@ type EnumOption struct {
 }
 
 // EnumOptions returns the field's allowed values paired with their descriptions
-// — from enum, or from options when enum is absent — coerced to the canonical
+// — from enum (with descriptions backfilled from options when the field carries
+// both forms), or from options when enum is absent — coerced to the canonical
 // type and ordered: numeric and boolean values are sorted; string values keep
 // source order (which can encode priority). Uncoercible literals are dropped.
 // Returns nil when the field declares no enum constraint.
@@ -122,9 +123,14 @@ func (f Field) EnumOptions() []EnumOption {
 	var out []EnumOption
 	switch {
 	case len(f.Enum) > 0:
+		// key by raw literal so enum "1" and option 1 align across JSON types
+		desc := make(map[string]string, len(f.Options))
+		for _, o := range f.Options {
+			desc[fmt.Sprintf("%v", o.Value)] = o.Description
+		}
 		for _, e := range f.Enum {
 			if v, ok := coerceLiteral(ct, e); ok {
-				out = append(out, EnumOption{Value: v})
+				out = append(out, EnumOption{Value: v, Description: desc[fmt.Sprintf("%v", e)]})
 			}
 		}
 	case len(f.Options) > 0:

internal/meta/normalize_test.go

@@ -80,6 +80,39 @@ func TestField_EnumOptions(t *testing.T) {
 	}
 }
 
+func TestField_EnumOptions_BothEnumAndOptions(t *testing.T) {
+	// enum is the value set; descriptions backfilled from options, empty where absent
+	f := Field{Type: "string", Enum: []any{"1", "2", "3", "4", "6"}, Options: []Option{
+		{Value: "1", Description: "from"},
+		{Value: "2", Description: "to"},
+		{Value: "6", Description: "subject"},
+	}}
+	want := []EnumOption{
+		{Value: "1", Description: "from"},
+		{Value: "2", Description: "to"},
+		{Value: "3", Description: ""},
+		{Value: "4", Description: ""},
+		{Value: "6", Description: "subject"},
+	}
+	if got := f.EnumOptions(); !reflect.DeepEqual(got, want) {
+		t.Errorf("EnumOptions(enum+options) = %+v, want %+v", got, want)
+	}
+
+	// enum values stored as strings match option values stored as numbers
+	fi := Field{Type: "integer", Enum: []any{"10", "2", "1"}, Options: []Option{
+		{Value: 1, Description: "one"},
+		{Value: 2, Description: "two"},
+	}}
+	wantI := []EnumOption{
+		{Value: int64(1), Description: "one"},
+		{Value: int64(2), Description: "two"},
+		{Value: int64(10), Description: ""},
+	}
+	if got := fi.EnumOptions(); !reflect.DeepEqual(got, wantI) {
+		t.Errorf("EnumOptions(integer enum+options) = %+v, want %+v", got, wantI)
+	}
+}
+
 func TestField_Enum_NumberAndBoolean(t *testing.T) {
 	// number: string-stored floats coerced to float64 and numerically sorted
 	if got := (Field{Type: "number", Enum: []any{"2.5", "1.5", "10"}}).EnumValues(); !reflect.DeepEqual(got, []any{1.5, 2.5, float64(10)}) {

internal/schema/assembler_test.go

@@ -472,6 +472,18 @@ func TestConvert_EnumDescriptions(t *testing.T) {
 	if bare.EnumDescriptions != nil {
 		t.Errorf("bare enum must have nil EnumDescriptions, got %v", bare.EnumDescriptions)
 	}
+
+	// enum + options both present -> enumDescriptions backfilled, aligned, "" where absent
+	both := Convert(meta.Field{Type: "string", Enum: []any{"1", "2", "3"}, Options: []meta.Option{
+		{Value: "1", Description: "from"},
+		{Value: "2", Description: "to"},
+	}})
+	if !reflect.DeepEqual(both.Enum, []interface{}{"1", "2", "3"}) {
+		t.Errorf("both Enum = %v", both.Enum)
+	}
+	if !reflect.DeepEqual(both.EnumDescriptions, []string{"from", "to", ""}) {
+		t.Errorf("both EnumDescriptions = %v, want [from to \"\"] aligned with enum", both.EnumDescriptions)
+	}
 }
 
 func TestBuildMeta_AffordanceFromMethod(t *testing.T) {

package.json

@@ -1,6 +1,6 @@
 {
   "name": "@larksuite/cli",
-  "version": "1.0.55",
+  "version": "1.0.56",
   "description": "The official CLI for Lark/Feishu open platform",
   "bin": {
     "lark-cli": "scripts/run.js"

scripts/fetch_meta.py

@@ -89,7 +89,7 @@ def main():
     count = len(data.get("services", []))
     print(f"fetch-meta: OK, {count} services from remote API", file=sys.stderr)
 
-    with open(OUT_PATH, "w") as fp:
+    with open(OUT_PATH, "w", encoding="utf-8", newline="\n") as fp:
         json.dump(data, fp, ensure_ascii=False, indent=2)
         fp.write("\n")
 

shortcuts/apps/git_credential.go

@@ -17,6 +17,7 @@ import (
 	"text/tabwriter"
 	"time"
 
+	"github.com/google/uuid"
 	larkcore "github.com/larksuite/oapi-sdk-go/v3/core"
 	"github.com/spf13/cobra"
 
@@ -32,6 +33,7 @@ import (
 )
 
 const gitCredentialIssuePath = apiBasePath + "/apps/:app_id/git_info"
+const gitCredentialHelperReportedShortcut = appsService + ":+git-credential-helper"
 
 // gitCredentialIssueHint is the actionable next-step attached to a failed
 // Git-credential issuance. A 5xx is flagged retryable separately at the call site.
@@ -302,7 +304,12 @@ func (i factoryIssuer) Issue(ctx context.Context, appID string, profile gitcred.
 		HttpMethod: http.MethodGet,
 		ApiPath:    issuePath(appID),
 	}
-	resp, err := ac.DoSDKRequest(ctx, req, core.AsUser)
+	ctx = contextWithGitCredentialHelperShortcut(ctx)
+	var opts []larkcore.RequestOptionFunc
+	if optFn := cmdutil.ShortcutHeaderOpts(ctx); optFn != nil {
+		opts = append(opts, optFn)
+	}
+	resp, err := ac.DoSDKRequest(ctx, req, core.AsUser, opts...)
 	data, err := parseIssueCredentialData(resp, err, errclass.ClassifyContext{
 		Brand:    string(cfg.Brand),
 		AppID:    cfg.AppID,
@@ -314,6 +321,13 @@ func (i factoryIssuer) Issue(ctx context.Context, appID string, profile gitcred.
 	return issuedFromData(appID, data)
 }
 
+func contextWithGitCredentialHelperShortcut(ctx context.Context) context.Context {
+	if _, ok := cmdutil.ShortcutNameFromContext(ctx); ok {
+		return ctx
+	}
+	return cmdutil.ContextWithShortcut(ctx, gitCredentialHelperReportedShortcut, uuid.New().String())
+}
+
 func runGitCredentialHelper(ctx context.Context, f *cmdutil.Factory, appID, action string) error {
 	if f == nil || f.IOStreams == nil {
 		return nil

shortcuts/apps/git_credential_test.go

@@ -825,7 +825,7 @@ func TestRunGitCredentialHelperActions(t *testing.T) {
 func TestFactoryIssuerBranches(t *testing.T) {
 	factory, _, reg := newAppsExecuteFactory(t)
 	expiresAt := time.Now().Add(24 * time.Hour).Unix()
-	reg.Register(&httpmock.Stub{
+	issueStub := &httpmock.Stub{
 		Method: "GET",
 		URL:    "/open-apis/spark/v1/apps/app_xxx/git_info",
 		Body: map[string]interface{}{
@@ -836,7 +836,8 @@ func TestFactoryIssuerBranches(t *testing.T) {
 				"StatusCode": 0,
 			},
 		},
-	})
+	}
+	reg.Register(issueStub)
 	issued, err := (factoryIssuer{f: factory}).Issue(context.Background(), "app_xxx", gitcred.ProfileContext{})
 	if err != nil {
 		t.Fatalf("factory issuer returned error: %v", err)
@@ -844,6 +845,12 @@ func TestFactoryIssuerBranches(t *testing.T) {
 	if issued.PAT != "pat-token" {
 		t.Fatalf("PAT = %q", issued.PAT)
 	}
+	if got := issueStub.CapturedHeaders.Get(cmdutil.HeaderShortcut); got != gitCredentialHelperReportedShortcut {
+		t.Fatalf("%s = %q, want %q", cmdutil.HeaderShortcut, got, gitCredentialHelperReportedShortcut)
+	}
+	if got := issueStub.CapturedHeaders.Get(cmdutil.HeaderExecutionId); got == "" {
+		t.Fatalf("%s header missing", cmdutil.HeaderExecutionId)
+	}
 
 	factory.Config = func() (*core.CliConfig, error) { return nil, errors.New("config failed") }
 	if _, err := (factoryIssuer{f: factory}).Issue(context.Background(), "app_xxx", gitcred.ProfileContext{}); err == nil {
@@ -880,6 +887,20 @@ func TestFactoryIssuerBranches(t *testing.T) {
 	}
 }
 
+func TestContextWithGitCredentialHelperShortcutPreservesExistingShortcut(t *testing.T) {
+	ctx := cmdutil.ContextWithShortcut(context.Background(), "apps:+git-credential-init", "exec-existing")
+	got := contextWithGitCredentialHelperShortcut(ctx)
+
+	name, ok := cmdutil.ShortcutNameFromContext(got)
+	if !ok || name != "apps:+git-credential-init" {
+		t.Fatalf("shortcut = %q ok=%v, want existing shortcut", name, ok)
+	}
+	executionID, ok := cmdutil.ExecutionIdFromContext(got)
+	if !ok || executionID != "exec-existing" {
+		t.Fatalf("execution id = %q ok=%v, want existing execution id", executionID, ok)
+	}
+}
+
 func TestGitCredentialHelpersAndParsers(t *testing.T) {
 	if issuePath(" app/with space ") != "/open-apis/spark/v1/apps/app%2Fwith%20space/git_info" {
 		t.Fatalf("issuePath escaped incorrectly: %s", issuePath(" app/with space "))

shortcuts/common/runner.go

@@ -223,6 +223,12 @@ func (ctx *RuntimeContext) Float64(name string) float64 {
 	return v
 }
 
+// IntArray returns an int-array flag value (repeated flag, also supports CSV splitting).
+func (ctx *RuntimeContext) IntArray(name string) []int {
+	v, _ := ctx.Cmd.Flags().GetIntSlice(name)
+	return v
+}
+
 // StrArray returns a string-array flag value (repeated flag, no CSV splitting).
 func (ctx *RuntimeContext) StrArray(name string) []string {
 	v, _ := ctx.Cmd.Flags().GetStringArray(name)
@@ -1176,6 +1182,8 @@ func registerShortcutFlagsWithContext(ctx context.Context, cmd *cobra.Command, f
 			var d float64
 			fmt.Sscanf(fl.Default, "%g", &d)
 			cmd.Flags().Float64(fl.Name, d, desc)
+		case "int_array":
+			cmd.Flags().IntSlice(fl.Name, nil, desc)
 		case "string_array":
 			cmd.Flags().StringArray(fl.Name, nil, desc)
 		case "string_slice":

shortcuts/common/runner_args_test.go

@@ -4,9 +4,12 @@
 package common
 
 import (
+	"context"
+	"reflect"
 	"strings"
 	"testing"
 
+	"github.com/larksuite/cli/internal/cmdutil"
 	"github.com/spf13/cobra"
 )
 
@@ -56,3 +59,29 @@ func TestRejectPositionalArgs_NoArgs(t *testing.T) {
 		t.Fatalf("expected no error for empty args, got: %v", err)
 	}
 }
+
+func TestShortcutFlagIntArray(t *testing.T) {
+	f, _, _, _ := cmdutil.TestFactory(t, nil)
+	parent := &cobra.Command{Use: "root"}
+	var got []int
+	shortcut := Shortcut{
+		Service:     "slides",
+		Command:     "+screenshot",
+		Description: "capture screenshots",
+		Flags: []Flag{
+			{Name: "slide-number", Type: "int_array"},
+		},
+		Execute: func(ctx context.Context, runtime *RuntimeContext) error {
+			got = runtime.IntArray("slide-number")
+			return nil
+		},
+	}
+	shortcut.Mount(parent, f)
+	parent.SetArgs([]string{"+screenshot", "--as", "user", "--slide-number", "1", "--slide-number", "2,3"})
+	if err := parent.Execute(); err != nil {
+		t.Fatalf("Execute() error = %v", err)
+	}
+	if want := []int{1, 2, 3}; !reflect.DeepEqual(got, want) {
+		t.Fatalf("slide-number = %#v, want %#v", got, want)
+	}
+}

shortcuts/common/types.go

@@ -18,7 +18,7 @@ const (
 // Flag describes a CLI flag for a shortcut.
 type Flag struct {
 	Name     string // flag name (e.g. "calendar-id")
-	Type     string // "string" (default) | "bool" | "int" | "float64" | "string_array" | "string_slice"
+	Type     string // "string" (default) | "bool" | "int" | "float64" | "int_array" | "string_array" | "string_slice"
 	Default  string // default value as string
 	Desc     string // help text
 	Hidden   bool   // hidden from --help, still readable at runtime

shortcuts/contact/contact_search_user.go

@@ -85,6 +85,7 @@ type searchUserAPIData struct {
 	Items     []searchUserAPIItem `json:"items"`
 	HasMore   bool                `json:"has_more"`
 	PageToken string              `json:"page_token"`
+	Notice    string              `json:"notice"`
 }
 
 type searchUserAPIItem struct {
@@ -126,6 +127,7 @@ type searchUser struct {
 type searchUserResponse struct {
 	Users   []searchUser `json:"users"`
 	HasMore bool         `json:"has_more"`
+	Notice  string       `json:"notice,omitempty"`
 }
 
 var ContactSearchUser = common.Shortcut{
@@ -189,6 +191,7 @@ var ContactSearchUser = common.Shortcut{
 	Execute: executeSearchUser,
 }
 
+// executeSearchUser dispatches contact search to single-query or fanout mode.
 func executeSearchUser(ctx context.Context, runtime *common.RuntimeContext) error {
 	if strings.TrimSpace(runtime.Str("queries")) != "" {
 		return executeSearchUserFanout(ctx, runtime)
@@ -196,6 +199,7 @@ func executeSearchUser(ctx context.Context, runtime *common.RuntimeContext) erro
 	return executeSearchUserSingle(ctx, runtime)
 }
 
+// executeSearchUserSingle performs one contact search and preserves server notices.
 func executeSearchUserSingle(ctx context.Context, runtime *common.RuntimeContext) error {
 	body, err := buildSearchUserBody(runtime)
 	if err != nil {
@@ -222,7 +226,7 @@ func executeSearchUserSingle(ctx context.Context, runtime *common.RuntimeContext
 	}
 
 	users, hasMore := projectUsers(respData, runtime.Str("lang"), runtime.Config.Brand)
-	out := searchUserResponse{Users: users, HasMore: hasMore}
+	out := searchUserResponse{Users: users, HasMore: hasMore, Notice: respData.Notice}
 
 	runtime.OutFormat(out, &output.Meta{Count: len(users)}, func(w io.Writer) {
 		if len(users) == 0 {

shortcuts/contact/contact_search_user_fanout.go

@@ -45,22 +45,17 @@ type fanoutResult struct {
 	Query   string
 	Users   []searchUser
 	HasMore bool
+	Notice  string
 	ErrMsg  string // empty = success
 	Err     error  // original failure, kept for typed all-failed propagation
 }
 
-// isFanoutSummaryFormat gates the per-fanout stderr summary line. Includes csv
-// because that summary lives on stderr and never corrupts the csv stream on
-// stdout — single-query mode keeps the narrower isHumanReadableFormat predicate
-// for its refine hint, so adding csv here doesn't regress that path.
+// isFanoutSummaryFormat gates the per-fanout stderr summary line.
 func isFanoutSummaryFormat(format string) bool {
 	return format == "pretty" || format == "table" || format == "csv"
 }
 
-// runOneQuery converts every failure mode (transport, HTTP status, parse,
-// API code) into an ErrMsg string instead of returning a Go error. The
-// fanout dispatcher (Task 6) relies on this so a single failed query never
-// short-circuits the remaining workers.
+// runOneQuery converts one fanout request into either users or an error summary.
 func runOneQuery(ctx context.Context, runtime *common.RuntimeContext, index int, query string,
 	filter *searchUserAPIFilter) fanoutResult {
 	// Pre-check ctx so queued workers see cancellation before issuing a
@@ -94,9 +89,10 @@ func runOneQuery(ctx context.Context, runtime *common.RuntimeContext, index int,
 	}
 
 	users, hasMore := projectUsers(respData, runtime.Str("lang"), runtime.Config.Brand)
-	return fanoutResult{Index: index, Query: query, Users: users, HasMore: hasMore}
+	return fanoutResult{Index: index, Query: query, Users: users, HasMore: hasMore, Notice: respData.Notice}
 }
 
+// fanoutErrorResult records a failed fanout query without stopping other workers.
 func fanoutErrorResult(index int, query string, err error) fanoutResult {
 	if err == nil {
 		return fanoutResult{Index: index, Query: query}
@@ -113,17 +109,16 @@ type querySummary struct {
 	Query   string `json:"query"`
 	Error   string `json:"error,omitempty"`
 	HasMore bool   `json:"has_more"`
+	Notice  string `json:"notice,omitempty"`
 }
 
 type fanoutResponse struct {
 	Users   []fanoutUser   `json:"users"`
 	Queries []querySummary `json:"queries"`
+	Notice  string         `json:"notice,omitempty"`
 }
 
-// buildFanoutResponse walks results by Index (input order), flattens users[]
-// with matched_query, lists every input in queries[] (including successes),
-// and returns an error only when every query failed. The error wraps the
-// first failing query's ErrMsg so the CLI exits non-zero on full failure.
+// buildFanoutResponse flattens ordered fanout results and fails only when all queries fail.
 func buildFanoutResponse(queries []string, results []fanoutResult) (*fanoutResponse, error) {
 	indexed := make([]fanoutResult, len(queries))
 	for _, r := range results {
@@ -142,6 +137,7 @@ func buildFanoutResponse(queries []string, results []fanoutResult) (*fanoutRespo
 			Query:   queries[i],
 			Error:   r.ErrMsg,
 			HasMore: r.HasMore,
+			Notice:  r.Notice,
 		})
 		if r.ErrMsg != "" {
 			failed++
@@ -152,6 +148,9 @@ func buildFanoutResponse(queries []string, results []fanoutResult) (*fanoutRespo
 			}
 			continue
 		}
+		if out.Notice == "" {
+			out.Notice = r.Notice
+		}
 		for _, u := range r.Users {
 			out.Users = append(out.Users, fanoutUser{searchUser: u, MatchedQuery: queries[i]})
 		}

shortcuts/contact/contact_search_user_test.go

@@ -562,6 +562,7 @@ func mountAndRun(t *testing.T, s common.Shortcut, args []string, f *cmdutil.Fact
 	return parent.Execute()
 }
 
+// searchUserStub returns a representative user search response with a notice.
 func searchUserStub() *httpmock.Stub {
 	return &httpmock.Stub{
 		Method: "POST",
@@ -569,6 +570,7 @@ func searchUserStub() *httpmock.Stub {
 		Body: map[string]interface{}{
 			"code": 0, "msg": "ok",
 			"data": map[string]interface{}{
+				"notice": "The query is too long and has been truncated to the first 50 characters for search.",
 				"items": []interface{}{
 					map[string]interface{}{
 						"id": "ou_a",
@@ -590,6 +592,7 @@ func searchUserStub() *httpmock.Stub {
 	}
 }
 
+// TestSearchUser_Integration_PrettyRendersExpectedColumns verifies human output columns.
 func TestSearchUser_Integration_PrettyRendersExpectedColumns(t *testing.T) {
 	f, stdout, _, reg := cmdutil.TestFactory(t, searchUserDefaultConfig())
 	reg.Register(searchUserStub())
@@ -614,6 +617,7 @@ func TestSearchUser_Integration_PrettyRendersExpectedColumns(t *testing.T) {
 	}
 }
 
+// TestSearchUser_Integration_JSONStructuredFields verifies normalized JSON and notices.
 func TestSearchUser_Integration_JSONStructuredFields(t *testing.T) {
 	f, stdout, _, reg := cmdutil.TestFactory(t, searchUserDefaultConfig())
 	reg.Register(searchUserStub())
@@ -631,6 +635,9 @@ func TestSearchUser_Integration_JSONStructuredFields(t *testing.T) {
 	if !ok {
 		t.Fatalf("envelope.data: expected object, got %v\nraw=%s", got["data"], stdout.String())
 	}
+	if data["notice"] != "The query is too long and has been truncated to the first 50 characters for search." {
+		t.Fatalf("data.notice = %v", data["notice"])
+	}
 	users, _ := data["users"].([]interface{})
 	if len(users) != 1 {
 		t.Fatalf("users: expected 1, got %d (output=%s)", len(users), stdout.String())
@@ -1358,6 +1365,7 @@ func TestSearchUser_Integration_NoAutoPaginationFlags(t *testing.T) {
 	}
 }
 
+// TestFanout_FilterAppliedToEachQuery verifies shared fanout filters reach every request.
 func TestFanout_FilterAppliedToEachQuery(t *testing.T) {
 	f, stdout, _, reg := cmdutil.TestFactory(t, searchUserDefaultConfig())
 	stub := &httpmock.Stub{
@@ -1399,6 +1407,7 @@ func TestFanout_FilterAppliedToEachQuery(t *testing.T) {
 	}
 }
 
+// TestFanout_PartialFailure_ExitZero verifies partial fanout failures keep notices.
 func TestFanout_PartialFailure_ExitZero(t *testing.T) {
 	f, stdout, _, reg := cmdutil.TestFactory(t, searchUserDefaultConfig())
 	reg.Register(&httpmock.Stub{
@@ -1406,6 +1415,7 @@ func TestFanout_PartialFailure_ExitZero(t *testing.T) {
 		BodyFilter: func(b []byte) bool { return strings.Contains(string(b), `"alice"`) },
 		Body: map[string]interface{}{"code": 0, "msg": "ok",
 			"data": map[string]interface{}{
+				"notice":   "The query is too long and has been truncated to the first 50 characters for search.",
 				"items":    []interface{}{map[string]interface{}{"id": "ou_a"}},
 				"has_more": false,
 			}},
@@ -1432,10 +1442,17 @@ func TestFanout_PartialFailure_ExitZero(t *testing.T) {
 	if len(users) != 1 {
 		t.Errorf("users: expected 1 (alice), got %d; stdout=%s", len(users), stdout.String())
 	}
+	if data["notice"] != "The query is too long and has been truncated to the first 50 characters for search." {
+		t.Fatalf("data.notice = %v", data["notice"])
+	}
 	queries := data["queries"].([]interface{})
 	if len(queries) != 2 {
 		t.Fatalf("queries: expected 2, got %d", len(queries))
 	}
+	q0 := queries[0].(map[string]interface{})
+	if q0["notice"] != "The query is too long and has been truncated to the first 50 characters for search." {
+		t.Fatalf("queries[0].notice = %v", q0["notice"])
+	}
 	q1 := queries[1].(map[string]interface{})
 	if !strings.HasPrefix(q1["error"].(string), "HTTP 500") {
 		t.Errorf("queries[1].error: got %q", q1["error"])

shortcuts/doc/docs_search.go

@@ -74,6 +74,9 @@ var DocsSearch = common.Shortcut{
 			"page_token": data["page_token"],
 			"results":    normalizedItems,
 		}
+		if notice, _ := data["notice"].(string); notice != "" {
+			resultData["notice"] = notice
+		}
 
 		runtime.OutFormat(resultData, &output.Meta{Count: len(normalizedItems)}, func(w io.Writer) {
 			if len(normalizedItems) == 0 {

shortcuts/doc/docs_search_test.go

@@ -7,8 +7,48 @@ import (
 	"encoding/json"
 	"strings"
 	"testing"
+
+	"github.com/larksuite/cli/internal/cmdutil"
+	"github.com/larksuite/cli/internal/httpmock"
 )
 
+// TestDocsSearchExecutePassesThroughNotice verifies docs +search preserves notices.
+func TestDocsSearchExecutePassesThroughNotice(t *testing.T) {
+	const notice = "The query is too long and has been truncated to the first 50 characters for search."
+
+	f, stdout, _, reg := cmdutil.TestFactory(t, docsTestConfigWithAppID("docs-search-notice"))
+	reg.Register(&httpmock.Stub{
+		Method: "POST",
+		URL:    "/open-apis/search/v2/doc_wiki/search",
+		Body: map[string]interface{}{
+			"code": 0,
+			"msg":  "ok",
+			"data": map[string]interface{}{
+				"notice":     notice,
+				"res_units":  []interface{}{},
+				"total":      0,
+				"has_more":   false,
+				"page_token": "",
+			},
+		},
+	})
+
+	if err := mountAndRunDocs(t, DocsSearch, []string{"+search", "--query", "incident", "--format", "json", "--as", "user"}, f, stdout); err != nil {
+		t.Fatalf("DocsSearch.Execute() error = %v", err)
+	}
+	reg.Verify(t)
+
+	var env map[string]interface{}
+	if err := json.Unmarshal(stdout.Bytes(), &env); err != nil {
+		t.Fatalf("json.Unmarshal(stdout) error = %v\nstdout=%s", err, stdout.String())
+	}
+	data, _ := env["data"].(map[string]interface{})
+	if got, _ := data["notice"].(string); got != notice {
+		t.Fatalf("data.notice = %q, want %q; data=%#v", got, notice, data)
+	}
+}
+
+// TestAddIsoTimeFieldsSupportsJSONNumber verifies JSON numbers get ISO fields.
 func TestAddIsoTimeFieldsSupportsJSONNumber(t *testing.T) {
 	t.Parallel()
 

shortcuts/drive/drive_add_comment.go

@@ -121,7 +121,7 @@ const (
 var DriveAddComment = common.Shortcut{
 	Service:     "drive",
 	Command:     "+add-comment",
-	Description: "Add a comment to doc/docx/file/sheet/slides; file targets support selected extensions and full comments only",
+	Description: "Add a comment to doc/docx/file/sheet/slides/base(bitable); file targets support selected extensions and full comments only",
 	Risk:        "write",
 	Scopes: []string{
 		"drive:drive.metadata:readonly",
@@ -131,12 +131,12 @@ var DriveAddComment = common.Shortcut{
 	},
 	AuthTypes: []string{"user", "bot"},
 	Flags: []common.Flag{
-		{Name: "doc", Desc: "document URL/token, file URL/token, sheet/slides URL, or wiki URL that resolves to doc/docx/file/sheet/slides", Required: true},
-		{Name: "type", Desc: "document type: doc, docx, file, sheet, slides (required when --doc is a bare token; auto-detected for URLs)", Enum: []string{"doc", "docx", "file", "sheet", "slides"}},
+		{Name: "doc", Desc: "document URL/token, file URL/token, sheet/slides/base/bitable URL, or wiki URL that resolves to doc/docx/file/sheet/slides/base(bitable)", Required: true},
+		{Name: "type", Desc: "document type: doc, docx, file, sheet, slides, bitable, base (required when --doc is a bare token; auto-detected for URLs; use bitable as the wire value, base is accepted as a compatibility alias)", Enum: []string{"doc", "docx", "file", "sheet", "slides", "bitable", "base"}},
 		{Name: "content", Desc: "reply_elements JSON string", Required: true, Input: []string{common.File, common.Stdin}},
 		{Name: "full-comment", Type: "bool", Desc: "create a full-document comment; also the default when no location is provided"},
 		{Name: "selection-with-ellipsis", Desc: "target content locator (plain text or 'start...end')"},
-		{Name: "block-id", Desc: "for docx: anchor block ID; for sheet: <sheetId>!<cell> (e.g. a281f9!D6); for slides: <slide-block-type>!<xml-id> (e.g. shape!bPq)"},
+		{Name: "block-id", Desc: "for docx: anchor block ID; for sheet: <sheetId>!<cell>; for slides: <slide-block-type>!<xml-id>; for base(bitable): <table-id>!<record-id>!<view-id>"},
 	},
 	Validate: func(ctx context.Context, runtime *common.RuntimeContext) error {
 		docRef, err := parseCommentDocRef(runtime.Str("doc"), runtime.Str("type"))
@@ -148,6 +148,17 @@ var DriveAddComment = common.Shortcut{
 			return err
 		}
 
+		if docRef.Kind == "base" {
+			if runtime.Bool("full-comment") {
+				return errs.NewValidationError(errs.SubtypeInvalidArgument, "--full-comment is not applicable for base(bitable) comments; use --block-id <table-id>!<record-id>!<view-id>").WithParam("--full-comment")
+			}
+			if strings.TrimSpace(runtime.Str("selection-with-ellipsis")) != "" {
+				return errs.NewValidationError(errs.SubtypeInvalidArgument, "--selection-with-ellipsis is not applicable for base(bitable) comments; use --block-id <table-id>!<record-id>!<view-id>").WithParam("--selection-with-ellipsis")
+			}
+			_, err := parseBaseCommentAnchor(runtime)
+			return err
+		}
+
 		// Sheet comment validation.
 		if docRef.Kind == "sheet" {
 			blockID := strings.TrimSpace(runtime.Str("block-id"))
@@ -188,7 +199,7 @@ var DriveAddComment = common.Shortcut{
 			return validateFileCommentMode(mode, "")
 		}
 		if mode == commentModeLocal && docRef.Kind == "doc" {
-			return errs.NewValidationError(errs.SubtypeInvalidArgument, "local comments only support docx, sheet, and slides; old doc format only supports full comments")
+			return errs.NewValidationError(errs.SubtypeInvalidArgument, "local comments only support docx, sheet, slides, and base(bitable); old doc format only supports full comments")
 		}
 
 		return nil
@@ -215,6 +226,23 @@ var DriveAddComment = common.Shortcut{
 			resolvedToken = target.FileToken
 		}
 
+		if resolvedKind == "base" {
+			anchor, err := parseBaseCommentAnchor(runtime)
+			if err != nil {
+				return common.NewDryRunAPI().Set("error", err.Error())
+			}
+			commentBody := buildBaseCommentCreateV2Request(replyElements, anchor)
+			desc := "1-step request: create base(bitable) record-local comment"
+			if isWiki {
+				desc = "2-step orchestration: resolve wiki -> create base(bitable) record-local comment"
+			}
+			return common.NewDryRunAPI().
+				Desc(desc).
+				POST("/open-apis/drive/v1/files/:file_token/new_comments").
+				Body(commentBody).
+				Set("file_token", resolvedToken)
+		}
+
 		// Sheet comment dry-run.
 		if resolvedKind == "sheet" {
 			anchor, _ := parseSheetCellRef(blockID)
@@ -352,6 +380,14 @@ var DriveAddComment = common.Shortcut{
 	Execute: func(ctx context.Context, runtime *common.RuntimeContext) error {
 		// Sheet comment: direct URL or token fast path.
 		docRef, _ := parseCommentDocRef(runtime.Str("doc"), runtime.Str("type"))
+		if docRef.Kind == "base" {
+			return executeBaseComment(runtime, resolvedCommentTarget{
+				DocID:      docRef.Token,
+				FileToken:  docRef.Token,
+				FileType:   "base",
+				ResolvedBy: "base",
+			})
+		}
 		if docRef.Kind == "sheet" {
 			return executeSheetComment(runtime, docRef)
 		}
@@ -375,6 +411,9 @@ var DriveAddComment = common.Shortcut{
 		if target.FileType == "slides" {
 			return executeSlidesComment(runtime, commentDocRef{Kind: "slides", Token: target.FileToken})
 		}
+		if target.FileType == "base" {
+			return executeBaseComment(runtime, target)
+		}
 		if target.FileType == "file" {
 			return executeFileComment(runtime, target)
 		}
@@ -482,6 +521,12 @@ func parseCommentDocRef(input, docType string) (commentDocRef, error) {
 	if token, ok := extractURLToken(raw, "/sheets/"); ok {
 		return commentDocRef{Kind: "sheet", Token: token}, nil
 	}
+	if token, ok := extractURLToken(raw, "/base/"); ok {
+		return commentDocRef{Kind: "base", Token: token}, nil
+	}
+	if token, ok := extractURLToken(raw, "/bitable/"); ok {
+		return commentDocRef{Kind: "base", Token: token}, nil
+	}
 	if token, ok := extractURLToken(raw, "/file/"); ok {
 		return commentDocRef{Kind: "file", Token: token}, nil
 	}
@@ -495,7 +540,7 @@ func parseCommentDocRef(input, docType string) (commentDocRef, error) {
 		return commentDocRef{Kind: "doc", Token: token}, nil
 	}
 	if strings.Contains(raw, "://") {
-		return commentDocRef{}, errs.NewValidationError(errs.SubtypeInvalidArgument, "unsupported --doc input %q: use a doc/docx/file/sheet/slides URL, a token with --type, or a wiki URL that resolves to doc/docx/file/sheet/slides", raw).WithParam("--doc")
+		return commentDocRef{}, errs.NewValidationError(errs.SubtypeInvalidArgument, "unsupported --doc input %q: use a doc/docx/file/sheet/slides/base/bitable URL, a token with --type, or a wiki URL that resolves to doc/docx/file/sheet/slides/base(bitable)", raw).WithParam("--doc")
 	}
 	if strings.ContainsAny(raw, "/?#") {
 		return commentDocRef{}, errs.NewValidationError(errs.SubtypeInvalidArgument, "unsupported --doc input %q: use a token with --type, or a wiki URL", raw).WithParam("--doc")
@@ -504,7 +549,10 @@ func parseCommentDocRef(input, docType string) (commentDocRef, error) {
 	// Bare token: --type is required.
 	docType = strings.TrimSpace(docType)
 	if docType == "" {
-		return commentDocRef{}, errs.NewValidationError(errs.SubtypeInvalidArgument, "--type is required when --doc is a bare token (allowed values: doc, docx, file, sheet, slides)").WithParam("--type")
+		return commentDocRef{}, errs.NewValidationError(errs.SubtypeInvalidArgument, "--type is required when --doc is a bare token (allowed values: doc, docx, file, sheet, slides, bitable, base; use bitable as the wire value, base is accepted as a compatibility alias)").WithParam("--type")
+	}
+	if docType == "bitable" || docType == "base" {
+		return commentDocRef{Kind: "base", Token: raw}, nil
 	}
 	return commentDocRef{Kind: docType, Token: raw}, nil
 }
@@ -515,11 +563,11 @@ func resolveCommentTarget(ctx context.Context, runtime *common.RuntimeContext, i
 		return resolvedCommentTarget{}, err
 	}
 
-	if docRef.Kind == "docx" || docRef.Kind == "doc" || docRef.Kind == "file" || docRef.Kind == "sheet" || docRef.Kind == "slides" {
+	if docRef.Kind == "docx" || docRef.Kind == "doc" || docRef.Kind == "file" || docRef.Kind == "sheet" || docRef.Kind == "slides" || docRef.Kind == "base" {
 		if mode == commentModeLocal {
 			switch docRef.Kind {
 			case "doc":
-				return resolvedCommentTarget{}, errs.NewValidationError(errs.SubtypeInvalidArgument, "local comments only support docx, sheet, and slides; old doc format only supports full comments")
+				return resolvedCommentTarget{}, errs.NewValidationError(errs.SubtypeInvalidArgument, "local comments only support docx, sheet, slides, and base(bitable); old doc format only supports full comments")
 			case "file":
 				if err := validateFileCommentMode(mode, ""); err != nil {
 					return resolvedCommentTarget{}, err
@@ -557,6 +605,22 @@ func resolveCommentTarget(ctx context.Context, runtime *common.RuntimeContext, i
 	if objType == "slides" && strings.TrimSpace(runtime.Str("selection-with-ellipsis")) != "" {
 		return resolvedCommentTarget{}, errs.NewValidationError(errs.SubtypeInvalidArgument, "wiki resolved to %q, but --selection-with-ellipsis is not applicable for slide comments; use --block-id <slide-block-type>!<xml-id>", objType)
 	}
+	if objType == "bitable" || objType == "base" {
+		if runtime.Bool("full-comment") {
+			return resolvedCommentTarget{}, errs.NewValidationError(errs.SubtypeInvalidArgument, "wiki resolved to %q, but --full-comment is not applicable for base(bitable) comments; use --block-id <table-id>!<record-id>!<view-id>", objType).WithParam("--full-comment")
+		}
+		if strings.TrimSpace(runtime.Str("selection-with-ellipsis")) != "" {
+			return resolvedCommentTarget{}, errs.NewValidationError(errs.SubtypeInvalidArgument, "wiki resolved to %q, but --selection-with-ellipsis is not applicable for base(bitable) comments; use --block-id <table-id>!<record-id>!<view-id>", objType).WithParam("--selection-with-ellipsis")
+		}
+		fmt.Fprintf(runtime.IO().ErrOut, "Resolved wiki to base: %s\n", common.MaskToken(objToken))
+		return resolvedCommentTarget{
+			DocID:      objToken,
+			FileToken:  objToken,
+			FileType:   "base",
+			ResolvedBy: "wiki",
+			WikiToken:  docRef.Token,
+		}, nil
+	}
 	if objType == "sheet" {
 		// Sheet comments are handled via the sheet fast path in Execute.
 		fmt.Fprintf(runtime.IO().ErrOut, "Resolved wiki to %s: %s\n", objType, common.MaskToken(objToken))
@@ -592,10 +656,10 @@ func resolveCommentTarget(ctx context.Context, runtime *common.RuntimeContext, i
 		}, nil
 	}
 	if mode == commentModeLocal && objType != "docx" {
-		return resolvedCommentTarget{}, errs.NewValidationError(errs.SubtypeInvalidArgument, "wiki resolved to %q, but local comments only support docx, sheet, and slides; for sheet use --block-id <sheetId>!<cell>, for slides use --block-id <slide-block-type>!<xml-id>", objType)
+		return resolvedCommentTarget{}, errs.NewValidationError(errs.SubtypeInvalidArgument, "wiki resolved to %q, but local comments only support docx, sheet, slides, and base(bitable); for sheet use --block-id <sheetId>!<cell>, for slides use --block-id <slide-block-type>!<xml-id>, for base use --block-id <table-id>!<record-id>!<view-id>", objType)
 	}
 	if mode == commentModeFull && objType != "docx" && objType != "doc" {
-		return resolvedCommentTarget{}, errs.NewValidationError(errs.SubtypeInvalidArgument, "wiki resolved to %q, but comments only support doc/docx/file/sheet/slides", objType)
+		return resolvedCommentTarget{}, errs.NewValidationError(errs.SubtypeInvalidArgument, "wiki resolved to %q, but comments only support doc/docx/file/sheet/slides/base(bitable)", objType)
 	}
 
 	fmt.Fprintf(runtime.IO().ErrOut, "Resolved wiki to %s: %s\n", objType, common.MaskToken(objToken))
@@ -787,6 +851,12 @@ type sheetAnchor struct {
 	Row     int
 }
 
+type baseAnchor struct {
+	BlockID      string
+	BaseRecordID string
+	BaseViewID   string
+}
+
 func buildCommentCreateV2Request(fileType, blockID, slideBlockType string, replyElements []map[string]interface{}, sheet *sheetAnchor) map[string]interface{} {
 	body := map[string]interface{}{
 		"file_type":      fileType,
@@ -813,6 +883,18 @@ func buildCommentCreateV2Request(fileType, blockID, slideBlockType string, reply
 	return body
 }
 
+func buildBaseCommentCreateV2Request(replyElements []map[string]interface{}, anchor baseAnchor) map[string]interface{} {
+	return map[string]interface{}{
+		"file_type":      "bitable",
+		"reply_elements": replyElements,
+		"anchor": map[string]interface{}{
+			"block_id":       anchor.BlockID,
+			"base_record_id": anchor.BaseRecordID,
+			"base_view_id":   anchor.BaseViewID,
+		},
+	}
+}
+
 func anchorBlockIDForDryRun(blockID string) string {
 	if strings.TrimSpace(blockID) != "" {
 		return strings.TrimSpace(blockID)
@@ -820,6 +902,26 @@ func anchorBlockIDForDryRun(blockID string) string {
 	return "<anchor_block_id>"
 }
 
+func parseBaseCommentAnchor(runtime *common.RuntimeContext) (baseAnchor, error) {
+	blockID := strings.TrimSpace(runtime.Str("block-id"))
+	if blockID == "" {
+		return baseAnchor{}, errs.NewValidationError(errs.SubtypeInvalidArgument, "--block-id is required for base(bitable) record-local comments (format: <table-id>!<record-id>!<view-id>, e.g. tbl9mp6fj9kDKHQV!recBIBgGmb!vewc46MG1R)").WithParam("--block-id")
+	}
+	return parseBaseBlockRef(blockID)
+}
+
+func parseBaseBlockRef(blockID string) (baseAnchor, error) {
+	parts := strings.Split(strings.TrimSpace(blockID), "!")
+	if len(parts) != 3 || strings.TrimSpace(parts[0]) == "" || strings.TrimSpace(parts[1]) == "" || strings.TrimSpace(parts[2]) == "" {
+		return baseAnchor{}, errs.NewValidationError(errs.SubtypeInvalidArgument, "base(bitable) record-local comments require --block-id in <table-id>!<record-id>!<view-id> format, e.g. tbl9mp6fj9kDKHQV!recBIBgGmb!vewc46MG1R").WithParam("--block-id")
+	}
+	return baseAnchor{
+		BlockID:      strings.TrimSpace(parts[0]),
+		BaseRecordID: strings.TrimSpace(parts[1]),
+		BaseViewID:   strings.TrimSpace(parts[2]),
+	}, nil
+}
+
 func parseSlidesBlockRef(blockID string) (string, string, error) {
 	blockID = strings.TrimSpace(blockID)
 	if blockID == "" {
@@ -1030,6 +1132,53 @@ func executeSheetComment(runtime *common.RuntimeContext, docRef commentDocRef) e
 	return nil
 }
 
+func executeBaseComment(runtime *common.RuntimeContext, target resolvedCommentTarget) error {
+	replyElements, err := parseCommentReplyElements(runtime.Str("content"))
+	if err != nil {
+		return err
+	}
+	anchor, err := parseBaseCommentAnchor(runtime)
+	if err != nil {
+		return err
+	}
+
+	requestPath := fmt.Sprintf("/open-apis/drive/v1/files/%s/new_comments", validate.EncodePathSegment(target.FileToken))
+	requestBody := buildBaseCommentCreateV2Request(replyElements, anchor)
+
+	fmt.Fprintf(runtime.IO().ErrOut, "Creating base(bitable) record-local comment in %s (table=%s, record=%s, view=%s)\n",
+		common.MaskToken(target.FileToken), anchor.BlockID, anchor.BaseRecordID, anchor.BaseViewID)
+
+	data, err := runtime.CallAPITyped("POST", requestPath, nil, requestBody)
+	if err != nil {
+		return err
+	}
+
+	out := map[string]interface{}{
+		"file_token":     target.FileToken,
+		"file_type":      "bitable",
+		"resolved_by":    target.ResolvedBy,
+		"comment_mode":   "base_record",
+		"base_block_id":  anchor.BlockID,
+		"base_record_id": anchor.BaseRecordID,
+		"base_view_id":   anchor.BaseViewID,
+	}
+	if commentID := data["comment_id"]; commentID != nil {
+		out["comment_id"] = commentID
+	}
+	if replyID := data["reply_id"]; replyID != nil {
+		out["reply_id"] = replyID
+	}
+	if createdAt := firstPresentValue(data, "created_at", "create_time"); createdAt != nil {
+		out["created_at"] = createdAt
+	}
+	if target.WikiToken != "" {
+		out["wiki_token"] = target.WikiToken
+	}
+
+	runtime.Out(out, nil)
+	return nil
+}
+
 func executeFileComment(runtime *common.RuntimeContext, target resolvedCommentTarget) error {
 	replyElements, err := parseCommentReplyElements(runtime.Str("content"))
 	if err != nil {

shortcuts/drive/drive_add_comment_test.go

@@ -133,6 +133,20 @@ func TestParseCommentDocRef(t *testing.T) {
 			wantKind:  "file",
 			wantToken: "fileToken",
 		},
+		{
+			name:      "raw token with type bitable",
+			input:     "baseToken",
+			docType:   "bitable",
+			wantKind:  "base",
+			wantToken: "baseToken",
+		},
+		{
+			name:      "raw token with type base alias",
+			input:     "baseToken",
+			docType:   "base",
+			wantKind:  "base",
+			wantToken: "baseToken",
+		},
 		{
 			name:    "raw token without type",
 			input:   "xxxxxx",
@@ -156,6 +170,18 @@ func TestParseCommentDocRef(t *testing.T) {
 			wantKind:  "file",
 			wantToken: "boxcn123",
 		},
+		{
+			name:      "base url",
+			input:     "https://example.larksuite.com/base/baseToken123?table=tbl1",
+			wantKind:  "base",
+			wantToken: "baseToken123",
+		},
+		{
+			name:      "bitable url",
+			input:     "https://example.larksuite.com/bitable/baseToken456?table=tbl1",
+			wantKind:  "base",
+			wantToken: "baseToken456",
+		},
 		{
 			name:    "unsupported url",
 			input:   "https://example.com/not-a-doc",
@@ -726,6 +752,35 @@ func TestBuildCommentCreateV2RequestSheetOverridesBlockID(t *testing.T) {
 	}
 }
 
+func TestBuildBaseCommentCreateV2Request(t *testing.T) {
+	t.Parallel()
+	replyElements := []map[string]interface{}{
+		{"type": "text", "text": "base comment"},
+	}
+	got := buildBaseCommentCreateV2Request(replyElements, baseAnchor{
+		BlockID:      "tbl9mp6fj9kDKHQV",
+		BaseRecordID: "recBIBgGmb",
+		BaseViewID:   "vewc46MG1R",
+	})
+
+	if got["file_type"] != "bitable" {
+		t.Fatalf("expected file_type bitable, got %#v", got["file_type"])
+	}
+	anchor, ok := got["anchor"].(map[string]interface{})
+	if !ok {
+		t.Fatalf("expected anchor map, got %#v", got["anchor"])
+	}
+	if anchor["block_id"] != "tbl9mp6fj9kDKHQV" {
+		t.Fatalf("expected block_id tbl9mp6fj9kDKHQV, got %#v", anchor["block_id"])
+	}
+	if anchor["base_record_id"] != "recBIBgGmb" {
+		t.Fatalf("expected base_record_id recBIBgGmb, got %#v", anchor["base_record_id"])
+	}
+	if anchor["base_view_id"] != "vewc46MG1R" {
+		t.Fatalf("expected base_view_id vewc46MG1R, got %#v", anchor["base_view_id"])
+	}
+}
+
 // ── Sheet cell ref parsing tests ────────────────────────────────────────────
 
 func TestParseSheetCellRef(t *testing.T) {
@@ -985,6 +1040,78 @@ func TestFileCommentValidateRejectsSelectionWithEllipsis(t *testing.T) {
 	}
 }
 
+func TestBaseCommentValidateMissingBlockID(t *testing.T) {
+	f, stdout, _, _ := cmdutil.TestFactory(t, driveTestConfig())
+	err := mountAndRunDrive(t, DriveAddComment, []string{
+		"+add-comment",
+		"--doc", "https://example.larksuite.com/base/baseToken",
+		"--content", `[{"type":"text","text":"test"}]`,
+		"--as", "user",
+	}, f, stdout)
+	if err == nil || !strings.Contains(err.Error(), "--block-id is required") {
+		t.Fatalf("expected block-id required error, got: %v", err)
+	}
+}
+
+func TestBaseCommentValidateMalformedBlockID(t *testing.T) {
+	cases := []string{
+		"tbl9mp6fj9kDKHQV",
+		"tbl9mp6fj9kDKHQV!recBIBgGmb",
+		"tbl9mp6fj9kDKHQV!!vewc46MG1R",
+	}
+	for _, blockID := range cases {
+		t.Run(blockID, func(t *testing.T) {
+			f, stdout, _, _ := cmdutil.TestFactory(t, driveTestConfig())
+			err := mountAndRunDrive(t, DriveAddComment, []string{
+				"+add-comment",
+				"--doc", "https://example.larksuite.com/base/baseToken",
+				"--content", `[{"type":"text","text":"test"}]`,
+				"--block-id", blockID,
+				"--as", "user",
+			}, f, stdout)
+			if err == nil || !strings.Contains(err.Error(), "<table-id>!<record-id>!<view-id>") {
+				t.Fatalf("expected block-id format error, got: %v", err)
+			}
+		})
+	}
+}
+
+func TestBaseCommentValidateRejectsIncompatibleFlags(t *testing.T) {
+	cases := []struct {
+		name    string
+		args    []string
+		wantErr string
+	}{
+		{
+			name:    "full comment",
+			args:    []string{"--full-comment"},
+			wantErr: "--full-comment is not applicable for base(bitable) comments",
+		},
+		{
+			name:    "selection",
+			args:    []string{"--selection-with-ellipsis", "some text"},
+			wantErr: "--selection-with-ellipsis is not applicable for base(bitable) comments",
+		},
+	}
+	for _, tc := range cases {
+		t.Run(tc.name, func(t *testing.T) {
+			f, stdout, _, _ := cmdutil.TestFactory(t, driveTestConfig())
+			args := []string{
+				"+add-comment",
+				"--doc", "https://example.larksuite.com/base/baseToken",
+				"--content", `[{"type":"text","text":"test"}]`,
+				"--block-id", "tbl9mp6fj9kDKHQV!recBIBgGmb!vewc46MG1R",
+				"--as", "user",
+			}
+			args = append(args, tc.args...)
+			err := mountAndRunDrive(t, DriveAddComment, args, f, stdout)
+			if err == nil || !strings.Contains(err.Error(), tc.wantErr) {
+				t.Fatalf("expected %q error, got: %v", tc.wantErr, err)
+			}
+		})
+	}
+}
+
 // ── Slides comment execute tests ────────────────────────────────────────────
 
 func TestSlidesCommentExecuteSuccess(t *testing.T) {
@@ -1195,6 +1322,87 @@ func TestSheetCommentViaWikiMissingBlockID(t *testing.T) {
 	}
 }
 
+func TestBaseCommentExecuteSuccess(t *testing.T) {
+	f, stdout, _, reg := cmdutil.TestFactory(t, driveTestConfig())
+	createStub := &httpmock.Stub{
+		Method: "POST", URL: "/open-apis/drive/v1/files/baseToken/new_comments",
+		Body: map[string]interface{}{
+			"code": 0, "msg": "success",
+			"data": map[string]interface{}{
+				"comment_id": "baseComment123",
+				"reply_id":   "baseReply123",
+				"created_at": 1700000000,
+			},
+		},
+	}
+	reg.Register(createStub)
+	err := mountAndRunDrive(t, DriveAddComment, []string{
+		"+add-comment",
+		"--doc", "https://example.larksuite.com/base/baseToken",
+		"--content", `[{"type":"text","text":"请看这条记录"}]`,
+		"--block-id", "tbl9mp6fj9kDKHQV!recBIBgGmb!vewc46MG1R",
+		"--as", "user",
+	}, f, stdout)
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+
+	var requestBody map[string]interface{}
+	if err := json.Unmarshal(createStub.CapturedBody, &requestBody); err != nil {
+		t.Fatalf("failed to decode captured body: %v\nbody:\n%s", err, string(createStub.CapturedBody))
+	}
+	if got := mustStringField(t, requestBody, "file_type", "request.file_type"); got != "bitable" {
+		t.Fatalf("request file_type = %q, want bitable", got)
+	}
+	anchor := mustMapValue(t, requestBody["anchor"], "request.anchor")
+	if got := mustStringField(t, anchor, "block_id", "request.anchor.block_id"); got != "tbl9mp6fj9kDKHQV" {
+		t.Fatalf("request block_id = %q, want tbl9mp6fj9kDKHQV", got)
+	}
+	if got := mustStringField(t, anchor, "base_record_id", "request.anchor.base_record_id"); got != "recBIBgGmb" {
+		t.Fatalf("request base_record_id = %q, want recBIBgGmb", got)
+	}
+	if got := mustStringField(t, anchor, "base_view_id", "request.anchor.base_view_id"); got != "vewc46MG1R" {
+		t.Fatalf("request base_view_id = %q, want vewc46MG1R", got)
+	}
+
+	out := decodeJSONMap(t, stdout.String())
+	data := mustMapValue(t, out["data"], "data")
+	if got := mustStringField(t, data, "file_type", "data.file_type"); got != "bitable" {
+		t.Fatalf("stdout file_type = %q, want bitable\nstdout:\n%s", got, stdout.String())
+	}
+	if got := mustStringField(t, data, "comment_mode", "data.comment_mode"); got != "base_record" {
+		t.Fatalf("stdout comment_mode = %q, want base_record\nstdout:\n%s", got, stdout.String())
+	}
+	if got := mustStringField(t, data, "reply_id", "data.reply_id"); got != "baseReply123" {
+		t.Fatalf("stdout reply_id = %q, want baseReply123\nstdout:\n%s", got, stdout.String())
+	}
+}
+
+func TestBaseCommentExecuteBareToken(t *testing.T) {
+	f, stdout, _, reg := cmdutil.TestFactory(t, driveTestConfig())
+	reg.Register(&httpmock.Stub{
+		Method: "POST", URL: "/open-apis/drive/v1/files/baseBareToken/new_comments",
+		Body: map[string]interface{}{
+			"code": 0, "msg": "success",
+			"data": map[string]interface{}{"comment_id": "baseBareComment"},
+		},
+	})
+	err := mountAndRunDrive(t, DriveAddComment, []string{
+		"+add-comment",
+		"--doc", "baseBareToken",
+		"--type", "bitable",
+		"--content", `[{"type":"text","text":"ok"}]`,
+		"--block-id", "tbl9mp6fj9kDKHQV!recBIBgGmb!vewc46MG1R",
+		"--as", "user",
+	}, f, stdout)
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	if !strings.Contains(stdout.String(), "baseBareComment") {
+		t.Fatalf("stdout missing comment_id: %s", stdout.String())
+	}
+}
+
 func TestFileCommentExecuteSuccess(t *testing.T) {
 	f, stdout, _, reg := cmdutil.TestFactory(t, driveTestConfig())
 	reg.Register(&httpmock.Stub{
@@ -1433,6 +1641,40 @@ func TestDryRunSlidesDirectURL(t *testing.T) {
 	}
 }
 
+func TestDryRunBaseDirectURL(t *testing.T) {
+	f, stdout, _, _ := cmdutil.TestFactory(t, driveTestConfig())
+	err := mountAndRunDrive(t, DriveAddComment, []string{
+		"+add-comment",
+		"--doc", "https://example.larksuite.com/base/baseToken",
+		"--content", `[{"type":"text","text":"test"}]`,
+		"--block-id", "tbl9mp6fj9kDKHQV!recBIBgGmb!vewc46MG1R",
+		"--dry-run", "--as", "user",
+	}, f, stdout)
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	if !strings.Contains(stdout.String(), "record-local comment") {
+		t.Fatalf("dry-run output missing record-local comment: %s", stdout.String())
+	}
+	out := decodeJSONMap(t, stdout.String())
+	api := mustSliceValue(t, out["api"], "api")
+	call := mustMapValue(t, api[0], "api[0]")
+	body := mustMapValue(t, call["body"], "api[0].body")
+	anchor := mustMapValue(t, body["anchor"], "api[0].body.anchor")
+	if got := mustStringField(t, body, "file_type", "api[0].body.file_type"); got != "bitable" {
+		t.Fatalf("dry-run body.file_type = %q, want bitable\nstdout:\n%s", got, stdout.String())
+	}
+	if got := mustStringField(t, anchor, "block_id", "api[0].body.anchor.block_id"); got != "tbl9mp6fj9kDKHQV" {
+		t.Fatalf("dry-run body.anchor.block_id = %q, want tbl9mp6fj9kDKHQV\nstdout:\n%s", got, stdout.String())
+	}
+	if got := mustStringField(t, anchor, "base_record_id", "api[0].body.anchor.base_record_id"); got != "recBIBgGmb" {
+		t.Fatalf("dry-run body.anchor.base_record_id = %q, want recBIBgGmb\nstdout:\n%s", got, stdout.String())
+	}
+	if got := mustStringField(t, anchor, "base_view_id", "api[0].body.anchor.base_view_id"); got != "vewc46MG1R" {
+		t.Fatalf("dry-run body.anchor.base_view_id = %q, want vewc46MG1R\nstdout:\n%s", got, stdout.String())
+	}
+}
+
 func TestDryRunWikiResolvesToSlides(t *testing.T) {
 	f, stdout, _, reg := cmdutil.TestFactory(t, driveTestConfig())
 	reg.Register(&httpmock.Stub{
@@ -1636,25 +1878,92 @@ func TestResolveWikiToDocxFullComment(t *testing.T) {
 	}
 }
 
-func TestResolveWikiToUnsupportedType(t *testing.T) {
-	f, stdout, _, reg := cmdutil.TestFactory(t, driveTestConfig())
-	reg.Register(&httpmock.Stub{
-		Method: "GET", URL: "/open-apis/wiki/v2/spaces/get_node",
-		Body: map[string]interface{}{
-			"code": 0, "msg": "success",
-			"data": map[string]interface{}{
-				"node": map[string]interface{}{"obj_type": "bitable", "obj_token": "bitToken"},
-			},
+func TestResolveWikiToBaseComment(t *testing.T) {
+	for _, objType := range []string{"bitable", "base"} {
+		t.Run(objType, func(t *testing.T) {
+			f, stdout, _, reg := cmdutil.TestFactory(t, driveTestConfig())
+			reg.Register(&httpmock.Stub{
+				Method: "GET", URL: "/open-apis/wiki/v2/spaces/get_node",
+				Body: map[string]interface{}{
+					"code": 0, "msg": "success",
+					"data": map[string]interface{}{
+						"node": map[string]interface{}{"obj_type": objType, "obj_token": "bitToken"},
+					},
+				},
+			})
+			reg.Register(&httpmock.Stub{
+				Method: "POST", URL: "/open-apis/drive/v1/files/bitToken/new_comments",
+				Body: map[string]interface{}{
+					"code": 0, "msg": "success",
+					"data": map[string]interface{}{"comment_id": "wikiBaseComment", "reply_id": "wikiBaseReply"},
+				},
+			})
+			err := mountAndRunDrive(t, DriveAddComment, []string{
+				"+add-comment",
+				"--doc", "https://example.larksuite.com/wiki/wikiToken",
+				"--content", `[{"type":"text","text":"test"}]`,
+				"--block-id", "tbl9mp6fj9kDKHQV!recBIBgGmb!vewc46MG1R",
+				"--as", "user",
+			}, f, stdout)
+			if err != nil {
+				t.Fatalf("unexpected error: %v", err)
+			}
+			if !strings.Contains(stdout.String(), "wikiBaseComment") {
+				t.Fatalf("stdout missing comment_id: %s", stdout.String())
+			}
+			out := decodeJSONMap(t, stdout.String())
+			data := mustMapValue(t, out["data"], "data")
+			if got := mustStringField(t, data, "file_type", "data.file_type"); got != "bitable" {
+				t.Fatalf("stdout file_type = %q, want bitable\nstdout:\n%s", got, stdout.String())
+			}
+			if got := mustStringField(t, data, "wiki_token", "data.wiki_token"); got != "wikiToken" {
+				t.Fatalf("stdout wiki_token = %q, want wikiToken\nstdout:\n%s", got, stdout.String())
+			}
+		})
+	}
+}
+
+func TestResolveWikiToBaseRejectsIncompatibleFlags(t *testing.T) {
+	cases := []struct {
+		name    string
+		args    []string
+		wantErr string
+	}{
+		{
+			name:    "full comment",
+			args:    []string{"--full-comment"},
+			wantErr: "--full-comment is not applicable for base(bitable) comments",
 		},
-	})
-	err := mountAndRunDrive(t, DriveAddComment, []string{
-		"+add-comment",
-		"--doc", "https://example.larksuite.com/wiki/wikiToken",
-		"--content", `[{"type":"text","text":"test"}]`,
-		"--as", "user",
-	}, f, stdout)
-	if err == nil || !strings.Contains(err.Error(), "only support doc/docx/file/sheet/slides") {
-		t.Fatalf("expected unsupported type error, got: %v", err)
+		{
+			name:    "selection",
+			args:    []string{"--selection-with-ellipsis", "some text"},
+			wantErr: "--selection-with-ellipsis is not applicable for base(bitable) comments",
+		},
+	}
+	for _, tc := range cases {
+		t.Run(tc.name, func(t *testing.T) {
+			f, stdout, _, reg := cmdutil.TestFactory(t, driveTestConfig())
+			reg.Register(&httpmock.Stub{
+				Method: "GET", URL: "/open-apis/wiki/v2/spaces/get_node",
+				Body: map[string]interface{}{
+					"code": 0, "msg": "success",
+					"data": map[string]interface{}{
+						"node": map[string]interface{}{"obj_type": "bitable", "obj_token": "bitToken"},
+					},
+				},
+			})
+			args := []string{
+				"+add-comment",
+				"--doc", "https://example.larksuite.com/wiki/wikiToken",
+				"--content", `[{"type":"text","text":"test"}]`,
+				"--as", "user",
+			}
+			args = append(args, tc.args...)
+			err := mountAndRunDrive(t, DriveAddComment, args, f, stdout)
+			if err == nil || !strings.Contains(err.Error(), tc.wantErr) {
+				t.Fatalf("expected %q error, got: %v", tc.wantErr, err)
+			}
+		})
 	}
 }
 
@@ -1735,7 +2044,7 @@ func TestDocOldFormatLocalCommentRejected(t *testing.T) {
 		"--block-id", "blk_123",
 		"--as", "user",
 	}, f, stdout)
-	if err == nil || !strings.Contains(err.Error(), "only support docx, sheet, and slides") {
+	if err == nil || !strings.Contains(err.Error(), "only support docx, sheet, slides, and base(bitable)") {
 		t.Fatalf("expected local comment rejection for old doc, got: %v", err)
 	}
 }

shortcuts/drive/drive_search.go

@@ -149,6 +149,9 @@ var DriveSearch = common.Shortcut{
 			"page_token": data["page_token"],
 			"results":    normalizedItems,
 		}
+		if notice, _ := data["notice"].(string); notice != "" {
+			resultData["notice"] = notice
+		}
 
 		runtime.OutFormat(resultData, &output.Meta{Count: len(normalizedItems)}, func(w io.Writer) {
 			renderDriveSearchTable(w, data, normalizedItems)

shortcuts/drive/drive_search_test.go

@@ -14,12 +14,49 @@ import (
 	"time"
 
 	"github.com/larksuite/cli/errs"
+	"github.com/larksuite/cli/internal/cmdutil"
 	"github.com/larksuite/cli/internal/errclass"
+	"github.com/larksuite/cli/internal/httpmock"
 	"github.com/larksuite/cli/internal/output"
 )
 
-// TestClampOpenedTimeWindow covers the 3-month / 1-year boundary logic that
-// narrows --opened-since / --opened-until and generates the multi-slice notice.
+// TestDriveSearchExecutePassesThroughNotice verifies drive +search preserves notices.
+func TestDriveSearchExecutePassesThroughNotice(t *testing.T) {
+	const notice = "The query is too long and has been truncated to the first 50 characters for search."
+
+	f, stdout, _, reg := cmdutil.TestFactory(t, driveTestConfig())
+	reg.Register(&httpmock.Stub{
+		Method: "POST",
+		URL:    "/open-apis/search/v2/doc_wiki/search",
+		Body: map[string]interface{}{
+			"code": 0,
+			"msg":  "ok",
+			"data": map[string]interface{}{
+				"notice":     notice,
+				"res_units":  []interface{}{},
+				"total":      0,
+				"has_more":   false,
+				"page_token": "",
+			},
+		},
+	})
+
+	if err := mountAndRunDrive(t, DriveSearch, []string{"+search", "--query", "incident", "--format", "json", "--as", "user"}, f, stdout); err != nil {
+		t.Fatalf("DriveSearch.Execute() error = %v", err)
+	}
+	reg.Verify(t)
+
+	var env map[string]interface{}
+	if err := json.Unmarshal(stdout.Bytes(), &env); err != nil {
+		t.Fatalf("json.Unmarshal(stdout) error = %v\nstdout=%s", err, stdout.String())
+	}
+	data, _ := env["data"].(map[string]interface{})
+	if got, _ := data["notice"].(string); got != notice {
+		t.Fatalf("data.notice = %q, want %q; data=%#v", got, notice, data)
+	}
+}
+
+// TestClampOpenedTimeWindow covers opened-time clamping and slice notices.
 func TestClampOpenedTimeWindow(t *testing.T) {
 	t.Parallel()
 

shortcuts/im/builders_test.go

@@ -26,9 +26,7 @@ func mustMarshalDryRun(t *testing.T, v interface{}) string {
 	return string(b)
 }
 
-// newTestRuntimeContext builds a *common.RuntimeContext backed by a cobra
-// command whose flags are populated from the provided string and bool maps,
-// for unit-testing shortcut bodies, validators, and dry-run shapes.
+// newTestRuntimeContext builds a RuntimeContext with string and bool test flags.
 func newTestRuntimeContext(t *testing.T, stringFlags map[string]string, boolFlags map[string]bool) *common.RuntimeContext {
 	t.Helper()
 
@@ -59,9 +57,38 @@ func newTestRuntimeContext(t *testing.T, stringFlags map[string]string, boolFlag
 	return &common.RuntimeContext{Cmd: cmd}
 }
 
-// newMessagesSearchTestRuntimeContext is the messages-search variant of
-// newTestRuntimeContext: registers the search-specific --page-size flag
-// before applying caller-provided values.
+// newChatSearchTestRuntimeContext builds a chat-search RuntimeContext with typed flags.
+func newChatSearchTestRuntimeContext(t *testing.T, stringFlags map[string]string, boolFlags map[string]bool) *common.RuntimeContext {
+	t.Helper()
+
+	cmd := &cobra.Command{Use: "test"}
+	cmd.Flags().Int("page-size", 20, "")
+	for name := range stringFlags {
+		if name == "page-size" {
+			continue
+		}
+		cmd.Flags().String(name, "", "")
+	}
+	for name := range boolFlags {
+		cmd.Flags().Bool(name, false, "")
+	}
+	if err := cmd.ParseFlags(nil); err != nil {
+		t.Fatalf("ParseFlags() error = %v", err)
+	}
+	for name, val := range stringFlags {
+		if err := cmd.Flags().Set(name, val); err != nil {
+			t.Fatalf("Flags().Set(%q) error = %v", name, err)
+		}
+	}
+	for name, val := range boolFlags {
+		if err := cmd.Flags().Set(name, map[bool]string{true: "true", false: "false"}[val]); err != nil {
+			t.Fatalf("Flags().Set(%q) error = %v", name, err)
+		}
+	}
+	return &common.RuntimeContext{Cmd: cmd}
+}
+
+// newMessagesSearchTestRuntimeContext builds a messages-search RuntimeContext.
 func newMessagesSearchTestRuntimeContext(t *testing.T, stringFlags map[string]string, boolFlags map[string]bool) *common.RuntimeContext {
 	t.Helper()
 
@@ -231,6 +258,7 @@ func TestIsMediaKey(t *testing.T) {
 	}
 }
 
+// TestShortcutValidateBranches covers direct shortcut validation branches.
 func TestShortcutValidateBranches(t *testing.T) {
 
 	t.Run("ImChatCreate valid", func(t *testing.T) {
@@ -297,7 +325,7 @@ func TestShortcutValidateBranches(t *testing.T) {
 	})
 
 	t.Run("ImChatSearch invalid page size", func(t *testing.T) {
-		runtime := newTestRuntimeContext(t, map[string]string{
+		runtime := newChatSearchTestRuntimeContext(t, map[string]string{
 			"query":     "ok",
 			"page-size": "0",
 		}, nil)
@@ -307,12 +335,13 @@ func TestShortcutValidateBranches(t *testing.T) {
 		}
 	})
 
-	t.Run("ImChatSearch query too long", func(t *testing.T) {
-		runtime := newTestRuntimeContext(t, map[string]string{
-			"query": strings.Repeat("q", 65),
+	t.Run("ImChatSearch allows long query for server-side notice", func(t *testing.T) {
+		runtime := newChatSearchTestRuntimeContext(t, map[string]string{
+			"query":     strings.Repeat("q", 81),
+			"page-size": "20",
 		}, nil)
 		err := ImChatSearch.Validate(context.Background(), runtime)
-		if err == nil || !strings.Contains(err.Error(), "--query exceeds the maximum of 64 characters") {
+		if err != nil {
 			t.Fatalf("ImChatSearch.Validate() error = %v", err)
 		}
 	})
@@ -607,6 +636,7 @@ func TestShortcutValidateBranches(t *testing.T) {
 	})
 }
 
+// TestMessagesSearchPaginationConfig verifies page-all and page-limit behavior.
 func TestMessagesSearchPaginationConfig(t *testing.T) {
 	t.Run("default single page", func(t *testing.T) {
 		runtime := newMessagesSearchTestRuntimeContext(t, nil, nil)
@@ -650,8 +680,7 @@ func TestMessagesSearchPaginationConfig(t *testing.T) {
 	})
 }
 
-// TestShortcutDryRunShapes verifies that each shortcut's DryRun function
-// produces the expected API path, query parameters, and request body.
+// TestShortcutDryRunShapes verifies shortcut dry-run API paths and payloads.
 func TestShortcutDryRunShapes(t *testing.T) {
 	t.Run("ImChatCreate dry run includes params and body", func(t *testing.T) {
 		cmd := &cobra.Command{Use: "test"}
@@ -674,19 +703,19 @@ func TestShortcutDryRunShapes(t *testing.T) {
 	})
 
 	t.Run("ImChatSearch dry run includes built params", func(t *testing.T) {
-		runtime := newTestRuntimeContext(t, map[string]string{
+		runtime := newChatSearchTestRuntimeContext(t, map[string]string{
 			"query":      "team-alpha",
 			"page-size":  "50",
 			"page-token": "next_page",
 		}, nil)
 		got := mustMarshalDryRun(t, ImChatSearch.DryRun(context.Background(), runtime))
-		if !strings.Contains(got, `"/open-apis/im/v2/chats/search"`) || !strings.Contains(got, `"page_size":20`) || !strings.Contains(got, `"query":"\"team-alpha\""`) {
+		if !strings.Contains(got, `"/open-apis/im/v2/chats/search"`) || !strings.Contains(got, `"page_size":50`) || !strings.Contains(got, `"query":"\"team-alpha\""`) {
 			t.Fatalf("ImChatSearch.DryRun() = %s", got)
 		}
 	})
 
 	t.Run("ImChatSearch dry run still works with --exclude-muted set", func(t *testing.T) {
-		runtime := newTestRuntimeContext(t, map[string]string{
+		runtime := newChatSearchTestRuntimeContext(t, map[string]string{
 			"query": "team-alpha",
 		}, map[string]bool{
 			"exclude-muted": true,

shortcuts/im/convert_lib/card_userdsl_test.go

@@ -0,0 +1,993 @@
+// Copyright (c) 2026 Lark Technologies Pte. Ltd.
+// SPDX-License-Identifier: MIT
+
+package convertlib
+
+import (
+	"encoding/json"
+	"strings"
+	"testing"
+)
+
+func TestConvertInteractiveEventContent(t *testing.T) {
+	// invalid JSON → fallback
+	if got := ConvertInteractiveEventContent("not-json", nil); got != "[interactive card]" {
+		t.Fatalf("invalid JSON = %q, want [interactive card]", got)
+	}
+	// missing user_dsl → fallback
+	if got := ConvertInteractiveEventContent(`{"other":"field"}`, nil); got != "[interactive card]" {
+		t.Fatalf("missing user_dsl = %q, want [interactive card]", got)
+	}
+	// empty user_dsl → fallback
+	if got := ConvertInteractiveEventContent(`{"user_dsl":""}`, nil); got != "[interactive card]" {
+		t.Fatalf("empty user_dsl = %q, want [interactive card]", got)
+	}
+	// user_dsl that is not a string (wrong type) → fallback
+	if got := ConvertInteractiveEventContent(`{"user_dsl":123}`, nil); got != "[interactive card]" {
+		t.Fatalf("non-string user_dsl = %q, want [interactive card]", got)
+	}
+	// valid user-2 card → <card> output
+	userDsl := `{"schema":"2.0","header":{"title":{"tag":"plain_text","content":"Hello"}},"body":{"elements":[{"tag":"markdown","content":"world"}]}}`
+	rawContent := `{"user_dsl":"` + strings.ReplaceAll(userDsl, `"`, `\"`) + `"}`
+	got := ConvertInteractiveEventContent(rawContent, nil)
+	if !strings.HasPrefix(got, `<card title="Hello">`) {
+		t.Fatalf("valid card = %q, want prefix <card title=\"Hello\">", got)
+	}
+	if !strings.Contains(got, "world") {
+		t.Fatalf("valid card = %q, want to contain 'world'", got)
+	}
+}
+
+func makeMentionCard(mdContent string) string {
+	obj := map[string]interface{}{
+		"schema": "2.0",
+		"header": map[string]interface{}{
+			"title": map[string]interface{}{"tag": "plain_text", "content": "T"},
+		},
+		"body": map[string]interface{}{
+			"elements": []interface{}{
+				map[string]interface{}{"tag": "markdown", "content": mdContent},
+			},
+		},
+	}
+	dslBytes, _ := json.Marshal(obj)
+	raw, _ := json.Marshal(map[string]interface{}{"user_dsl": string(dslBytes)})
+	return string(raw)
+}
+
+func TestConvertInteractiveEventContentMentions(t *testing.T) {
+	mentions := []interface{}{
+		map[string]interface{}{
+			"key":  "@_user_1",
+			"name": "test-user",
+			"id":   map[string]interface{}{"open_id": "fake-uid-001"},
+		},
+	}
+
+	// quoted attrs: mention_key="key"
+	got := ConvertInteractiveEventContent(makeMentionCard(`hi <at mention_key="@_user_1">n</at> done`), mentions)
+	if !strings.Contains(got, "@test-user(fake-uid-001)") {
+		t.Fatalf("quoted mention_key not resolved, got: %s", got)
+	}
+
+	// unquoted attrs (real Lark format): <at id=ou_xxx mention_key=@_user_1></at>
+	got = ConvertInteractiveEventContent(makeMentionCard(`hi <at id=fake-uid-001 mention_key=@_user_1></at> done`), mentions)
+	if !strings.Contains(got, "@test-user(fake-uid-001)") {
+		t.Fatalf("unquoted mention_key not resolved, got: %s", got)
+	}
+
+	// mentions_key variant (unquoted)
+	got = ConvertInteractiveEventContent(makeMentionCard(`hi <at mentions_key=@_user_1></at> done`), mentions)
+	if !strings.Contains(got, "@test-user(fake-uid-001)") {
+		t.Fatalf("unquoted mentions_key not resolved, got: %s", got)
+	}
+
+	// degradation 1: no mention_key/mentions_key attr → fall back to @id (unquoted)
+	got = ConvertInteractiveEventContent(makeMentionCard(`hi <at id=fake-uid-001></at> done`), mentions)
+	if !strings.Contains(got, "@fake-uid-001") {
+		t.Fatalf("no mention_key unquoted: expected @id fallback, got: %s", got)
+	}
+
+	// degradation 2: mention_key not found in mentions → fall back to @id
+	got = ConvertInteractiveEventContent(makeMentionCard(`hi <at id=fake-uid-001 mention_key=@_unknown></at> done`), mentions)
+	if !strings.Contains(got, "@fake-uid-001") {
+		t.Fatalf("key not in mentions: expected @id fallback, got: %s", got)
+	}
+
+	// multi-mention: ids=id1,id2,id3 mentions_key=k1,,k3
+	// k1 hits → @name(id1), k2 empty → @id2 fallback, k3 not found → @id3 fallback
+	got = ConvertInteractiveEventContent(
+		makeMentionCard(`<at ids=fake-uid-001,fake-uid-002,fake-uid-003 mentions_key=@_user_1,,@_unknown></at>`),
+		mentions,
+	)
+	want := "@test-user(fake-uid-001)@fake-uid-002@fake-uid-003"
+	if !strings.Contains(got, want) {
+		t.Fatalf("multi-mention unquoted: want %q in output, got: %s", want, got)
+	}
+}
+
+func TestUserDslConverterSchema(t *testing.T) {
+	c := &userDslConverter{}
+
+	// user-2.ts: schema field present, header at root, body.elements
+	schema2 := cardObj{
+		"schema": "2.0",
+		"header": cardObj{
+			"title":    cardObj{"tag": "plain_text", "content": "Schema2 Title"},
+			"subtitle": cardObj{"tag": "plain_text", "content": "Sub"},
+		},
+		"body": cardObj{
+			"elements": []interface{}{
+				cardObj{"tag": "markdown", "content": "body text"},
+			},
+		},
+	}
+	got := c.convert(schema2)
+	if got != "<card title=\"Schema2 Title\" subtitle=\"Sub\">\nbody text\n</card>" {
+		t.Fatalf("schema2 = %q", got)
+	}
+
+	// user-1.ts: no schema field, i18n_header.zh_cn, elements at root
+	schema1 := cardObj{
+		"i18n_header": cardObj{
+			"zh_cn": cardObj{
+				"title": cardObj{"tag": "plain_text", "content": "Schema1 Title"},
+			},
+		},
+		"elements": []interface{}{
+			cardObj{"tag": "hr"},
+		},
+	}
+	got = c.convert(schema1)
+	if got != "<card title=\"Schema1 Title\">\n---\n</card>" {
+		t.Fatalf("schema1 = %q", got)
+	}
+
+	// user-1.ts: no schema, direct header (real Lark event format)
+	schema1Direct := cardObj{
+		"header": cardObj{
+			"title":    cardObj{"tag": "plain_text", "content": "Direct Header Title"},
+			"subtitle": cardObj{"tag": "plain_text", "content": "Direct Sub"},
+		},
+		"elements": []interface{}{
+			cardObj{"tag": "markdown", "content": "direct body"},
+		},
+	}
+	got = c.convert(schema1Direct)
+	if got != "<card title=\"Direct Header Title\" subtitle=\"Direct Sub\">\ndirect body\n</card>" {
+		t.Fatalf("schema1 direct header = %q", got)
+	}
+
+	// no header, no elements → fallback
+	got = c.convert(cardObj{})
+	if got != "[interactive card]" {
+		t.Fatalf("empty card = %q, want [interactive card]", got)
+	}
+
+	// card with title only → valid (not "[interactive card]")
+	titleOnly := cardObj{
+		"schema": "2.0",
+		"header": cardObj{"title": cardObj{"tag": "plain_text", "content": "TitleOnly"}},
+		"body":   cardObj{"elements": []interface{}{}},
+	}
+	got = c.convert(titleOnly)
+	if !strings.Contains(got, "TitleOnly") {
+		t.Fatalf("title-only card = %q, want to contain TitleOnly", got)
+	}
+}
+
+func TestUserDslConverterDispatch(t *testing.T) {
+	c := &userDslConverter{}
+
+	tests := []struct {
+		name     string
+		elem     cardObj
+		want     string
+		contains string
+	}{
+		{
+			name: "plain_text",
+			elem: cardObj{"tag": "plain_text", "content": "hello"},
+			want: "hello",
+		},
+		{
+			name: "markdown",
+			elem: cardObj{"tag": "markdown", "content": "**bold**"},
+			want: "**bold**",
+		},
+		{
+			name: "hr",
+			elem: cardObj{"tag": "hr"},
+			want: "---",
+		},
+		{
+			name: "br",
+			elem: cardObj{"tag": "br"},
+			want: "\n",
+		},
+		{
+			name: "img with img_key",
+			elem: cardObj{
+				"tag":     "img",
+				"img_key": "img_v3_abc",
+				"alt":     cardObj{"tag": "plain_text", "content": "Banner"},
+			},
+			want: "🖼️ Banner(img_key:img_v3_abc)",
+		},
+		{
+			name: "img_combination",
+			elem: cardObj{
+				"tag": "img_combination",
+				"img_list": []interface{}{
+					cardObj{"img_key": "k1"},
+					cardObj{"img_key": "k2"},
+				},
+			},
+			want: "🖼️ 2 image(s)(keys:k1,k2)",
+		},
+		{
+			name: "button with behaviors default_url",
+			elem: cardObj{
+				"tag":  "button",
+				"text": cardObj{"tag": "plain_text", "content": "Open"},
+				"behaviors": []interface{}{
+					cardObj{"type": "open_url", "default_url": "https://example.com"},
+				},
+			},
+			want: "[Open](https://example.com)",
+		},
+		{
+			name: "button disabled",
+			elem: cardObj{
+				"tag":      "button",
+				"text":     cardObj{"tag": "plain_text", "content": "Nope"},
+				"disabled": true,
+			},
+			want: "[Nope ✗]",
+		},
+		{
+			name: "button no url",
+			elem: cardObj{
+				"tag":  "button",
+				"text": cardObj{"tag": "plain_text", "content": "Submit"},
+			},
+			want: "[Submit]",
+		},
+		{
+			name: "action wrapper (user-1 style)",
+			elem: cardObj{
+				"tag": "action",
+				"actions": []interface{}{
+					cardObj{"tag": "button", "text": cardObj{"tag": "plain_text", "content": "A"}},
+					cardObj{"tag": "button", "text": cardObj{"tag": "plain_text", "content": "B"}},
+				},
+			},
+			want: "[A] [B]",
+		},
+		{
+			name: "overflow",
+			elem: cardObj{
+				"tag": "overflow",
+				"options": []interface{}{
+					cardObj{"text": cardObj{"tag": "plain_text", "content": "Edit"}},
+					cardObj{"text": cardObj{"tag": "plain_text", "content": "Delete"}},
+				},
+			},
+			want: "⋮ Edit, Delete",
+		},
+		{
+			name: "select_static no selection",
+			elem: cardObj{
+				"tag": "select_static",
+				"options": []interface{}{
+					cardObj{"text": cardObj{"tag": "plain_text", "content": "Option1"}, "value": "1"},
+					cardObj{"text": cardObj{"tag": "plain_text", "content": "Option2"}, "value": "2"},
+				},
+			},
+			want: "{Option1 / Option2 ▼}",
+		},
+		{
+			name: "select_static with initial_option",
+			elem: cardObj{
+				"tag":            "select_static",
+				"initial_option": "2",
+				"options": []interface{}{
+					cardObj{"text": cardObj{"tag": "plain_text", "content": "Option1"}, "value": "1"},
+					cardObj{"text": cardObj{"tag": "plain_text", "content": "Option2"}, "value": "2"},
+				},
+			},
+			want: "{Option1 / ✓Option2}",
+		},
+		{
+			name: "multi_select_static with selected_values",
+			elem: cardObj{
+				"tag":             "multi_select_static",
+				"selected_values": []interface{}{"A"},
+				"options": []interface{}{
+					cardObj{"text": cardObj{"tag": "plain_text", "content": "OptA"}, "value": "A"},
+					cardObj{"text": cardObj{"tag": "plain_text", "content": "OptB"}, "value": "B"},
+				},
+			},
+			want: "{✓OptA / OptB}(multi)",
+		},
+		{
+			name: "select_person no options no selection shows placeholder",
+			elem: cardObj{
+				"tag":         "select_person",
+				"placeholder": cardObj{"tag": "plain_text", "content": "请选择"},
+			},
+			want: "{请选择 ▼}",
+		},
+		{
+			name: "select_person with initial_option synthesizes from ID",
+			elem: cardObj{
+				"tag":            "select_person",
+				"initial_option": "fake-open-id-001",
+			},
+			want: "{✓fake-open-id-001}",
+		},
+		{
+			name: "multi_select_person with selected_values shows IDs and multi",
+			elem: cardObj{
+				"tag":             "multi_select_person",
+				"selected_values": []interface{}{"fake-open-id-001", "fake-open-id-002"},
+			},
+			want: "{✓fake-open-id-001 / ✓fake-open-id-002}(multi)",
+		},
+		{
+			name: "multi_select_person no selection shows placeholder",
+			elem: cardObj{
+				"tag":         "multi_select_person",
+				"placeholder": cardObj{"tag": "plain_text", "content": "添加人员"},
+			},
+			want: "{添加人员 ▼}(multi)",
+		},
+		{
+			name: "input with default_value",
+			elem: cardObj{
+				"tag":           "input",
+				"label":         cardObj{"tag": "plain_text", "content": "Reason"},
+				"default_value": "prefilled",
+			},
+			want: "Reason: prefilled___",
+		},
+		{
+			name: "input with placeholder",
+			elem: cardObj{
+				"tag":         "input",
+				"placeholder": cardObj{"tag": "plain_text", "content": "Type here"},
+			},
+			want: "Type here_____",
+		},
+		{
+			name: "date_picker with initial_date",
+			elem: cardObj{
+				"tag":          "date_picker",
+				"initial_date": "2026-01-01",
+			},
+			want: "📅 2026-01-01",
+		},
+		{
+			name: "date_picker placeholder",
+			elem: cardObj{
+				"tag":         "date_picker",
+				"placeholder": cardObj{"tag": "plain_text", "content": "Pick date"},
+			},
+			want: "📅 Pick date",
+		},
+		{
+			name: "picker_time with initial_time",
+			elem: cardObj{
+				"tag":          "picker_time",
+				"initial_time": "14:30",
+			},
+			want: "🕐 14:30",
+		},
+		{
+			name: "checker unchecked",
+			elem: cardObj{
+				"tag":  "checker",
+				"text": cardObj{"tag": "plain_text", "content": "Task A"},
+			},
+			want: "[ ] Task A",
+		},
+		{
+			name: "checker checked",
+			elem: cardObj{
+				"tag":     "checker",
+				"checked": true,
+				"text":    cardObj{"tag": "plain_text", "content": "Task B"},
+			},
+			want: "[x] Task B",
+		},
+		{
+			name: "chart with chart_spec",
+			elem: cardObj{
+				"tag": "chart",
+				"chart_spec": cardObj{
+					"title":  cardObj{"text": "Sales"},
+					"type":   "bar",
+					"xField": "month",
+					"yField": "value",
+					"data": cardObj{"values": []interface{}{
+						cardObj{"month": "Jan", "value": float64(10)},
+						cardObj{"month": "Feb", "value": float64(20)},
+					}},
+				},
+			},
+			want: "📊 Sales (Bar chart)\nSummary: Jan:10, Feb:20",
+		},
+		{
+			name: "chart with compound xField array",
+			elem: cardObj{
+				"tag": "chart",
+				"chart_spec": cardObj{
+					"title":  cardObj{"text": "Sales"},
+					"type":   "bar",
+					"xField": []interface{}{"month", "category"},
+					"yField": "value",
+					"data": cardObj{"values": []interface{}{
+						cardObj{"month": "Jan", "category": "A", "value": float64(10)},
+						cardObj{"month": "Feb", "category": "B", "value": float64(20)},
+					}},
+				},
+			},
+			want: "📊 Sales (Bar chart)\nSummary: Jan:10, Feb:20",
+		},
+		{
+			name: "chart no custom title uses type name",
+			elem: cardObj{
+				"tag": "chart",
+				"chart_spec": cardObj{
+					"type":          "pie",
+					"categoryField": "label",
+					"valueField":    "val",
+					"data": cardObj{"values": []interface{}{
+						cardObj{"label": "A", "val": float64(1)},
+					}},
+				},
+			},
+			want: "📊 Pie chart\nSummary: A:1",
+		},
+		{
+			name: "chart vchart array data format",
+			elem: cardObj{
+				"tag": "chart",
+				"chart_spec": cardObj{
+					"type":   "bar",
+					"xField": "x",
+					"yField": "y",
+					"data": []interface{}{
+						cardObj{"id": "s1", "values": []interface{}{
+							cardObj{"x": "Jan", "y": float64(5)},
+						}},
+						cardObj{"id": "s2", "values": []interface{}{
+							cardObj{"x": "Feb", "y": float64(8)},
+						}},
+					},
+				},
+			},
+			want: "📊 Bar chart\nSummary: Jan:5, Feb:8",
+		},
+		{
+			name: "text_tag",
+			elem: cardObj{
+				"tag":  "text_tag",
+				"text": cardObj{"tag": "plain_text", "content": "新功能"},
+			},
+			want: "「新功能」",
+		},
+		{
+			name: "avatar with user_id",
+			elem: cardObj{"tag": "avatar", "user_id": "fake-open-id-001"},
+			want: "👤(id:fake-open-id-001)",
+		},
+		{
+			name: "avatar no user_id",
+			elem: cardObj{"tag": "avatar"},
+			want: "👤",
+		},
+		{
+			name: "select_img no selection",
+			elem: cardObj{
+				"tag": "select_img",
+				"options": []interface{}{
+					cardObj{"value": "v1", "img_key": "img_k1"},
+					cardObj{"value": "v2", "img_key": "img_k2"},
+				},
+			},
+			want: "{🖼️ Image 1(v1)(img_key:img_k1) / 🖼️ Image 2(v2)(img_key:img_k2)}",
+		},
+		{
+			name: "select_img with selected",
+			elem: cardObj{
+				"tag":             "select_img",
+				"selected_values": []interface{}{"v1"},
+				"options": []interface{}{
+					cardObj{"value": "v1", "img_key": "img_k1"},
+					cardObj{"value": "v2", "img_key": "img_k2"},
+				},
+			},
+			want: "{✓🖼️ Image 1(v1)(img_key:img_k1) / 🖼️ Image 2(v2)(img_key:img_k2)}",
+		},
+		{
+			name: "repeat delegates to elements",
+			elem: cardObj{
+				"tag": "repeat",
+				"elements": []interface{}{
+					cardObj{"tag": "markdown", "content": "item A"},
+					cardObj{"tag": "markdown", "content": "item B"},
+				},
+			},
+			want: "item A\nitem B",
+		},
+		{
+			name: "audio with file_key",
+			elem: cardObj{"tag": "audio", "file_key": "file_abc123"},
+			want: "🎵 Audio(key:file_abc123)",
+		},
+		{
+			name: "audio fallback audio_id",
+			elem: cardObj{"tag": "audio", "audio_id": "audio_xyz"},
+			want: "🎵 Audio(key:audio_xyz)",
+		},
+		{
+			name: "video with file_key",
+			elem: cardObj{"tag": "video", "file_key": "video_abc"},
+			want: "🎬 Video(key:video_abc)",
+		},
+		{
+			name: "custom_icon returns empty",
+			elem: cardObj{"tag": "custom_icon", "img_key": "some_key"},
+			want: "",
+		},
+		{
+			name: "standard_icon returns empty",
+			elem: cardObj{"tag": "standard_icon", "token": "alarm_outlined"},
+			want: "",
+		},
+		{
+			name: "button disabled with disabled_tips",
+			elem: cardObj{
+				"tag":           "button",
+				"text":          cardObj{"tag": "plain_text", "content": "Submit"},
+				"disabled":      true,
+				"disabled_tips": cardObj{"tag": "plain_text", "content": "Not allowed"},
+			},
+			want: "[Submit ✗](tips:\"Not allowed\")",
+		},
+		{
+			name: "button with confirm",
+			elem: cardObj{
+				"tag":  "button",
+				"text": cardObj{"tag": "plain_text", "content": "Delete"},
+				"confirm": cardObj{
+					"title": cardObj{"tag": "plain_text", "content": "确认"},
+					"text":  cardObj{"tag": "plain_text", "content": "不可撤销"},
+				},
+			},
+			want: "[Delete](confirm:\"确认: 不可撤销\")",
+		},
+		{
+			name: "overflow with url",
+			elem: cardObj{
+				"tag": "overflow",
+				"options": []interface{}{
+					cardObj{"text": cardObj{"tag": "plain_text", "content": "Open"}, "url": "https://example.com"},
+					cardObj{"text": cardObj{"tag": "plain_text", "content": "Copy"}, "value": "copy"},
+				},
+			},
+			want: "⋮ [Open](https://example.com), Copy(copy)",
+		},
+		{
+			name: "select_static with initial_index",
+			elem: cardObj{
+				"tag":           "select_static",
+				"initial_index": float64(1),
+				"options": []interface{}{
+					cardObj{"text": cardObj{"tag": "plain_text", "content": "First"}, "value": "a"},
+					cardObj{"text": cardObj{"tag": "plain_text", "content": "Second"}, "value": "b"},
+				},
+			},
+			want: "{First / ✓Second}",
+		},
+		{
+			name: "div text with notation size",
+			elem: cardObj{
+				"tag": "div",
+				"text": cardObj{
+					"tag":       "plain_text",
+					"content":   "小字注释",
+					"text_size": "notation",
+				},
+			},
+			want: "📝 小字注释",
+		},
+		{
+			name: "form",
+			elem: cardObj{
+				"tag": "form",
+				"elements": []interface{}{
+					cardObj{"tag": "markdown", "content": "fill this"},
+				},
+			},
+			want: "<form>\nfill this\n</form>",
+		},
+		{
+			name: "collapsible_panel collapsed",
+			elem: cardObj{
+				"tag":      "collapsible_panel",
+				"expanded": false,
+				"header":   cardObj{"title": cardObj{"tag": "plain_text", "content": "Details"}},
+				"elements": []interface{}{
+					cardObj{"tag": "markdown", "content": "inner"},
+				},
+			},
+			want: "▶ Details\n    inner\n▲",
+		},
+		{
+			name: "collapsible_panel expanded",
+			elem: cardObj{
+				"tag":      "collapsible_panel",
+				"expanded": true,
+				"header":   cardObj{"title": cardObj{"tag": "plain_text", "content": "Details"}},
+				"elements": []interface{}{
+					cardObj{"tag": "markdown", "content": "inner"},
+				},
+			},
+			want: "▼ Details\n    inner\n▲",
+		},
+		{
+			name: "interactive_container with behaviors",
+			elem: cardObj{
+				"tag": "interactive_container",
+				"behaviors": []interface{}{
+					cardObj{"type": "open_url", "default_url": "https://example.com"},
+				},
+				"elements": []interface{}{
+					cardObj{"tag": "markdown", "content": "Click here"},
+				},
+			},
+			want: "<clickable url=\"https://example.com\">\nClick here\n</clickable>",
+		},
+		{
+			name: "interactive_container no url",
+			elem: cardObj{
+				"tag": "interactive_container",
+				"elements": []interface{}{
+					cardObj{"tag": "markdown", "content": "No link"},
+				},
+			},
+			want: "<clickable>\nNo link\n</clickable>",
+		},
+		{
+			name: "column_set with buttons → space-joined",
+			elem: cardObj{
+				"tag": "column_set",
+				"columns": []interface{}{
+					cardObj{"tag": "column", "elements": []interface{}{
+						cardObj{"tag": "button", "text": cardObj{"tag": "plain_text", "content": "X"}},
+					}},
+					cardObj{"tag": "column", "elements": []interface{}{
+						cardObj{"tag": "button", "text": cardObj{"tag": "plain_text", "content": "Y"}},
+					}},
+				},
+			},
+			want: "[X] [Y]",
+		},
+		{
+			name: "person",
+			elem: cardObj{"tag": "person", "user_id": "fake-open-id-002"},
+			want: "fake-open-id-002",
+		},
+		{
+			name: "unknown tag fallback to content",
+			elem: cardObj{"tag": "mystery", "content": "mystery content"},
+			want: "mystery content",
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			got := c.convertElement(tt.elem, 0)
+			if tt.contains != "" {
+				if !strings.Contains(got, tt.contains) {
+					t.Fatalf("convertElement(%s) = %q, want to contain %q", tt.name, got, tt.contains)
+				}
+				return
+			}
+			if got != tt.want {
+				t.Fatalf("convertElement(%s) = %q, want %q", tt.name, got, tt.want)
+			}
+		})
+	}
+}
+
+func TestUserDslExtractButtonURL(t *testing.T) {
+	c := &userDslConverter{}
+
+	// direct url field wins first
+	got := c.extractButtonURL(cardObj{
+		"url":       "https://example.com/direct",
+		"multi_url": cardObj{"url": "https://example.com/multi"},
+		"behaviors": []interface{}{
+			cardObj{"type": "open_url", "default_url": "https://example.com/behavior"},
+		},
+	})
+	if got != "https://example.com/direct" {
+		t.Fatalf("direct url = %q, want https://example.com/direct", got)
+	}
+
+	// multi_url.url when no direct url
+	got = c.extractButtonURL(cardObj{
+		"multi_url": cardObj{"url": "https://example.com/multi"},
+		"behaviors": []interface{}{
+			cardObj{"type": "open_url", "default_url": "https://example.com/behavior"},
+		},
+	})
+	if got != "https://example.com/multi" {
+		t.Fatalf("multi_url = %q, want https://example.com/multi", got)
+	}
+
+	// behaviors default_url as last resort
+	got = c.extractButtonURL(cardObj{
+		"behaviors": []interface{}{
+			cardObj{"type": "open_url", "default_url": "https://example.com/behavior"},
+		},
+	})
+	if got != "https://example.com/behavior" {
+		t.Fatalf("behaviors = %q, want https://example.com/behavior", got)
+	}
+
+	// non-open_url behavior is ignored
+	got = c.extractButtonURL(cardObj{
+		"behaviors": []interface{}{
+			cardObj{"type": "callback", "default_url": "https://example.com/callback"},
+		},
+	})
+	if got != "" {
+		t.Fatalf("non-open_url = %q, want empty", got)
+	}
+
+	// no url anywhere → empty
+	got = c.extractButtonURL(cardObj{"text": cardObj{"content": "No URL"}})
+	if got != "" {
+		t.Fatalf("no url = %q, want empty", got)
+	}
+}
+
+func TestUserDslExtractTableCellValue(t *testing.T) {
+	c := &userDslConverter{}
+
+	// nil
+	if got := c.extractUserDslTableCellValue(nil); got != "" {
+		t.Fatalf("nil = %q, want empty", got)
+	}
+	// string
+	if got := c.extractUserDslTableCellValue("hello"); got != "hello" {
+		t.Fatalf("string = %q, want 'hello'", got)
+	}
+	// float64 integer
+	if got := c.extractUserDslTableCellValue(float64(42)); got != "42" {
+		t.Fatalf("int float = %q, want '42'", got)
+	}
+	// float64 decimal
+	if got := c.extractUserDslTableCellValue(float64(3.14)); got != "3.14" {
+		t.Fatalf("float = %q, want '3.14'", got)
+	}
+	// []interface{} with text tags → 「text」 format
+	got := c.extractUserDslTableCellValue([]interface{}{
+		cardObj{"text": "S2", "color": "blue"},
+		cardObj{"text": "M1", "color": "red"},
+	})
+	if got != "「S2」 「M1」" {
+		t.Fatalf("tag array = %q, want '「S2」 「M1」'", got)
+	}
+	// cardObj with content field
+	got = c.extractUserDslTableCellValue(cardObj{"content": "cell content"})
+	if got != "cell content" {
+		t.Fatalf("cardObj with content = %q, want 'cell content'", got)
+	}
+}
+
+func TestUserDslConvertTable(t *testing.T) {
+	c := &userDslConverter{}
+
+	got := c.convertTable(cardObj{
+		"columns": []interface{}{
+			cardObj{"display_name": "客户名称", "name": "customer_name"},
+			cardObj{"display_name": "规模", "name": "scale"},
+			cardObj{"display_name": "金额", "name": "arr"},
+		},
+		"rows": []interface{}{
+			cardObj{
+				"customer_name": "飞书科技",
+				"scale":         []interface{}{cardObj{"text": "S2", "color": "blue"}},
+				"arr":           float64(16800),
+			},
+		},
+	})
+	want := "| 客户名称 | 规模 | 金额 |\n|------|------|------|\n| 飞书科技 | 「S2」 | 16800 |"
+	if got != want {
+		t.Fatalf("convertTable() = %q, want %q", got, want)
+	}
+
+	// no columns → empty
+	if got := c.convertTable(cardObj{}); got != "" {
+		t.Fatalf("no columns = %q, want empty", got)
+	}
+}
+
+func TestLarkMdMentionResolution(t *testing.T) {
+	mentions := []interface{}{
+		map[string]interface{}{
+			"key":  "@_user_1",
+			"name": "test-user",
+			"id":   map[string]interface{}{"open_id": "fake-uid-001"},
+		},
+	}
+
+	// lark_md in div.text — the real Lark event format (C01 case)
+	card := map[string]interface{}{
+		"elements": []interface{}{
+			map[string]interface{}{
+				"tag": "div",
+				"text": map[string]interface{}{
+					"tag":     "lark_md",
+					"content": "Hello <at id=fake-uid-001></at> check this.",
+				},
+			},
+		},
+	}
+	dslBytes, _ := json.Marshal(card)
+	raw, _ := json.Marshal(map[string]interface{}{"user_dsl": string(dslBytes)})
+	got := ConvertInteractiveEventContent(string(raw), mentions)
+	if strings.Contains(got, "<at") {
+		t.Fatalf("div.text lark_md: raw <at> tag not resolved, got: %s", got)
+	}
+	if !strings.Contains(got, "@fake-uid-001") {
+		t.Fatalf("div.text lark_md: @id not in output, got: %s", got)
+	}
+
+	// lark_md in note.elements (C02 case)
+	card = map[string]interface{}{
+		"elements": []interface{}{
+			map[string]interface{}{
+				"tag": "note",
+				"elements": []interface{}{
+					map[string]interface{}{
+						"tag":     "lark_md",
+						"content": "Note: <at id=fake-uid-001></at> check.",
+					},
+				},
+			},
+		},
+	}
+	dslBytes, _ = json.Marshal(card)
+	raw, _ = json.Marshal(map[string]interface{}{"user_dsl": string(dslBytes)})
+	got = ConvertInteractiveEventContent(string(raw), mentions)
+	if strings.Contains(got, "<at") {
+		t.Fatalf("note lark_md: raw <at> tag not resolved, got: %s", got)
+	}
+	if !strings.Contains(got, "@fake-uid-001") {
+		t.Fatalf("note lark_md: @id not in output, got: %s", got)
+	}
+
+	// mention_key resolution via mentions map
+	card = map[string]interface{}{
+		"elements": []interface{}{
+			map[string]interface{}{
+				"tag": "div",
+				"text": map[string]interface{}{
+					"tag":     "lark_md",
+					"content": `Hi <at mention_key="@_user_1">n</at> done.`,
+				},
+			},
+		},
+	}
+	dslBytes, _ = json.Marshal(card)
+	raw, _ = json.Marshal(map[string]interface{}{"user_dsl": string(dslBytes)})
+	got = ConvertInteractiveEventContent(string(raw), mentions)
+	if !strings.Contains(got, "@test-user(fake-uid-001)") {
+		t.Fatalf("div.text lark_md mention_key: want @test-user(fake-uid-001), got: %s", got)
+	}
+}
+
+func TestConvertUserDslCardEndToEnd(t *testing.T) {
+	// user-2.ts format — matches structure of docs/user-dsl/user-example-2.json
+	schema2JSON := `{
+		"schema": "2.0",
+		"header": {
+			"title": {"tag": "plain_text", "content": "飞书卡片组件展示"},
+			"template": "blue"
+		},
+		"body": {
+			"elements": [
+				{"tag": "markdown", "content": "### 基础文本"},
+				{"tag": "hr"},
+				{
+					"tag": "img",
+					"img_key": "img_v3_02122_abc",
+					"alt": {"tag": "plain_text", "content": "示例图片"}
+				},
+				{
+					"tag": "button",
+					"text": {"tag": "plain_text", "content": "主要按钮"},
+					"behaviors": [{"type": "open_url", "default_url": "https://example.com"}]
+				},
+				{
+					"tag": "table",
+					"columns": [
+						{"display_name": "名称", "name": "name"},
+						{"display_name": "数值", "name": "value"}
+					],
+					"rows": [
+						{"name": "项目A", "value": 100},
+						{"name": "项目B", "value": 200}
+					]
+				}
+			]
+		}
+	}`
+
+	got := convertUserDslCard(schema2JSON, nil)
+
+	if !strings.HasPrefix(got, `<card title="飞书卡片组件展示">`) {
+		t.Fatalf("e2e schema2: missing card title prefix, got: %s", got)
+	}
+	if !strings.Contains(got, "### 基础文本") {
+		t.Fatal("e2e schema2: missing markdown content")
+	}
+	if !strings.Contains(got, "---") {
+		t.Fatal("e2e schema2: missing hr")
+	}
+	if !strings.Contains(got, "🖼️ 示例图片(img_key:img_v3_02122_abc)") {
+		t.Fatalf("e2e schema2: missing image, got: %s", got)
+	}
+	if !strings.Contains(got, "[主要按钮](https://example.com)") {
+		t.Fatalf("e2e schema2: missing button, got: %s", got)
+	}
+	if !strings.Contains(got, "| 名称 | 数值 |") {
+		t.Fatal("e2e schema2: missing table header")
+	}
+	if !strings.Contains(got, "| 项目A | 100 |") {
+		t.Fatalf("e2e schema2: missing table row, got: %s", got)
+	}
+	if !strings.HasSuffix(got, "</card>") {
+		t.Fatalf("e2e schema2: missing </card> suffix, got: %s", got)
+	}
+
+	// user-1.ts format
+	schema1JSON := `{
+		"i18n_header": {
+			"zh_cn": {
+				"title": {"tag": "plain_text", "content": "Schema1 卡片"},
+				"template": "blue"
+			}
+		},
+		"elements": [
+			{"tag": "markdown", "content": "Hello **World**"},
+			{
+				"tag": "action",
+				"actions": [
+					{
+						"tag": "button",
+						"text": {"tag": "plain_text", "content": "跳转"},
+						"behaviors": [{"type": "open_url", "default_url": "https://example.com"}]
+					}
+				]
+			}
+		]
+	}`
+
+	got = convertUserDslCard(schema1JSON, nil)
+	if !strings.HasPrefix(got, `<card title="Schema1 卡片">`) {
+		t.Fatalf("e2e schema1: missing card title, got: %s", got)
+	}
+	if !strings.Contains(got, "Hello **World**") {
+		t.Fatal("e2e schema1: missing markdown")
+	}
+	if !strings.Contains(got, "[跳转](https://example.com)") {
+		t.Fatalf("e2e schema1: missing button, got: %s", got)
+	}
+}

shortcuts/im/convert_lib/helpers_test.go

@@ -122,7 +122,7 @@ func TestExtractPostBlocksText(t *testing.T) {
 	}
 
 	got := extractPostBlocksText(blocks)
-	want := "hello @Alice [docs](https://example.com)\n[Image: img_123]"
+	want := "hello @Alice [docs](https://example.com)\n![Image](img_123)"
 	if got != want {
 		t.Fatalf("extractPostBlocksText() = %q, want %q", got, want)
 	}

shortcuts/im/convert_lib/text.go

@@ -39,16 +39,16 @@ func (postConverter) Convert(ctx *ConvertContext) string {
 	if title, _ := body["title"].(string); title != "" {
 		parts = append(parts, title)
 	}
-	if blocks, _ := body["content"].([]interface{}); len(blocks) > 0 {
-		for _, para := range blocks {
-			elems, _ := para.([]interface{})
-			var line strings.Builder
-			for _, el := range elems {
-				elem, _ := el.(map[string]interface{})
-				line.WriteString(renderPostElem(elem))
-			}
-			parts = append(parts, line.String())
+	// Prefer content_v2 blocks; fallback to content blocks
+	blocks := selectContentBlocks(body)
+	for _, para := range blocks {
+		elems, _ := para.([]interface{})
+		var line strings.Builder
+		for _, el := range elems {
+			elem, _ := el.(map[string]interface{})
+			line.WriteString(renderPostElem(elem))
 		}
+		parts = append(parts, line.String())
 	}
 
 	result := strings.TrimSpace(strings.Join(parts, "\n"))
@@ -58,6 +58,17 @@ func (postConverter) Convert(ctx *ConvertContext) string {
 	return ResolveMentionKeys(result, ctx.MentionMap)
 }
 
+// selectContentBlocks returns content_v2 blocks when present and non-empty;
+// otherwise falls back to content blocks. This implements the content_v2
+// priority rule for post messages.
+func selectContentBlocks(body map[string]interface{}) []interface{} {
+	if v2, ok := body["content_v2"].([]interface{}); ok && len(v2) > 0 {
+		return v2
+	}
+	blocks, _ := body["content"].([]interface{})
+	return blocks
+}
+
 func unwrapPostLocale(parsed map[string]interface{}) map[string]interface{} {
 	if _, ok := parsed["content"]; ok {
 		return parsed
@@ -114,10 +125,14 @@ func renderPostElem(el map[string]interface{}) string {
 		var rendered string
 		switch {
 		case userId == "@_all" || userId == "all":
-			rendered = "@all"
+			rendered = `<at user_id="all"></at>`
 		default:
 			if name, _ := el["user_name"].(string); name != "" {
-				rendered = "@" + name
+				if userId != "" && strings.HasPrefix(userId, "ou") {
+					rendered = fmt.Sprintf(`<at user_id="%s">%s</at>`, userId, name)
+				} else {
+					rendered = "@" + name
+				}
 			} else {
 				rendered = "@" + userId
 			}
@@ -138,7 +153,7 @@ func renderPostElem(el map[string]interface{}) string {
 	case "img":
 		key, _ := el["image_key"].(string)
 		if key != "" {
-			return fmt.Sprintf("[Image: %s]", key)
+			return fmt.Sprintf("![Image](%s)", key)
 		}
 		return "[Image]"
 	case "media":

shortcuts/im/convert_lib/text_test.go

@@ -93,9 +93,13 @@ func TestRenderPostElem(t *testing.T) {
 	}{
 		{name: "text", el: map[string]interface{}{"tag": "text", "text": "hello"}, want: "hello"},
 		{name: "link", el: map[string]interface{}{"tag": "a", "text": "doc", "href": "https://example.com"}, want: "[doc](https://example.com)"},
-		{name: "mention all", el: map[string]interface{}{"tag": "at", "user_id": "@_all"}, want: "@all"},
-		{name: "mention user", el: map[string]interface{}{"tag": "at", "user_name": "Alice"}, want: "@Alice"},
-		{name: "image", el: map[string]interface{}{"tag": "img", "image_key": "img_123"}, want: "[Image: img_123]"},
+		{name: "mention all", el: map[string]interface{}{"tag": "at", "user_id": "@_all"}, want: `<at user_id="all"></at>`},
+		{name: "mention user with id", el: map[string]interface{}{"tag": "at", "user_id": "ou_user_1", "user_name": "Alice"}, want: `<at user_id="ou_user_1">Alice</at>`},
+		{name: "mention user name only", el: map[string]interface{}{"tag": "at", "user_name": "Alice"}, want: "@Alice"},
+		{name: "mention user id only", el: map[string]interface{}{"tag": "at", "user_id": "@_user_1"}, want: "@@_user_1"},
+		{name: "image", el: map[string]interface{}{"tag": "img", "image_key": "img_123"}, want: "![Image](img_123)"},
+		{name: "image no key", el: map[string]interface{}{"tag": "img"}, want: "[Image]"},
+		{name: "md text", el: map[string]interface{}{"tag": "md", "text": "##### 标题\n\n<at user_id=\"ou_xxx\">Alice</at> 你好"}, want: "##### 标题\n\n<at user_id=\"ou_xxx\">Alice</at> 你好"},
 		{name: "media", el: map[string]interface{}{"tag": "media", "file_key": "file_123"}, want: "[Media: file_123]"},
 		{name: "code block", el: map[string]interface{}{"tag": "code_block", "language": "go", "text": "fmt.Println(1)"}, want: "\n```go\nfmt.Println(1)\n```\n"},
 		{name: "hr", el: map[string]interface{}{"tag": "hr"}, want: "\n---\n"},
@@ -144,3 +148,87 @@ func TestRenderPostElemEmotionStyleMd(t *testing.T) {
 		})
 	}
 }
+
+func TestSelectContentBlocks(t *testing.T) {
+	tests := []struct {
+		name string
+		body map[string]interface{}
+		want int
+	}{
+		{
+			name: "content_v2 present and non-empty",
+			body: map[string]interface{}{
+				"content":    []interface{}{[]interface{}{map[string]interface{}{"tag": "text", "text": "old"}}},
+				"content_v2": []interface{}{[]interface{}{map[string]interface{}{"tag": "md", "text": "new"}}},
+			},
+			want: 1,
+		},
+		{
+			name: "content_v2 empty array",
+			body: map[string]interface{}{
+				"content":    []interface{}{[]interface{}{map[string]interface{}{"tag": "text", "text": "old"}}},
+				"content_v2": []interface{}{},
+			},
+			want: 1,
+		},
+		{
+			name: "content_v2 nil",
+			body: map[string]interface{}{
+				"content": []interface{}{[]interface{}{map[string]interface{}{"tag": "text", "text": "old"}}},
+			},
+			want: 1,
+		},
+		{
+			name: "content_v2 wrong type",
+			body: map[string]interface{}{
+				"content":    []interface{}{[]interface{}{map[string]interface{}{"tag": "text", "text": "old"}}},
+				"content_v2": "not_an_array",
+			},
+			want: 1,
+		},
+		{
+			name: "both missing",
+			body: map[string]interface{}{},
+			want: 0,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			got := selectContentBlocks(tt.body)
+			if len(got) != tt.want {
+				t.Fatalf("selectContentBlocks() len = %d, want %d", len(got), tt.want)
+			}
+		})
+	}
+}
+
+func TestPostConverterConvertContentV2(t *testing.T) {
+	// AC-M1-H1: content_v2 present → use content_v2 blocks (md passthrough)
+	ctx := &ConvertContext{
+		RawContent: `{"content_v2":[[{"tag":"md","text":"##### 标题\n\n<at user_id=\"ou_xxx\">Alice</at> 你好"}]],"content":[[{"tag":"text","text":"old path"}]]}`,
+	}
+	want := "##### 标题\n\n<at user_id=\"ou_xxx\">Alice</at> 你好"
+	if got := (postConverter{}).Convert(ctx); got != want {
+		t.Fatalf("postConverter.Convert(content_v2) = %q, want %q", got, want)
+	}
+
+	// AC-M1-H2: no content_v2 → use content blocks with new at/img format
+	ctx2 := &ConvertContext{
+		RawContent: `{"content":[[{"tag":"at","user_id":"ou_xxx","user_name":"Bob"},{"tag":"text","text":" "},{"tag":"img","image_key":"img_123"}]]}`,
+		Mentions:   []interface{}{map[string]interface{}{"key": "ou_xxx", "id": "ou_bob", "name": "Bob"}},
+	}
+	want2 := `<at user_id="ou_xxx">Bob</at> ![Image](img_123)`
+	if got := (postConverter{}).Convert(ctx2); got != want2 {
+		t.Fatalf("postConverter.Convert(content) = %q, want %q", got, want2)
+	}
+
+	// AC-M1-E1: content_v2 empty → fallback to content
+	ctx3 := &ConvertContext{
+		RawContent: `{"content_v2":[],"content":[[{"tag":"text","text":"fallback path"}]]}`,
+	}
+	want3 := "fallback path"
+	if got := (postConverter{}).Convert(ctx3); got != want3 {
+		t.Fatalf("postConverter.Convert(empty content_v2) = %q, want %q", got, want3)
+	}
+}

shortcuts/im/im_chat_search.go

@@ -29,7 +29,7 @@ var ImChatSearch = common.Shortcut{
 	AuthTypes:   []string{"user", "bot"},
 	HasFormat:   true,
 	Flags: []common.Flag{
-		{Name: "query", Desc: "search keyword (max 64 chars)"},
+		{Name: "query", Desc: "search keyword (server may return data.notice for overly long input)"},
 		{Name: "search-types", Desc: "chat types, comma-separated (private, external, public_joined, public_not_joined)"},
 		{Name: "chat-modes", Desc: "filter by chat mode, comma-separated (group, topic)"},
 		{Name: "member-ids", Desc: "filter by member open_ids, comma-separated"},
@@ -50,7 +50,7 @@ var ImChatSearch = common.Shortcut{
 			Params(params).
 			Body(body)
 	},
-	// Validate enforces query/member-ids presence, --query rune cap, search-types
+	// Validate enforces query/member-ids presence, search-types
 	// enum, --member-ids count and format, and --page-size bounds.
 	Validate: func(ctx context.Context, runtime *common.RuntimeContext) error {
 		query := runtime.Str("query")
@@ -58,9 +58,6 @@ var ImChatSearch = common.Shortcut{
 		if query == "" && memberIDs == "" {
 			return errs.NewValidationError(errs.SubtypeInvalidArgument, "--query and --member-ids cannot both be empty; provide at least one (e.g. --query \"team-name\" or --member-ids \"ou_xxx\")")
 		}
-		if query != "" && len([]rune(query)) > 64 {
-			return errs.NewValidationError(errs.SubtypeInvalidArgument, "--query exceeds the maximum of 64 characters (got %d)", len([]rune(query))).WithParam("--query")
-		}
 		if st := runtime.Str("search-types"); st != "" {
 			allowed := map[string]struct{}{
 				"private":           {},
@@ -151,6 +148,9 @@ var ImChatSearch = common.Shortcut{
 			"has_more":   hasMore,
 			"page_token": pageToken,
 		}
+		if notice, _ := resData["notice"].(string); notice != "" {
+			outData["notice"] = notice
+		}
 		if mfOut.Meta.Applied != "" {
 			outData["filter"] = MuteFilterMetaToMap(mfOut.Meta)
 		}

shortcuts/im/im_messages_search.go

@@ -91,7 +91,7 @@ var ImMessagesSearch = common.Shortcut{
 			return err
 		}
 
-		rawItems, hasMore, nextPageToken, truncatedByLimit, pageLimit, err := searchMessages(runtime, req)
+		rawItems, hasMore, nextPageToken, truncatedByLimit, pageLimit, notice, err := searchMessages(runtime, req)
 		if err != nil {
 			return err
 		}
@@ -103,6 +103,9 @@ var ImMessagesSearch = common.Shortcut{
 				"has_more":   hasMore,
 				"page_token": nextPageToken,
 			}
+			if notice != "" {
+				outData["notice"] = notice
+			}
 			runtime.OutFormat(outData, nil, func(w io.Writer) {
 				fmt.Fprintln(w, "No matching messages found.")
 			})
@@ -131,6 +134,9 @@ var ImMessagesSearch = common.Shortcut{
 				"page_token":  nextPageToken,
 				"note":        "failed to fetch message details, returning ID list only",
 			}
+			if notice != "" {
+				outData["notice"] = notice
+			}
 			runtime.OutFormat(outData, nil, func(w io.Writer) {
 				fmt.Fprintf(w, "Found %d messages (failed to fetch details):\n", len(messageIds))
 				for _, id := range messageIds {
@@ -206,6 +212,9 @@ var ImMessagesSearch = common.Shortcut{
 			"has_more":   hasMore,
 			"page_token": nextPageToken,
 		}
+		if notice != "" {
+			outData["notice"] = notice
+		}
 		runtime.OutFormat(outData, nil, func(w io.Writer) {
 			if len(enriched) == 0 {
 				fmt.Fprintln(w, "No matching messages found.")
@@ -377,6 +386,7 @@ func buildMessagesSearchRequest(runtime *common.RuntimeContext) (*messagesSearch
 	}, nil
 }
 
+// messagesSearchPaginationConfig derives auto-pagination mode and page limit.
 func messagesSearchPaginationConfig(runtime *common.RuntimeContext) (autoPaginate bool, pageLimit int) {
 	autoPaginate = runtime.Bool("page-all")
 	if runtime.Cmd != nil && runtime.Cmd.Flags().Changed("page-limit") {
@@ -392,7 +402,8 @@ func messagesSearchPaginationConfig(runtime *common.RuntimeContext) (autoPaginat
 	return autoPaginate, pageLimit
 }
 
-func searchMessages(runtime *common.RuntimeContext, req *messagesSearchRequest) ([]interface{}, bool, string, bool, int, error) {
+// searchMessages fetches message search pages and returns the first server notice.
+func searchMessages(runtime *common.RuntimeContext, req *messagesSearchRequest) ([]interface{}, bool, string, bool, int, string, error) {
 	autoPaginate, pageLimit := messagesSearchPaginationConfig(runtime)
 	pageToken := ""
 	if tokens := req.params["page_token"]; len(tokens) > 0 {
@@ -410,6 +421,7 @@ func searchMessages(runtime *common.RuntimeContext, req *messagesSearchRequest)
 		lastPageToken    string
 		truncatedByLimit bool
 		pageCount        int
+		notice           string
 	)
 
 	for {
@@ -423,9 +435,12 @@ func searchMessages(runtime *common.RuntimeContext, req *messagesSearchRequest)
 
 		searchData, err := runtime.DoAPIJSONTyped(http.MethodPost, "/open-apis/im/v1/messages/search", params, req.body)
 		if err != nil {
-			return nil, false, "", false, pageLimit, err
+			return nil, false, "", false, pageLimit, "", err
 		}
 
+		if notice == "" {
+			notice, _ = searchData["notice"].(string)
+		}
 		items, _ := searchData["items"].([]interface{})
 		allItems = append(allItems, items...)
 		lastHasMore, lastPageToken = common.PaginationMeta(searchData)
@@ -441,9 +456,10 @@ func searchMessages(runtime *common.RuntimeContext, req *messagesSearchRequest)
 		pageToken = lastPageToken
 	}
 
-	return allItems, lastHasMore, lastPageToken, truncatedByLimit, pageLimit, nil
+	return allItems, lastHasMore, lastPageToken, truncatedByLimit, pageLimit, notice, nil
 }
 
+// batchMGetMessages fetches message details in API-sized batches.
 func batchMGetMessages(runtime *common.RuntimeContext, messageIds []string) ([]interface{}, error) {
 	var items []interface{}
 	for _, batch := range chunkStrings(messageIds, messagesSearchMGetBatchSize) {
@@ -457,6 +473,7 @@ func batchMGetMessages(runtime *common.RuntimeContext, messageIds []string) ([]i
 	return items, nil
 }
 
+// batchQueryChatContexts fetches chat metadata best-effort for message rows.
 func batchQueryChatContexts(runtime *common.RuntimeContext, chatIds []string) map[string]map[string]interface{} {
 	chatContexts := map[string]map[string]interface{}{}
 	// Best-effort: a failed chunk only loses its own entries.
@@ -466,6 +483,7 @@ func batchQueryChatContexts(runtime *common.RuntimeContext, chatIds []string) ma
 	return chatContexts
 }
 
+// chunkStrings splits a string slice into fixed-size batches.
 func chunkStrings(items []string, chunkSize int) [][]string {
 	if len(items) == 0 || chunkSize <= 0 {
 		return nil

shortcuts/im/im_search_notice_test.go

@@ -0,0 +1,129 @@
+// Copyright (c) 2026 Lark Technologies Pte. Ltd.
+// SPDX-License-Identifier: MIT
+
+package im
+
+import (
+	"bytes"
+	"context"
+	"encoding/json"
+	"fmt"
+	"net/http"
+	"strings"
+	"testing"
+
+	"github.com/larksuite/cli/shortcuts/common"
+	"github.com/spf13/cobra"
+)
+
+// TestImChatSearchExecutePassesThroughNotice verifies chat search notice output.
+func TestImChatSearchExecutePassesThroughNotice(t *testing.T) {
+	const notice = "The query is too long and has been truncated to the first 50 characters for search."
+	longQuery := strings.Repeat("q", 81)
+
+	runtime := newBotShortcutRuntime(t, shortcutRoundTripFunc(func(req *http.Request) (*http.Response, error) {
+		if !strings.Contains(req.URL.Path, "/open-apis/im/v2/chats/search") {
+			return nil, fmt.Errorf("unexpected request: %s", req.URL.String())
+		}
+		var body map[string]interface{}
+		if err := json.NewDecoder(req.Body).Decode(&body); err != nil {
+			return nil, fmt.Errorf("decode request body: %w", err)
+		}
+		if got, _ := body["query"].(string); got != longQuery {
+			return nil, fmt.Errorf("body.query = %q, want %q", got, longQuery)
+		}
+		return shortcutJSONResponse(200, map[string]interface{}{
+			"code": 0,
+			"data": map[string]interface{}{
+				"notice":     notice,
+				"items":      []interface{}{},
+				"total":      0,
+				"has_more":   false,
+				"page_token": "",
+			},
+		}), nil
+	}))
+	runtime.Cmd = newChatSearchNoticeTestCommand(t, longQuery)
+	runtime.Format = "json"
+
+	if err := ImChatSearch.Execute(context.Background(), runtime); err != nil {
+		t.Fatalf("ImChatSearch.Execute() error = %v", err)
+	}
+
+	data := decodeShortcutData(t, runtime)
+	if got, _ := data["notice"].(string); got != notice {
+		t.Fatalf("data.notice = %q, want %q; data=%#v", got, notice, data)
+	}
+}
+
+// TestImMessagesSearchExecutePassesThroughNotice verifies message search notice output.
+func TestImMessagesSearchExecutePassesThroughNotice(t *testing.T) {
+	const notice = "The query is too long and has been truncated to the first 50 characters for search."
+
+	runtime := newMessagesSearchRuntime(t, map[string]string{
+		"query": "incident",
+	}, nil, shortcutRoundTripFunc(func(req *http.Request) (*http.Response, error) {
+		if !strings.Contains(req.URL.Path, "/open-apis/im/v1/messages/search") {
+			return nil, fmt.Errorf("unexpected request: %s", req.URL.String())
+		}
+		return shortcutJSONResponse(200, map[string]interface{}{
+			"code": 0,
+			"data": map[string]interface{}{
+				"notice":     notice,
+				"items":      []interface{}{},
+				"has_more":   false,
+				"page_token": "",
+			},
+		}), nil
+	}))
+	runtime.Format = "json"
+
+	if err := ImMessagesSearch.Execute(context.Background(), runtime); err != nil {
+		t.Fatalf("ImMessagesSearch.Execute() error = %v", err)
+	}
+
+	data := decodeShortcutData(t, runtime)
+	if got, _ := data["notice"].(string); got != notice {
+		t.Fatalf("data.notice = %q, want %q; data=%#v", got, notice, data)
+	}
+}
+
+// newChatSearchNoticeTestCommand builds a typed chat-search command for notice tests.
+func newChatSearchNoticeTestCommand(t *testing.T, query string) *cobra.Command {
+	t.Helper()
+
+	cmd := &cobra.Command{Use: "test"}
+	for _, name := range []string{"query", "search-types", "member-ids", "sort-by", "page-token"} {
+		cmd.Flags().String(name, "", "")
+	}
+	for _, name := range []string{"is-manager", "disable-search-by-user", "exclude-muted"} {
+		cmd.Flags().Bool(name, false, "")
+	}
+	cmd.Flags().Int("page-size", 20, "")
+	if err := cmd.ParseFlags(nil); err != nil {
+		t.Fatalf("ParseFlags() error = %v", err)
+	}
+	if err := cmd.Flags().Set("query", query); err != nil {
+		t.Fatalf("Flags().Set(query) error = %v", err)
+	}
+	return cmd
+}
+
+// decodeShortcutData extracts the JSON envelope data object from shortcut output.
+func decodeShortcutData(t *testing.T, runtime *common.RuntimeContext) map[string]interface{} {
+	t.Helper()
+
+	out, ok := runtime.Factory.IOStreams.Out.(*bytes.Buffer)
+	if !ok {
+		t.Fatalf("stdout buffer has type %T", runtime.Factory.IOStreams.Out)
+	}
+	var env map[string]interface{}
+	if err := json.Unmarshal(out.Bytes(), &env); err != nil {
+		t.Fatalf("json.Unmarshal(stdout) error = %v\nstdout=%s", err, out.String())
+	}
+	data, ok := env["data"].(map[string]interface{})
+	if !ok {
+		t.Fatalf("envelope data missing or wrong type: %#v", env)
+	}
+	return data
+}

shortcuts/mail/mail_triage.go

@@ -159,6 +159,7 @@ var MailTriage = common.Shortcut{
 		var messages []map[string]interface{}
 		var hasMore bool
 		var nextPageToken string
+		var notice string
 
 		useSearch, err := resolveTriagePath(parsed, query, filter)
 		if err != nil {
@@ -189,6 +190,9 @@ var MailTriage = common.Shortcut{
 				if err != nil {
 					return err
 				}
+				if notice == "" {
+					notice, _ = searchData["notice"].(string)
+				}
 				pageMessages := buildTriageMessagesFromSearchItems(searchData["items"])
 				messages = append(messages, pageMessages...)
 				pageHasMore, _ := searchData["has_more"].(bool)
@@ -282,8 +286,14 @@ var MailTriage = common.Shortcut{
 				"has_more":   hasMore,
 				"page_token": nextPageToken,
 			}
+			if notice != "" {
+				outData["notice"] = notice
+			}
 			output.PrintJson(runtime.IO().Out, outData)
 		default: // "table"
+			if notice != "" {
+				fmt.Fprintf(runtime.IO().ErrOut, "notice: %s\n", notice)
+			}
 			if len(messages) == 0 {
 				fmt.Fprintln(runtime.IO().ErrOut, "No messages found.")
 				return nil

shortcuts/mail/mail_triage_test.go

@@ -1478,14 +1478,16 @@ func boolPtr(v bool) *bool { return &v }
 
 // --- mailbox_id preservation tests ---
 
+// TestMailTriageStructuredOutputPreservesMailboxID verifies mailbox and notice metadata.
 func TestMailTriageStructuredOutputPreservesMailboxID(t *testing.T) {
 	tests := []struct {
-		name      string
-		mailbox   string
-		format    string
-		args      []string
-		register  func(*httpmock.Registry, string)
-		wantCount int
+		name       string
+		mailbox    string
+		format     string
+		args       []string
+		register   func(*httpmock.Registry, string)
+		wantCount  int
+		wantNotice string
 	}{
 		{
 			name:    "list json default mailbox",
@@ -1522,9 +1524,10 @@ func TestMailTriageStructuredOutputPreservesMailboxID(t *testing.T) {
 			register: func(reg *httpmock.Registry, mailbox string) {
 				registerMailTriageSearchStub(reg, mailbox, []interface{}{
 					mailTriageSearchItem("search_pub_001", "Shared search"),
-				}, false, "")
+				}, false, "", "The query is too long and has been truncated to the first 50 characters for search.")
 			},
-			wantCount: 1,
+			wantCount:  1,
+			wantNotice: "The query is too long and has been truncated to the first 50 characters for search.",
 		},
 		{
 			name:    "empty list json keeps top-level mailbox",
@@ -1559,6 +1562,9 @@ func TestMailTriageStructuredOutputPreservesMailboxID(t *testing.T) {
 			if data["mailbox_id"] != tt.mailbox {
 				t.Fatalf("top-level mailbox_id mismatch: got %v, want %q", data["mailbox_id"], tt.mailbox)
 			}
+			if tt.wantNotice != "" && data["notice"] != tt.wantNotice {
+				t.Fatalf("notice mismatch: got %v, want %q", data["notice"], tt.wantNotice)
+			}
 			messages := mailTriageMessagesFromOutput(t, data)
 			if len(messages) != tt.wantCount {
 				t.Fatalf("message count mismatch: got %d, want %d", len(messages), tt.wantCount)
@@ -1572,6 +1578,7 @@ func TestMailTriageStructuredOutputPreservesMailboxID(t *testing.T) {
 	}
 }
 
+// TestMailTriageMissingMessageMetadataStillGetsMailboxID verifies fallback rows keep mailbox IDs.
 func TestMailTriageMissingMessageMetadataStillGetsMailboxID(t *testing.T) {
 	f, stdout, _, reg := mailShortcutTestFactory(t)
 	defer reg.Verify(t)
@@ -1604,6 +1611,7 @@ func TestMailTriageMissingMessageMetadataStillGetsMailboxID(t *testing.T) {
 	}
 }
 
+// TestMailTriageTableOutputPreservesMailboxContext verifies public mailbox table hints.
 func TestMailTriageTableOutputPreservesMailboxContext(t *testing.T) {
 	tests := []struct {
 		name              string
@@ -1654,6 +1662,33 @@ func TestMailTriageTableOutputPreservesMailboxContext(t *testing.T) {
 	}
 }
 
+// TestMailTriageDefaultTableOutputPrintsSearchNoticeToStderr verifies stderr notices.
+func TestMailTriageDefaultTableOutputPrintsSearchNoticeToStderr(t *testing.T) {
+	const notice = "The query is too long and has been truncated to the first 50 characters for search."
+
+	f, stdout, stderr, reg := mailShortcutTestFactory(t)
+	defer reg.Verify(t)
+
+	registerMailTriageSearchStub(reg, "me", []interface{}{
+		mailTriageSearchItem("msg_search_notice", "Search notice result"),
+	}, false, "", notice)
+
+	if err := runMountedMailShortcut(t, MailTriage, []string{
+		"+triage",
+		"--query", strings.Repeat("q", 81),
+	}, f, stdout); err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+
+	if out := stdout.String(); !strings.Contains(out, "msg_search_notice") {
+		t.Fatalf("stdout should contain table row, got:\n%s", out)
+	}
+	if errOut := stderr.String(); !strings.Contains(errOut, "notice: "+notice) {
+		t.Fatalf("stderr should contain search notice, got:\n%s", errOut)
+	}
+}
+
+// decodeMailTriageJSONOutput decodes structured triage output for assertions.
 func decodeMailTriageJSONOutput(t *testing.T, stdout interface{ Bytes() []byte }) map[string]interface{} {
 	t.Helper()
 	var data map[string]interface{}
@@ -1663,6 +1698,7 @@ func decodeMailTriageJSONOutput(t *testing.T, stdout interface{ Bytes() []byte }
 	return data
 }
 
+// mailTriageMessagesFromOutput extracts triage messages as object maps.
 func mailTriageMessagesFromOutput(t *testing.T, data map[string]interface{}) []map[string]interface{} {
 	t.Helper()
 	rawMessages, ok := data["messages"].([]interface{})
@@ -1715,7 +1751,8 @@ func registerMailTriageBatchStub(reg *httpmock.Registry, mailbox string, message
 	})
 }
 
-func registerMailTriageSearchStub(reg *httpmock.Registry, mailbox string, items []interface{}, hasMore bool, pageToken string) {
+// registerMailTriageSearchStub registers a mailbox search response for triage tests.
+func registerMailTriageSearchStub(reg *httpmock.Registry, mailbox string, items []interface{}, hasMore bool, pageToken string, notices ...string) {
 	data := map[string]interface{}{
 		"items":    items,
 		"has_more": hasMore,
@@ -1723,6 +1760,9 @@ func registerMailTriageSearchStub(reg *httpmock.Registry, mailbox string, items
 	if pageToken != "" {
 		data["page_token"] = pageToken
 	}
+	if len(notices) > 0 && notices[0] != "" {
+		data["notice"] = notices[0]
+	}
 	reg.Register(&httpmock.Stub{
 		Method: "POST",
 		URL:    mailboxPath(mailbox, "search"),

shortcuts/minutes/minutes_search.go

@@ -308,6 +308,9 @@ var MinutesSearch = common.Shortcut{
 			"has_more":   data["has_more"],
 			"page_token": data["page_token"],
 		}
+		if notice, _ := data["notice"].(string); notice != "" {
+			outData["notice"] = notice
+		}
 
 		runtime.OutFormat(outData, &output.Meta{Count: len(rows)}, func(w io.Writer) {
 			if len(rows) == 0 {

shortcuts/minutes/minutes_search_test.go

@@ -609,6 +609,8 @@ func TestMinutesSearchExecuteShowsPaginationHintForTableFormat(t *testing.T) {
 func TestMinutesSearchExecuteJSONCountUsesRenderedRows(t *testing.T) {
 	t.Parallel()
 
+	const notice = "The query is too long and has been truncated to the first 50 characters for search."
+
 	f, stdout, _, reg := cmdutil.TestFactory(t, defaultConfig())
 	reg.Register(&httpmock.Stub{
 		Method: "POST",
@@ -617,6 +619,7 @@ func TestMinutesSearchExecuteJSONCountUsesRenderedRows(t *testing.T) {
 			"code": 0,
 			"msg":  "ok",
 			"data": map[string]interface{}{
+				"notice": notice,
 				"items": []interface{}{
 					nil,
 					map[string]interface{}{
@@ -641,6 +644,9 @@ func TestMinutesSearchExecuteJSONCountUsesRenderedRows(t *testing.T) {
 	reg.Verify(t)
 
 	var envelope struct {
+		Data struct {
+			Notice string `json:"notice"`
+		} `json:"data"`
 		Meta struct {
 			Count int `json:"count"`
 		} `json:"meta"`
@@ -651,6 +657,9 @@ func TestMinutesSearchExecuteJSONCountUsesRenderedRows(t *testing.T) {
 	if envelope.Meta.Count != 1 {
 		t.Fatalf("meta.count = %d, want 1", envelope.Meta.Count)
 	}
+	if envelope.Data.Notice != notice {
+		t.Fatalf("data.notice = %q, want %q", envelope.Data.Notice, notice)
+	}
 }
 
 // TestMinuteSearchFieldExtractors verifies field extractors read populated metadata correctly.

shortcuts/okr/okr_batch_create.go

@@ -0,0 +1,385 @@
+// Copyright (c) 2026 Lark Technologies Pte. Ltd.
+// SPDX-License-Identifier: MIT
+
+package okr
+
+import (
+	"context"
+	"encoding/json"
+	"fmt"
+	"io"
+	"strconv"
+	"strings"
+	"time"
+
+	"github.com/larksuite/cli/errs"
+	"github.com/larksuite/cli/shortcuts/common"
+)
+
+// batchCreateKR represents a key result in the batch create input.
+type batchCreateKR struct {
+	Text    string   `json:"text"`
+	Mention []string `json:"mention,omitempty"`
+}
+
+// batchCreateObjective represents an objective in the batch create input.
+type batchCreateObjective struct {
+	Text    string          `json:"text"`
+	Mention []string        `json:"mention,omitempty"`
+	KRs     []batchCreateKR `json:"krs,omitempty"`
+}
+
+// createdObjective tracks a created objective and its KR IDs for output.
+// KRs are automatically deleted by the backend when the objective is deleted (no need to delete them separately during rollback).
+type createdObjective struct {
+	ObjectiveID string
+	KRIDs       []string // for output response only, not used in rollback
+}
+
+// parseBatchCreateInput parses and validates the JSON input.
+func parseBatchCreateInput(input string) ([]batchCreateObjective, error) {
+	var objectives []batchCreateObjective
+	if err := json.Unmarshal([]byte(input), &objectives); err != nil {
+		return nil, errs.NewValidationError(errs.SubtypeInvalidArgument, "--input must be valid JSON array: %s", err).WithParam("--input").WithCause(err)
+	}
+	if len(objectives) == 0 {
+		return nil, errs.NewValidationError(errs.SubtypeInvalidArgument, "--input must contain at least one objective").WithParam("--input")
+	}
+	for i, obj := range objectives {
+		if strings.TrimSpace(obj.Text) == "" {
+			return nil, errs.NewValidationError(errs.SubtypeInvalidArgument, "objective[%d].text is required and cannot be empty", i).WithParam("--input")
+		}
+		for j, kr := range obj.KRs {
+			if strings.TrimSpace(kr.Text) == "" {
+				return nil, errs.NewValidationError(errs.SubtypeInvalidArgument, "objective[%d].krs[%d].text is required and cannot be empty", i, j).WithParam("--input")
+			}
+		}
+	}
+	return objectives, nil
+}
+
+// buildContentBlock converts text and mentions to a ContentBlock.
+func buildContentBlock(text string, mentions []string) *ContentBlock {
+	elements := make([]ContentParagraphElement, 0, len(mentions)+1)
+
+	// Add text element
+	textElem := ContentParagraphElement{
+		ParagraphElementType: ParagraphElementTypeTextRun.Ptr(),
+		TextRun: &ContentTextRun{
+			Text: &text,
+		},
+	}
+	elements = append(elements, textElem)
+
+	// Add mention elements
+	for _, mention := range mentions {
+		mentionElem := ContentParagraphElement{
+			ParagraphElementType: ParagraphElementTypeMention.Ptr(),
+			Mention: &ContentMention{
+				UserID: &mention,
+			},
+		}
+		elements = append(elements, mentionElem)
+	}
+
+	return &ContentBlock{
+		Blocks: []ContentBlockElement{
+			{
+				BlockElementType: BlockElementTypeParagraph.Ptr(),
+				Paragraph: &ContentParagraph{
+					Elements: elements,
+				},
+			},
+		},
+	}
+}
+
+// createObjective calls the API to create an objective.
+func createObjective(ctx context.Context, runtime *common.RuntimeContext, cycleID, userIDType string, obj batchCreateObjective) (string, error) {
+	content := buildContentBlock(obj.Text, obj.Mention)
+	body := map[string]interface{}{
+		"content": content,
+	}
+	queryParams := map[string]interface{}{
+		"cycle_id":     cycleID,
+		"user_id_type": userIDType,
+	}
+
+	path := fmt.Sprintf("/open-apis/okr/v2/cycles/%s/objectives", cycleID)
+	data, err := runtime.CallAPITyped("POST", path, queryParams, body)
+	if err != nil {
+		return "", wrapOkrNetworkErr(err, "failed to create objective")
+	}
+
+	objectiveID, ok := data["objective_id"].(string)
+	if !ok {
+		return "", errs.NewInternalError(errs.SubtypeUnknown, "create objective response missing objective_id")
+	}
+	return objectiveID, nil
+}
+
+// createKR calls the API to create a key result.
+func createKR(ctx context.Context, runtime *common.RuntimeContext, objectiveID, userIDType string, kr batchCreateKR) (string, error) {
+	content := buildContentBlock(kr.Text, kr.Mention)
+	body := map[string]interface{}{
+		"content": content,
+	}
+	queryParams := map[string]interface{}{
+		"objective_id": objectiveID,
+		"user_id_type": userIDType,
+	}
+
+	path := fmt.Sprintf("/open-apis/okr/v2/objectives/%s/key_results", objectiveID)
+	data, err := runtime.CallAPITyped("POST", path, queryParams, body)
+	if err != nil {
+		return "", wrapOkrNetworkErr(err, "failed to create key result")
+	}
+
+	krID, ok := data["key_result_id"].(string)
+	if !ok {
+		return "", errs.NewInternalError(errs.SubtypeUnknown, "create key result response missing key_result_id")
+	}
+	return krID, nil
+}
+
+// deleteObjective deletes an objective (used for rollback).
+func deleteObjective(ctx context.Context, runtime *common.RuntimeContext, objectiveID string) error {
+	queryParams := map[string]interface{}{
+		"objective_id": objectiveID,
+		"yes":          true,
+	}
+	path := fmt.Sprintf("/open-apis/okr/v2/objectives/%s", objectiveID)
+	_, err := runtime.CallAPITyped("DELETE", path, queryParams, nil)
+	if err != nil {
+		return wrapOkrNetworkErr(err, "failed to delete objective %s during rollback", objectiveID)
+	}
+	return nil
+}
+
+// rollback deletes created objectives in reverse order.
+// KRs are automatically deleted by the backend when the objective is deleted.
+func rollback(ctx context.Context, runtime *common.RuntimeContext, created []createdObjective) []error {
+	var errsList []error
+
+	// Iterate in reverse order
+	for i := len(created) - 1; i >= 0; i-- {
+		obj := created[i]
+
+		// Delete the objective (backend automatically deletes its KRs)
+		if err := deleteObjective(ctx, runtime, obj.ObjectiveID); err != nil {
+			//nolint:forbidigo // intermediate wrap for rollback error collection; final error is typed via buildRollbackError
+			errsList = append(errsList, fmt.Errorf("objective %s: %w", obj.ObjectiveID, err))
+		}
+
+		// Rate limiting between deletions
+		if i > 0 {
+			time.Sleep(500 * time.Millisecond)
+		}
+	}
+
+	return errsList
+}
+
+// OKRBatchCreate batch creates objectives and their key results.
+var OKRBatchCreate = common.Shortcut{
+	Service:     "okr",
+	Command:     "+batch-create",
+	Description: "Batch create OKR objectives and key results with rollback on failure",
+	Risk:        "write",
+	Scopes:      []string{"okr:okr.content:writeonly"},
+	AuthTypes:   []string{"user", "bot"},
+	HasFormat:   true,
+	Flags: []common.Flag{
+		{Name: "cycle-id", Desc: "OKR cycle ID (int64)", Required: true},
+		{Name: "input", Desc: "JSON array of objectives: [{\"text\":\"...\",\"mention\":[\"...\"],\"krs\":[{\"text\":\"...\",\"mention\":[\"...\"]}]}]", Input: []string{common.File, common.Stdin}, Required: true},
+		{Name: "user-id-type", Default: "open_id", Desc: "user ID type: open_id | union_id | user_id"},
+	},
+	Validate: func(ctx context.Context, runtime *common.RuntimeContext) error {
+		cycleID := runtime.Str("cycle-id")
+		if id, err := strconv.ParseInt(cycleID, 10, 64); err != nil || id <= 0 {
+			return errs.NewValidationError(errs.SubtypeInvalidArgument, "--cycle-id must be a positive int64").WithParam("--cycle-id")
+		}
+
+		input := runtime.Str("input")
+		if err := common.RejectDangerousCharsTyped("--input", input); err != nil {
+			return err
+		}
+		if _, err := parseBatchCreateInput(input); err != nil {
+			return err
+		}
+
+		idType := runtime.Str("user-id-type")
+		if idType != "open_id" && idType != "union_id" && idType != "user_id" {
+			return errs.NewValidationError(errs.SubtypeInvalidArgument, "--user-id-type must be one of: open_id | union_id | user_id").WithParam("--user-id-type")
+		}
+
+		return nil
+	},
+	DryRun: func(ctx context.Context, runtime *common.RuntimeContext) *common.DryRunAPI {
+		cycleID := runtime.Str("cycle-id")
+		userIDType := runtime.Str("user-id-type")
+		objectives, _ := parseBatchCreateInput(runtime.Str("input"))
+
+		apis := common.NewDryRunAPI()
+
+		for i, obj := range objectives {
+			// Objective creation
+			objContent := buildContentBlock(obj.Text, obj.Mention)
+			objBody := map[string]interface{}{
+				"content": objContent,
+			}
+			objParams := map[string]interface{}{
+				"cycle_id":     cycleID,
+				"user_id_type": userIDType,
+			}
+			objPath := fmt.Sprintf("/open-apis/okr/v2/cycles/%s/objectives", cycleID)
+			apis = apis.
+				POST(objPath).
+				Params(objParams).
+				Body(objBody).
+				Desc(fmt.Sprintf("Create objective[%d]: %s", i, obj.Text))
+
+			// KR creations
+			for j, kr := range obj.KRs {
+				krContent := buildContentBlock(kr.Text, kr.Mention)
+				krBody := map[string]interface{}{
+					"content": krContent,
+				}
+				krParams := map[string]interface{}{
+					"objective_id": "<objective_id_from_previous_call>",
+					"user_id_type": userIDType,
+				}
+				krPath := "/open-apis/okr/v2/objectives/<objective_id>/key_results"
+				apis = apis.
+					POST(krPath).
+					Params(krParams).
+					Body(krBody).
+					Desc(fmt.Sprintf("Create objective[%d].krs[%d]: %s", i, j, kr.Text))
+			}
+		}
+
+		return apis
+	},
+	Execute: func(ctx context.Context, runtime *common.RuntimeContext) error {
+		cycleID := runtime.Str("cycle-id")
+		userIDType := runtime.Str("user-id-type")
+		objectives, err := parseBatchCreateInput(runtime.Str("input"))
+		if err != nil {
+			return err
+		}
+
+		var created []createdObjective
+
+		for i, obj := range objectives {
+			// Rate limiting between objectives
+			if i > 0 {
+				time.Sleep(500 * time.Millisecond)
+			}
+
+			// Create objective
+			objectiveID, err := createObjective(ctx, runtime, cycleID, userIDType, obj)
+			if err != nil {
+				if len(created) == 0 {
+					return err
+				}
+				rollbackErrs := rollback(ctx, runtime, created)
+				return buildRollbackError(err, rollbackErrs, created)
+			}
+
+			createdObj := createdObjective{
+				ObjectiveID: objectiveID,
+			}
+
+			// Create KRs
+			for j, kr := range obj.KRs {
+				// Rate limiting between KRs
+				if j > 0 {
+					time.Sleep(500 * time.Millisecond)
+				}
+
+				krID, err := createKR(ctx, runtime, objectiveID, userIDType, kr)
+				if err != nil {
+					created = append(created, createdObj)
+					rollbackErrs := rollback(ctx, runtime, created)
+					return buildRollbackError(err, rollbackErrs, created)
+				}
+
+				createdObj.KRIDs = append(createdObj.KRIDs, krID)
+			}
+
+			created = append(created, createdObj)
+		}
+
+		// Build response
+		respCreated := make([]map[string]interface{}, 0, len(created))
+		for _, obj := range created {
+			respCreated = append(respCreated, map[string]interface{}{
+				"objective_id": obj.ObjectiveID,
+				"krs":          obj.KRIDs,
+			})
+		}
+
+		result := map[string]interface{}{
+			"ok":   true,
+			"data": map[string]interface{}{"created": respCreated},
+		}
+
+		runtime.OutFormat(result, nil, func(w io.Writer) {
+			fmt.Fprintf(w, "Successfully created %d objective(s)\n", len(created))
+			for i, obj := range created {
+				fmt.Fprintf(w, "Objective[%d] ID: %s (%d KR(s))\n", i, obj.ObjectiveID, len(obj.KRIDs))
+				for j, krID := range obj.KRIDs {
+					fmt.Fprintf(w, "  KR[%d] ID: %s\n", j, krID)
+				}
+			}
+		})
+
+		return nil
+	},
+}
+
+// buildRollbackError constructs an error that includes both the original failure
+// and any rollback failures, with a list of residual IDs that could not be cleaned up.
+// KRs are automatically deleted by the backend when the objective is deleted, so we only
+// need to track objective IDs for residual cleanup.
+func buildRollbackError(originalErr error, rollbackErrs []error, created []createdObjective) error {
+	var residualIDs []string
+
+	// Only collect residual IDs when rollback had failures
+	// If rollback succeeded (len(rollbackErrs) == 0), all objectives were deleted
+	if len(rollbackErrs) > 0 {
+		for _, obj := range created {
+			residualIDs = append(residualIDs, fmt.Sprintf("objective:%s", obj.ObjectiveID))
+		}
+	}
+
+	msg := fmt.Sprintf("batch create failed, rolling back: %v", originalErr)
+	if len(rollbackErrs) > 0 {
+		var rollbackMsgs []string
+		for _, e := range rollbackErrs {
+			rollbackMsgs = append(rollbackMsgs, e.Error())
+		}
+		msg += fmt.Sprintf("; rollback also had %d failure(s): %s", len(rollbackErrs), strings.Join(rollbackMsgs, "; "))
+	}
+	if len(residualIDs) > 0 {
+		msg += fmt.Sprintf("; residual objectives that may need manual cleanup (KRs auto-deleted with objective): %s", strings.Join(residualIDs, ", "))
+	}
+
+	// Preserve the original error's type information if it's already a typed error
+	if prob, ok := errs.ProblemOf(originalErr); ok {
+		switch prob.Category {
+		case errs.CategoryAPI:
+			return errs.NewAPIError(prob.Subtype, "%s", msg).WithCause(originalErr)
+		case errs.CategoryNetwork:
+			return errs.NewNetworkError(prob.Subtype, "%s", msg).WithCause(originalErr)
+		case errs.CategoryValidation:
+			return errs.NewValidationError(prob.Subtype, "%s", msg).WithCause(originalErr)
+		case errs.CategoryInternal:
+			return errs.NewInternalError(prob.Subtype, "%s", msg).WithCause(originalErr)
+		default:
+			return errs.NewInternalError(prob.Subtype, "%s", msg).WithCause(originalErr)
+		}
+	}
+
+	return errs.NewInternalError(errs.SubtypeUnknown, "%s", msg).WithCause(originalErr)
+}

shortcuts/okr/okr_batch_create_test.go

@@ -0,0 +1,593 @@
+// Copyright (c) 2026 Lark Technologies Pte. Ltd.
+// SPDX-License-Identifier: MIT
+
+package okr
+
+import (
+	"bytes"
+	"errors"
+	"strings"
+	"testing"
+
+	"github.com/larksuite/cli/errs"
+	"github.com/larksuite/cli/internal/cmdutil"
+	"github.com/larksuite/cli/internal/core"
+	"github.com/larksuite/cli/internal/httpmock"
+	"github.com/spf13/cobra"
+)
+
+func batchCreateTestConfig(t *testing.T) *core.CliConfig {
+	t.Helper()
+	return &core.CliConfig{
+		AppID:     "test-okr-batch-create",
+		AppSecret: "secret-okr-batch-create",
+		Brand:     core.BrandFeishu,
+	}
+}
+
+func runBatchCreateShortcut(t *testing.T, f *cmdutil.Factory, stdout *bytes.Buffer, args []string) error {
+	t.Helper()
+	parent := &cobra.Command{Use: "okr"}
+	OKRBatchCreate.Mount(parent, f)
+	parent.SetArgs(args)
+	parent.SilenceErrors = true
+	parent.SilenceUsage = true
+	if stdout != nil {
+		stdout.Reset()
+	}
+	return parent.Execute()
+}
+
+const validBatchCreateInput = `[
+  {"text":"Objective 1","mention":["ou_123"],"krs":[{"text":"KR 1.1","mention":["ou_456"]}]},
+  {"text":"Objective 2","krs":[{"text":"KR 2.1"},{"text":"KR 2.2"}]}
+]`
+
+// --- Validate tests ---
+
+func TestBatchCreateValidate_MissingCycleID(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, _ := cmdutil.TestFactory(t, batchCreateTestConfig(t))
+	err := runBatchCreateShortcut(t, f, stdout, []string{
+		"+batch-create",
+		"--input", validBatchCreateInput,
+	})
+	// cobra Required:true reports flag name without "--" prefix
+	if err == nil || !strings.Contains(err.Error(), "cycle-id") {
+		t.Fatalf("expected --cycle-id required error, got: %v", err)
+	}
+}
+
+func TestBatchCreateValidate_InvalidCycleID(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, _ := cmdutil.TestFactory(t, batchCreateTestConfig(t))
+	err := runBatchCreateShortcut(t, f, stdout, []string{
+		"+batch-create",
+		"--cycle-id", "abc",
+		"--input", validBatchCreateInput,
+	})
+	if err == nil {
+		t.Fatal("expected error for invalid --cycle-id")
+	}
+	_, ok := errs.ProblemOf(err)
+	if !ok {
+		t.Fatalf("expected typed error, got: %v", err)
+	}
+	validationErr, ok := err.(*errs.ValidationError)
+	if !ok {
+		t.Fatalf("expected ValidationError, got: %T", err)
+	}
+	if validationErr.Param != "--cycle-id" {
+		t.Fatalf("expected param --cycle-id, got %q", validationErr.Param)
+	}
+}
+
+func TestBatchCreateValidate_MissingInput(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, _ := cmdutil.TestFactory(t, batchCreateTestConfig(t))
+	err := runBatchCreateShortcut(t, f, stdout, []string{
+		"+batch-create",
+		"--cycle-id", "123",
+	})
+	// cobra Required:true reports flag name without "--" prefix
+	if err == nil || !strings.Contains(err.Error(), "input") {
+		t.Fatalf("expected --input required error, got: %v", err)
+	}
+}
+
+func TestBatchCreateValidate_InvalidInputJSON(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, _ := cmdutil.TestFactory(t, batchCreateTestConfig(t))
+	err := runBatchCreateShortcut(t, f, stdout, []string{
+		"+batch-create",
+		"--cycle-id", "123",
+		"--input", "not-json",
+	})
+	if err == nil {
+		t.Fatal("expected error for invalid --input JSON")
+	}
+	_, ok := errs.ProblemOf(err)
+	if !ok {
+		t.Fatalf("expected typed error, got: %v", err)
+	}
+	validationErr, ok := err.(*errs.ValidationError)
+	if !ok {
+		t.Fatalf("expected ValidationError, got: %T", err)
+	}
+	if validationErr.Param != "--input" {
+		t.Fatalf("expected param --input, got %q", validationErr.Param)
+	}
+}
+
+func TestBatchCreateValidate_EmptyInputArray(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, _ := cmdutil.TestFactory(t, batchCreateTestConfig(t))
+	err := runBatchCreateShortcut(t, f, stdout, []string{
+		"+batch-create",
+		"--cycle-id", "123",
+		"--input", "[]",
+	})
+	if err == nil {
+		t.Fatal("expected error for empty --input array")
+	}
+	_, ok := errs.ProblemOf(err)
+	if !ok {
+		t.Fatalf("expected typed error, got: %v", err)
+	}
+	validationErr, ok := err.(*errs.ValidationError)
+	if !ok {
+		t.Fatalf("expected ValidationError, got: %T", err)
+	}
+	if validationErr.Param != "--input" {
+		t.Fatalf("expected param --input, got %q", validationErr.Param)
+	}
+}
+
+func TestBatchCreateValidate_EmptyObjectiveText(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, _ := cmdutil.TestFactory(t, batchCreateTestConfig(t))
+	err := runBatchCreateShortcut(t, f, stdout, []string{
+		"+batch-create",
+		"--cycle-id", "123",
+		"--input", `[{"text":"","krs":[{"text":"KR 1"}]}]`,
+	})
+	if err == nil {
+		t.Fatal("expected error for empty objective text")
+	}
+	_, ok := errs.ProblemOf(err)
+	if !ok {
+		t.Fatalf("expected typed error, got: %v", err)
+	}
+	validationErr, ok := err.(*errs.ValidationError)
+	if !ok {
+		t.Fatalf("expected ValidationError, got: %T", err)
+	}
+	if validationErr.Param != "--input" {
+		t.Fatalf("expected param --input, got %q", validationErr.Param)
+	}
+	if !strings.Contains(err.Error(), "objective[0].text") {
+		t.Fatalf("expected error to mention objective[0].text, got: %v", err)
+	}
+}
+
+func TestBatchCreateValidate_EmptyKRText(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, _ := cmdutil.TestFactory(t, batchCreateTestConfig(t))
+	err := runBatchCreateShortcut(t, f, stdout, []string{
+		"+batch-create",
+		"--cycle-id", "123",
+		"--input", `[{"text":"Obj 1","krs":[{"text":""}]}]`,
+	})
+	if err == nil {
+		t.Fatal("expected error for empty KR text")
+	}
+	_, ok := errs.ProblemOf(err)
+	if !ok {
+		t.Fatalf("expected typed error, got: %v", err)
+	}
+	validationErr, ok := err.(*errs.ValidationError)
+	if !ok {
+		t.Fatalf("expected ValidationError, got: %T", err)
+	}
+	if validationErr.Param != "--input" {
+		t.Fatalf("expected param --input, got %q", validationErr.Param)
+	}
+	if !strings.Contains(err.Error(), "objective[0].krs[0].text") {
+		t.Fatalf("expected error to mention objective[0].krs[0].text, got: %v", err)
+	}
+}
+
+func TestBatchCreateValidate_InvalidUserIDType(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, _ := cmdutil.TestFactory(t, batchCreateTestConfig(t))
+	err := runBatchCreateShortcut(t, f, stdout, []string{
+		"+batch-create",
+		"--cycle-id", "123",
+		"--input", validBatchCreateInput,
+		"--user-id-type", "invalid",
+	})
+	if err == nil {
+		t.Fatal("expected error for invalid --user-id-type")
+	}
+	_, ok := errs.ProblemOf(err)
+	if !ok {
+		t.Fatalf("expected typed error, got: %v", err)
+	}
+	validationErr, ok := err.(*errs.ValidationError)
+	if !ok {
+		t.Fatalf("expected ValidationError, got: %T", err)
+	}
+	if validationErr.Param != "--user-id-type" {
+		t.Fatalf("expected param --user-id-type, got %q", validationErr.Param)
+	}
+}
+
+func TestBatchCreateValidate_Valid(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, reg := cmdutil.TestFactory(t, batchCreateTestConfig(t))
+	reg.Register(&httpmock.Stub{
+		Method: "POST",
+		URL:    "/open-apis/okr/v2/cycles/123/objectives",
+		Body: map[string]interface{}{
+			"code": 0,
+			"msg":  "ok",
+			"data": map[string]interface{}{
+				"objective_id": "100",
+			},
+		},
+	})
+	reg.Register(&httpmock.Stub{
+		Method: "POST",
+		URL:    "/open-apis/okr/v2/objectives/100/key_results",
+		Body: map[string]interface{}{
+			"code": 0,
+			"msg":  "ok",
+			"data": map[string]interface{}{
+				"key_result_id": "200",
+			},
+		},
+	})
+	reg.Register(&httpmock.Stub{
+		Method: "POST",
+		URL:    "/open-apis/okr/v2/cycles/123/objectives",
+		Body: map[string]interface{}{
+			"code": 0,
+			"msg":  "ok",
+			"data": map[string]interface{}{
+				"objective_id": "101",
+			},
+		},
+	})
+	reg.Register(&httpmock.Stub{
+		Method: "POST",
+		URL:    "/open-apis/okr/v2/objectives/101/key_results",
+		Body: map[string]interface{}{
+			"code": 0,
+			"msg":  "ok",
+			"data": map[string]interface{}{
+				"key_result_id": "201",
+			},
+		},
+	})
+	reg.Register(&httpmock.Stub{
+		Method: "POST",
+		URL:    "/open-apis/okr/v2/objectives/101/key_results",
+		Body: map[string]interface{}{
+			"code": 0,
+			"msg":  "ok",
+			"data": map[string]interface{}{
+				"key_result_id": "202",
+			},
+		},
+	})
+	err := runBatchCreateShortcut(t, f, stdout, []string{
+		"+batch-create",
+		"--cycle-id", "123",
+		"--input", validBatchCreateInput,
+	})
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+}
+
+// --- DryRun tests ---
+
+func TestBatchCreateDryRun(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, _ := cmdutil.TestFactory(t, batchCreateTestConfig(t))
+	err := runBatchCreateShortcut(t, f, stdout, []string{
+		"+batch-create",
+		"--cycle-id", "123",
+		"--input", validBatchCreateInput,
+		"--dry-run",
+	})
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	output := stdout.String()
+	if !strings.Contains(output, "/open-apis/okr/v2/cycles/123/objectives") {
+		t.Fatalf("dry-run output should contain objective creation API path, got: %s", output)
+	}
+	if !strings.Contains(output, "POST") {
+		t.Fatalf("dry-run output should contain POST method, got: %s", output)
+	}
+	if !strings.Contains(output, "/open-apis/okr/v2/objectives/") || !strings.Contains(output, "/key_results") {
+		t.Fatalf("dry-run output should contain KR creation API path, got: %s", output)
+	}
+	// Verify content is in the body
+	if !strings.Contains(output, "Objective 1") {
+		t.Fatalf("dry-run output should contain objective text, got: %s", output)
+	}
+	if !strings.Contains(output, "KR 1.1") {
+		t.Fatalf("dry-run output should contain KR text, got: %s", output)
+	}
+}
+
+// --- Execute tests ---
+
+func TestBatchCreateExecute_Success(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, reg := cmdutil.TestFactory(t, batchCreateTestConfig(t))
+	reg.Register(&httpmock.Stub{
+		Method: "POST",
+		URL:    "/open-apis/okr/v2/cycles/123/objectives",
+		Body: map[string]interface{}{
+			"code": 0,
+			"msg":  "ok",
+			"data": map[string]interface{}{
+				"objective_id": "100",
+			},
+		},
+	})
+	reg.Register(&httpmock.Stub{
+		Method: "POST",
+		URL:    "/open-apis/okr/v2/objectives/100/key_results",
+		Body: map[string]interface{}{
+			"code": 0,
+			"msg":  "ok",
+			"data": map[string]interface{}{
+				"key_result_id": "200",
+			},
+		},
+	})
+	err := runBatchCreateShortcut(t, f, stdout, []string{
+		"+batch-create",
+		"--cycle-id", "123",
+		"--input", `[{"text":"Obj 1","krs":[{"text":"KR 1"}]}]`,
+	})
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+
+	// Verify output
+	data := decodeEnvelope(t, stdout)
+	ok, _ := data["ok"].(bool)
+	if !ok {
+		t.Fatal("expected ok=true in output")
+	}
+	dataField, _ := data["data"].(map[string]interface{})
+	created, _ := dataField["created"].([]interface{})
+	if len(created) != 1 {
+		t.Fatalf("expected 1 created objective, got %d", len(created))
+	}
+	obj, _ := created[0].(map[string]interface{})
+	if obj["objective_id"] != "100" {
+		t.Fatalf("expected objective_id=100, got %v", obj["objective_id"])
+	}
+	krs, _ := obj["krs"].([]interface{})
+	if len(krs) != 1 || krs[0] != "200" {
+		t.Fatalf("expected krs=[200], got %v", krs)
+	}
+}
+
+func TestBatchCreateExecute_APIErrorOnObjective(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, reg := cmdutil.TestFactory(t, batchCreateTestConfig(t))
+	reg.Register(&httpmock.Stub{
+		Method: "POST",
+		URL:    "/open-apis/okr/v2/cycles/123/objectives",
+		Status: 400,
+		Body: map[string]interface{}{
+			"code": 1001001,
+			"msg":  "invalid parameters",
+		},
+	})
+	err := runBatchCreateShortcut(t, f, stdout, []string{
+		"+batch-create",
+		"--cycle-id", "123",
+		"--input", `[{"text":"Obj 1"}]`,
+	})
+	if err == nil {
+		t.Fatal("expected error for API failure")
+	}
+	// Should be a typed error from the API
+	prob, ok := errs.ProblemOf(err)
+	if !ok {
+		t.Fatalf("expected typed error, got: %v", err)
+	}
+	if prob.Category != "api" {
+		t.Fatalf("expected api category, got %q", prob.Category)
+	}
+}
+
+func TestBatchCreateExecute_APIErrorOnKR_TriggersRollback(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, reg := cmdutil.TestFactory(t, batchCreateTestConfig(t))
+	// First objective creation succeeds
+	reg.Register(&httpmock.Stub{
+		Method: "POST",
+		URL:    "/open-apis/okr/v2/cycles/123/objectives",
+		Body: map[string]interface{}{
+			"code": 0,
+			"msg":  "ok",
+			"data": map[string]interface{}{
+				"objective_id": "100",
+			},
+		},
+	})
+	// KR creation fails
+	reg.Register(&httpmock.Stub{
+		Method: "POST",
+		URL:    "/open-apis/okr/v2/objectives/100/key_results",
+		Status: 400,
+		Body: map[string]interface{}{
+			"code": 1001001,
+			"msg":  "invalid parameters",
+		},
+	})
+	// Rollback: delete the created objective
+	reg.Register(&httpmock.Stub{
+		Method: "DELETE",
+		URL:    "/open-apis/okr/v2/objectives/100",
+		Body: map[string]interface{}{
+			"code": 0,
+			"msg":  "ok",
+			"data": map[string]interface{}{
+				"objective_id": "100",
+			},
+		},
+	})
+	err := runBatchCreateShortcut(t, f, stdout, []string{
+		"+batch-create",
+		"--cycle-id", "123",
+		"--input", `[{"text":"Obj 1","krs":[{"text":"KR 1"}]}]`,
+	})
+	if err == nil {
+		t.Fatal("expected error for KR creation failure")
+	}
+	// Error should mention rollback
+	if !strings.Contains(err.Error(), "rolling back") && !strings.Contains(err.Error(), "rollback") {
+		t.Fatalf("expected error to mention rollback, got: %v", err)
+	}
+	// Assert typed error metadata
+	prob, ok := errs.ProblemOf(err)
+	if !ok {
+		t.Fatalf("expected typed error, got: %v", err)
+	}
+	if prob.Category != errs.CategoryAPI {
+		t.Fatalf("expected api category (preserved from original error), got %q", prob.Category)
+	}
+	// Assert cause preservation
+	var apiErr *errs.APIError
+	if !errors.As(err, &apiErr) {
+		t.Fatalf("expected error to wrap APIError, got: %T", err)
+	}
+	if !errors.Is(err, apiErr) {
+		t.Fatal("expected errors.Is to find the wrapped APIError")
+	}
+}
+
+func TestBatchCreateExecute_RollbackDeleteFails(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, reg := cmdutil.TestFactory(t, batchCreateTestConfig(t))
+	// Objective creation succeeds
+	reg.Register(&httpmock.Stub{
+		Method: "POST",
+		URL:    "/open-apis/okr/v2/cycles/123/objectives",
+		Body: map[string]interface{}{
+			"code": 0,
+			"msg":  "ok",
+			"data": map[string]interface{}{
+				"objective_id": "100",
+			},
+		},
+	})
+	// KR creation fails
+	reg.Register(&httpmock.Stub{
+		Method: "POST",
+		URL:    "/open-apis/okr/v2/objectives/100/key_results",
+		Status: 400,
+		Body: map[string]interface{}{
+			"code": 1001001,
+			"msg":  "invalid parameters",
+		},
+	})
+	// Rollback delete also fails
+	reg.Register(&httpmock.Stub{
+		Method: "DELETE",
+		URL:    "/open-apis/okr/v2/objectives/100",
+		Status: 500,
+		Body: map[string]interface{}{
+			"code": 9999999,
+			"msg":  "internal error",
+		},
+	})
+	err := runBatchCreateShortcut(t, f, stdout, []string{
+		"+batch-create",
+		"--cycle-id", "123",
+		"--input", `[{"text":"Obj 1","krs":[{"text":"KR 1"}]}]`,
+	})
+	if err == nil {
+		t.Fatal("expected error for KR creation failure")
+	}
+	// Error should mention residual resources
+	if !strings.Contains(err.Error(), "residual") && !strings.Contains(err.Error(), "manual cleanup") {
+		t.Fatalf("expected error to mention residual resources, got: %v", err)
+	}
+	if !strings.Contains(err.Error(), "objective:100") {
+		t.Fatalf("expected error to list residual objective ID, got: %v", err)
+	}
+}
+
+// --- Unit tests for helper functions ---
+
+func TestParseBatchCreateInput_Valid(t *testing.T) {
+	t.Parallel()
+	input := `[{"text":"Obj 1","mention":["ou_123"],"krs":[{"text":"KR 1","mention":["ou_456"]}]}]`
+	objs, err := parseBatchCreateInput(input)
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	if len(objs) != 1 {
+		t.Fatalf("expected 1 objective, got %d", len(objs))
+	}
+	if objs[0].Text != "Obj 1" {
+		t.Fatalf("expected text 'Obj 1', got %q", objs[0].Text)
+	}
+	if len(objs[0].Mention) != 1 || objs[0].Mention[0] != "ou_123" {
+		t.Fatalf("expected mention ['ou_123'], got %v", objs[0].Mention)
+	}
+	if len(objs[0].KRs) != 1 {
+		t.Fatalf("expected 1 KR, got %d", len(objs[0].KRs))
+	}
+	if objs[0].KRs[0].Text != "KR 1" {
+		t.Fatalf("expected KR text 'KR 1', got %q", objs[0].KRs[0].Text)
+	}
+}
+
+func TestBuildContentBlock(t *testing.T) {
+	t.Parallel()
+	cb := buildContentBlock("Test text", []string{"ou_123", "ou_456"})
+	if cb == nil {
+		t.Fatal("expected non-nil ContentBlock")
+	}
+	if len(cb.Blocks) != 1 {
+		t.Fatalf("expected 1 block, got %d", len(cb.Blocks))
+	}
+	block := cb.Blocks[0]
+	if block.BlockElementType == nil || *block.BlockElementType != BlockElementTypeParagraph {
+		t.Fatalf("expected paragraph block type")
+	}
+	if block.Paragraph == nil {
+		t.Fatal("expected non-nil paragraph")
+	}
+	// Should have 3 elements: 1 text + 2 mentions
+	if len(block.Paragraph.Elements) != 3 {
+		t.Fatalf("expected 3 paragraph elements, got %d", len(block.Paragraph.Elements))
+	}
+	// First element should be textRun
+	if block.Paragraph.Elements[0].ParagraphElementType == nil ||
+		*block.Paragraph.Elements[0].ParagraphElementType != ParagraphElementTypeTextRun {
+		t.Fatal("expected first element to be textRun")
+	}
+	if block.Paragraph.Elements[0].TextRun == nil || *block.Paragraph.Elements[0].TextRun.Text != "Test text" {
+		t.Fatalf("expected text 'Test text', got %v", block.Paragraph.Elements[0].TextRun)
+	}
+	// Second and third should be mentions
+	for i := 1; i <= 2; i++ {
+		if block.Paragraph.Elements[i].ParagraphElementType == nil ||
+			*block.Paragraph.Elements[i].ParagraphElementType != ParagraphElementTypeMention {
+			t.Fatalf("expected element %d to be mention", i)
+		}
+	}
+}

shortcuts/okr/okr_indicator_update.go

@@ -0,0 +1,178 @@
+// Copyright (c) 2026 Lark Technologies Pte. Ltd.
+// SPDX-License-Identifier: MIT
+
+package okr
+
+import (
+	"context"
+	"fmt"
+	"io"
+	"math"
+	"strconv"
+
+	"github.com/larksuite/cli/errs"
+	"github.com/larksuite/cli/shortcuts/common"
+)
+
+// parseIndicatorValue parses and validates the indicator value.
+func parseIndicatorValue(valueStr string) (float64, error) {
+	value, err := strconv.ParseFloat(valueStr, 64)
+	if err != nil {
+		return 0, errs.NewValidationError(errs.SubtypeInvalidArgument, "--value must be a number between -99999999999 and 99999999999").WithParam("--value").WithCause(err)
+	}
+	if math.IsNaN(value) || math.IsInf(value, 0) || value < -99999999999 || value > 99999999999 {
+		return 0, errs.NewValidationError(errs.SubtypeInvalidArgument, "--value must be a number between -99999999999 and 99999999999").WithParam("--value")
+	}
+	return value, nil
+}
+
+// fetchIndicatorID fetches the indicator ID for an objective or key result.
+// The indicators.list API returns a single indicator object (not a list),
+// which always exists (may be a default empty indicator).
+func fetchIndicatorID(ctx context.Context, runtime *common.RuntimeContext, level string, id string) (string, error) {
+	var path string
+	var params map[string]interface{}
+
+	if level == "objective" {
+		path = fmt.Sprintf("/open-apis/okr/v2/objectives/%s/indicators", id)
+		params = map[string]interface{}{"page_size": 100}
+	} else {
+		path = fmt.Sprintf("/open-apis/okr/v2/key_results/%s/indicators", id)
+		params = map[string]interface{}{"page_size": 100}
+	}
+
+	data, err := runtime.CallAPITyped("GET", path, params, nil)
+	if err != nil {
+		return "", wrapOkrNetworkErr(err, "failed to fetch indicators")
+	}
+
+	// Parse response to get indicator ID
+	// Response format: {"indicator": {"id": "...", ...}} (single object, not a list)
+	indicator, ok := data["indicator"].(map[string]interface{})
+	if !ok {
+		return "", errs.NewInternalError(errs.SubtypeUnknown, "indicator field not found in response")
+	}
+
+	indicatorID, ok := indicator["id"].(string)
+	if !ok || indicatorID == "" {
+		return "", errs.NewInternalError(errs.SubtypeUnknown, "indicator ID not found or empty")
+	}
+
+	return indicatorID, nil
+}
+
+// OKRIndicatorUpdate updates the current value of an indicator for an objective or key result.
+var OKRIndicatorUpdate = common.Shortcut{
+	Service:     "okr",
+	Command:     "+indicator-update",
+	Description: "Update the indicator current value for an objective or key result",
+	Risk:        "write",
+	Scopes:      []string{"okr:okr.content:writeonly"},
+	AuthTypes:   []string{"user", "bot"},
+	HasFormat:   true,
+	Flags: []common.Flag{
+		{Name: "level", Desc: "level to update: objective | key-result, Required.", Enum: []string{"objective", "key-result"}},
+		{Name: "id", Desc: "objective or key result ID (int64), Required."},
+		{Name: "value", Desc: "new current value for the indicator (number), Required."},
+	},
+	Validate: func(ctx context.Context, runtime *common.RuntimeContext) error {
+		level := runtime.Str("level")
+		if level == "" {
+			return errs.NewValidationError(errs.SubtypeInvalidArgument, "--level is required").WithParam("--level")
+		}
+		if level != "objective" && level != "key-result" {
+			return errs.NewValidationError(errs.SubtypeInvalidArgument, "--level must be one of: objective | key-result").WithParam("--level")
+		}
+
+		id := runtime.Str("id")
+		if id == "" {
+			return errs.NewValidationError(errs.SubtypeInvalidArgument, "--id is required").WithParam("--id")
+		}
+		if _, err := strconv.ParseInt(id, 10, 64); err != nil {
+			return errs.NewValidationError(errs.SubtypeInvalidArgument, "--id must be a valid int64").WithParam("--id")
+		}
+		if runtime.Str("value") == "" {
+			return errs.NewValidationError(errs.SubtypeInvalidArgument, "--value is required").WithParam("--value")
+		}
+		if _, err := parseIndicatorValue(runtime.Str("value")); err != nil {
+			return err
+		}
+
+		return nil
+	},
+	DryRun: func(ctx context.Context, runtime *common.RuntimeContext) *common.DryRunAPI {
+		level := runtime.Str("level")
+		id := runtime.Str("id")
+		value, _ := parseIndicatorValue(runtime.Str("value"))
+
+		apis := common.NewDryRunAPI()
+
+		var listPath string
+		if level == "objective" {
+			listPath = fmt.Sprintf("/open-apis/okr/v2/objectives/%s/indicators", id)
+		} else {
+			listPath = fmt.Sprintf("/open-apis/okr/v2/key_results/%s/indicators", id)
+		}
+
+		// First API: fetch indicator list
+		apis = apis.
+			GET(listPath).
+			Params(map[string]interface{}{"page_size": 100}).
+			Desc(fmt.Sprintf("Fetch indicators for the %s to get indicator ID", level))
+
+		// Second API: patch indicator value
+		patchPath := "/open-apis/okr/v2/indicators/:indicator_id"
+		patchBody := map[string]interface{}{
+			"current_value": value,
+		}
+		apis = apis.
+			PATCH(patchPath).
+			Body(patchBody).
+			Set("indicator_id", "<indicator_id_from_list>").
+			Desc("Update indicator current value")
+
+		return apis
+	},
+	Execute: func(ctx context.Context, runtime *common.RuntimeContext) error {
+		level := runtime.Str("level")
+		id := runtime.Str("id")
+		value, err := parseIndicatorValue(runtime.Str("value"))
+		if err != nil {
+			return err
+		}
+
+		// Step 1: Fetch indicator ID
+		indicatorID, err := fetchIndicatorID(ctx, runtime, level, id)
+		if err != nil {
+			return err
+		}
+
+		// Step 2: Update indicator value
+		patchPath := fmt.Sprintf("/open-apis/okr/v2/indicators/%s", indicatorID)
+		patchBody := map[string]interface{}{
+			"current_value": value,
+		}
+
+		_, err = runtime.CallAPITyped("PATCH", patchPath, nil, patchBody)
+		if err != nil {
+			return wrapOkrNetworkErr(err, "failed to update indicator value")
+		}
+
+		// Build response
+		result := map[string]interface{}{
+			"indicator_id":  indicatorID,
+			"current_value": value,
+			"level":         level,
+			"target_id":     id,
+		}
+
+		runtime.OutFormat(result, nil, func(w io.Writer) {
+			fmt.Fprintf(w, "Updated Indicator [%s]\n", indicatorID)
+			fmt.Fprintf(w, "  Level: %s\n", level)
+			fmt.Fprintf(w, "  Target ID: %s\n", id)
+			fmt.Fprintf(w, "  Current Value: %v\n", value)
+		})
+
+		return nil
+	},
+}

shortcuts/okr/okr_indicator_update_test.go

@@ -0,0 +1,391 @@
+// Copyright (c) 2026 Lark Technologies Pte. Ltd.
+// SPDX-License-Identifier: MIT
+
+package okr
+
+import (
+	"bytes"
+	"encoding/json"
+	"errors"
+	"strings"
+	"testing"
+
+	"github.com/larksuite/cli/errs"
+	"github.com/larksuite/cli/internal/cmdutil"
+	"github.com/larksuite/cli/internal/core"
+	"github.com/larksuite/cli/internal/httpmock"
+	"github.com/spf13/cobra"
+)
+
+func indicatorUpdateTestConfig(t *testing.T) *core.CliConfig {
+	t.Helper()
+	return &core.CliConfig{
+		AppID:     "test-okr-indicator-update",
+		AppSecret: "secret-okr-indicator-update",
+		Brand:     core.BrandFeishu,
+	}
+}
+
+func runIndicatorUpdateShortcut(t *testing.T, f *cmdutil.Factory, stdout *bytes.Buffer, args []string) error {
+	t.Helper()
+	parent := &cobra.Command{Use: "okr"}
+	OKRIndicatorUpdate.Mount(parent, f)
+	parent.SetArgs(args)
+	parent.SilenceErrors = true
+	parent.SilenceUsage = true
+	if stdout != nil {
+		stdout.Reset()
+	}
+	return parent.Execute()
+}
+
+// --- Validate tests ---
+
+func TestIndicatorUpdateValidate_MissingLevel(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, _ := cmdutil.TestFactory(t, indicatorUpdateTestConfig(t))
+	err := runIndicatorUpdateShortcut(t, f, stdout, []string{
+		"+indicator-update",
+		"--id", "123",
+		"--value", "50",
+	})
+	// cobra Required:true reports flag name without "--" prefix
+	if err == nil || !strings.Contains(err.Error(), "level") {
+		t.Fatalf("expected --level required error, got: %v", err)
+	}
+}
+
+func TestIndicatorUpdateValidate_InvalidLevel(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, _ := cmdutil.TestFactory(t, indicatorUpdateTestConfig(t))
+	err := runIndicatorUpdateShortcut(t, f, stdout, []string{
+		"+indicator-update",
+		"--level", "invalid",
+		"--id", "123",
+		"--value", "50",
+	})
+	if err == nil {
+		t.Fatal("expected error for invalid level")
+	}
+	_, ok := errs.ProblemOf(err)
+	if !ok {
+		t.Fatalf("expected typed error, got: %v", err)
+	}
+	validationErr, ok := err.(*errs.ValidationError)
+	if !ok {
+		t.Fatalf("expected ValidationError, got: %T", err)
+	}
+	if validationErr.Param != "--level" {
+		t.Fatalf("expected param --level, got %q", validationErr.Param)
+	}
+}
+
+func TestIndicatorUpdateValidate_MissingID(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, _ := cmdutil.TestFactory(t, indicatorUpdateTestConfig(t))
+	err := runIndicatorUpdateShortcut(t, f, stdout, []string{
+		"+indicator-update",
+		"--level", "objective",
+		"--value", "50",
+	})
+	if err == nil || !strings.Contains(err.Error(), "id") {
+		t.Fatalf("expected --id required error, got: %v", err)
+	}
+}
+
+func TestIndicatorUpdateValidate_InvalidID(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, _ := cmdutil.TestFactory(t, indicatorUpdateTestConfig(t))
+	err := runIndicatorUpdateShortcut(t, f, stdout, []string{
+		"+indicator-update",
+		"--level", "objective",
+		"--id", "not-a-number",
+		"--value", "50",
+	})
+	if err == nil {
+		t.Fatal("expected error for invalid id")
+	}
+	_, ok := errs.ProblemOf(err)
+	if !ok {
+		t.Fatalf("expected typed error, got: %v", err)
+	}
+	validationErr, ok := err.(*errs.ValidationError)
+	if !ok {
+		t.Fatalf("expected ValidationError, got: %T", err)
+	}
+	if validationErr.Param != "--id" {
+		t.Fatalf("expected param --id, got %q", validationErr.Param)
+	}
+}
+
+func TestIndicatorUpdateValidate_MissingValue(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, _ := cmdutil.TestFactory(t, indicatorUpdateTestConfig(t))
+	err := runIndicatorUpdateShortcut(t, f, stdout, []string{
+		"+indicator-update",
+		"--level", "objective",
+		"--id", "123",
+	})
+	if err == nil || !strings.Contains(err.Error(), "value") {
+		t.Fatalf("expected --value required error, got: %v", err)
+	}
+}
+
+func TestIndicatorUpdateValidate_InvalidValue(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, _ := cmdutil.TestFactory(t, indicatorUpdateTestConfig(t))
+	err := runIndicatorUpdateShortcut(t, f, stdout, []string{
+		"+indicator-update",
+		"--level", "objective",
+		"--id", "123",
+		"--value", "not-a-number",
+	})
+	if err == nil {
+		t.Fatal("expected error for invalid value")
+	}
+	_, ok := errs.ProblemOf(err)
+	if !ok {
+		t.Fatalf("expected typed error, got: %v", err)
+	}
+	validationErr, ok := err.(*errs.ValidationError)
+	if !ok {
+		t.Fatalf("expected ValidationError, got: %T", err)
+	}
+	if validationErr.Param != "--value" {
+		t.Fatalf("expected param --value, got %q", validationErr.Param)
+	}
+}
+
+func TestIndicatorUpdateValidate_Valid(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, reg := cmdutil.TestFactory(t, indicatorUpdateTestConfig(t))
+	// Mock fetch indicators
+	reg.Register(&httpmock.Stub{
+		Method: "GET",
+		URL:    "/open-apis/okr/v2/objectives/123/indicators",
+		Body: map[string]interface{}{
+			"code": 0,
+			"msg":  "ok",
+			"data": map[string]interface{}{
+				"indicator": map[string]interface{}{
+					"id": "ind-456",
+				},
+			},
+		},
+	})
+	// Mock patch indicator
+	reg.Register(&httpmock.Stub{
+		Method: "PATCH",
+		URL:    "/open-apis/okr/v2/indicators/ind-456",
+		Body: map[string]interface{}{
+			"code": 0,
+			"msg":  "ok",
+		},
+	})
+	err := runIndicatorUpdateShortcut(t, f, stdout, []string{
+		"+indicator-update",
+		"--level", "objective",
+		"--id", "123",
+		"--value", "75.5",
+	})
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+}
+
+// --- Execute tests ---
+
+func TestIndicatorUpdateExecute_Objectives_Success(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, reg := cmdutil.TestFactory(t, indicatorUpdateTestConfig(t))
+	// Mock fetch indicators
+	reg.Register(&httpmock.Stub{
+		Method: "GET",
+		URL:    "/open-apis/okr/v2/objectives/123/indicators",
+		Body: map[string]interface{}{
+			"code": 0,
+			"msg":  "ok",
+			"data": map[string]interface{}{
+				"indicator": map[string]interface{}{
+					"id": "ind-456",
+				},
+			},
+		},
+	})
+	// Mock patch indicator
+	reg.Register(&httpmock.Stub{
+		Method: "PATCH",
+		URL:    "/open-apis/okr/v2/indicators/ind-456",
+		Body: map[string]interface{}{
+			"code": 0,
+			"msg":  "ok",
+		},
+		BodyFilter: func(body []byte) bool {
+			var data map[string]interface{}
+			if err := json.Unmarshal(body, &data); err != nil {
+				return false
+			}
+			val, ok := data["current_value"].(float64)
+			return ok && val == 75.5
+		},
+	})
+	err := runIndicatorUpdateShortcut(t, f, stdout, []string{
+		"+indicator-update",
+		"--level", "objective",
+		"--id", "123",
+		"--value", "75.5",
+	})
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+}
+
+func TestIndicatorUpdateExecute_KeyResults_Success(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, reg := cmdutil.TestFactory(t, indicatorUpdateTestConfig(t))
+	// Mock fetch indicators
+	reg.Register(&httpmock.Stub{
+		Method: "GET",
+		URL:    "/open-apis/okr/v2/key_results/456/indicators",
+		Body: map[string]interface{}{
+			"code": 0,
+			"msg":  "ok",
+			"data": map[string]interface{}{
+				"indicator": map[string]interface{}{
+					"id": "ind-789",
+				},
+			},
+		},
+	})
+	// Mock patch indicator
+	reg.Register(&httpmock.Stub{
+		Method: "PATCH",
+		URL:    "/open-apis/okr/v2/indicators/ind-789",
+		Body: map[string]interface{}{
+			"code": 0,
+			"msg":  "ok",
+		},
+	})
+	err := runIndicatorUpdateShortcut(t, f, stdout, []string{
+		"+indicator-update",
+		"--level", "key-result",
+		"--id", "456",
+		"--value", "100",
+	})
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+}
+
+func TestIndicatorUpdateExecute_FetchAPIError(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, reg := cmdutil.TestFactory(t, indicatorUpdateTestConfig(t))
+	// Mock fetch indicators - API error
+	reg.Register(&httpmock.Stub{
+		Method: "GET",
+		URL:    "/open-apis/okr/v2/objectives/123/indicators",
+		Body: map[string]interface{}{
+			"code": 9999,
+			"msg":  "fetch error",
+		},
+	})
+	err := runIndicatorUpdateShortcut(t, f, stdout, []string{
+		"+indicator-update",
+		"--level", "objective",
+		"--id", "123",
+		"--value", "50",
+	})
+	if err == nil {
+		t.Fatal("expected error for fetch API failure")
+	}
+	prob, ok := errs.ProblemOf(err)
+	if !ok {
+		t.Fatalf("expected typed error, got: %v", err)
+	}
+	if prob.Category != errs.CategoryAPI {
+		t.Fatalf("expected CategoryAPI, got %q", prob.Category)
+	}
+	var apiErr *errs.APIError
+	if !errors.As(err, &apiErr) {
+		t.Fatalf("expected error to be *errs.APIError, got: %T", err)
+	}
+	if !errors.Is(err, apiErr) {
+		t.Fatal("errors.Is should find the APIError in the chain")
+	}
+}
+
+func TestIndicatorUpdateExecute_PatchAPIError(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, reg := cmdutil.TestFactory(t, indicatorUpdateTestConfig(t))
+	// Mock fetch indicators
+	reg.Register(&httpmock.Stub{
+		Method: "GET",
+		URL:    "/open-apis/okr/v2/objectives/123/indicators",
+		Body: map[string]interface{}{
+			"code": 0,
+			"msg":  "ok",
+			"data": map[string]interface{}{
+				"indicator": map[string]interface{}{
+					"id": "ind-456",
+				},
+			},
+		},
+	})
+	// Mock patch indicator - API error
+	reg.Register(&httpmock.Stub{
+		Method: "PATCH",
+		URL:    "/open-apis/okr/v2/indicators/ind-456",
+		Body: map[string]interface{}{
+			"code": 9999,
+			"msg":  "patch error",
+		},
+	})
+	err := runIndicatorUpdateShortcut(t, f, stdout, []string{
+		"+indicator-update",
+		"--level", "objective",
+		"--id", "123",
+		"--value", "50",
+	})
+	if err == nil {
+		t.Fatal("expected error for patch API failure")
+	}
+	prob, ok := errs.ProblemOf(err)
+	if !ok {
+		t.Fatalf("expected typed error, got: %v", err)
+	}
+	if prob.Category != errs.CategoryAPI {
+		t.Fatalf("expected CategoryAPI, got %q", prob.Category)
+	}
+	var apiErr *errs.APIError
+	if !errors.As(err, &apiErr) {
+		t.Fatalf("expected error to be *errs.APIError, got: %T", err)
+	}
+	if !errors.Is(err, apiErr) {
+		t.Fatal("errors.Is should find the APIError in the chain")
+	}
+}
+
+// --- parseIndicatorValue tests ---
+
+func TestParseIndicatorValue_Valid(t *testing.T) {
+	t.Parallel()
+	tests := []string{"0", "100", "75.5", "-10", "0.001", "99999999999"}
+	for _, v := range tests {
+		result, err := parseIndicatorValue(v)
+		if err != nil {
+			t.Fatalf("expected no error for %q, got: %v", v, err)
+		}
+		_ = result
+	}
+}
+
+func TestParseIndicatorValue_Invalid(t *testing.T) {
+	t.Parallel()
+	tests := []string{"", "abc", "1e100000", "100000000000"}
+	for _, v := range tests {
+		_, err := parseIndicatorValue(v)
+		if err == nil {
+			t.Fatalf("expected error for %q", v)
+		}
+	}
+}

shortcuts/okr/okr_reorder.go

@@ -0,0 +1,448 @@
+// Copyright (c) 2026 Lark Technologies Pte. Ltd.
+// SPDX-License-Identifier: MIT
+
+package okr
+
+import (
+	"context"
+	"encoding/json"
+	"fmt"
+	"io"
+	"sort"
+	"strconv"
+	"strings"
+	"time"
+
+	"github.com/larksuite/cli/errs"
+	"github.com/larksuite/cli/shortcuts/common"
+)
+
+// reorderItem is the interface for items that have an ID.
+type reorderItem interface {
+	GetID() string
+}
+
+// reorderOp represents a single reorder operation.
+type reorderOp struct {
+	ID       string `json:"id"`
+	Position int32  `json:"position"`
+}
+
+// parseReorderOps parses and validates the --ops JSON array.
+func parseReorderOps(opsStr string) ([]reorderOp, error) {
+	var ops []reorderOp
+	if err := json.Unmarshal([]byte(opsStr), &ops); err != nil {
+		return nil, errs.NewValidationError(errs.SubtypeInvalidArgument, "--ops must be valid JSON array: %s", err).WithParam("--ops").WithCause(err)
+	}
+	if len(ops) == 0 {
+		return nil, errs.NewValidationError(errs.SubtypeInvalidArgument, "--ops must contain at least one operation").WithParam("--ops")
+	}
+
+	seen := make(map[string]bool)
+	seenPos := make(map[int32]bool)
+	for i, op := range ops {
+		if strings.TrimSpace(op.ID) == "" {
+			return nil, errs.NewValidationError(errs.SubtypeInvalidArgument, "ops[%d].id is required and cannot be empty", i).WithParam("--ops")
+		}
+		if op.Position <= 0 {
+			return nil, errs.NewValidationError(errs.SubtypeInvalidArgument, "ops[%d].position must be a positive integer", i).WithParam("--ops")
+		}
+		if seen[op.ID] {
+			return nil, errs.NewValidationError(errs.SubtypeInvalidArgument, "duplicate id %q in --ops", op.ID).WithParam("--ops")
+		}
+		if seenPos[op.Position] {
+			return nil, errs.NewValidationError(errs.SubtypeInvalidArgument, "duplicate position %d in --ops", op.Position).WithParam("--ops")
+		}
+		seen[op.ID] = true
+		seenPos[op.Position] = true
+	}
+	return ops, nil
+}
+
+// fetchObjectives fetches all objectives in a cycle.
+func fetchObjectives(ctx context.Context, runtime *common.RuntimeContext, cycleID string) ([]Objective, error) {
+	queryParams := map[string]interface{}{"page_size": "100"}
+	var objectives []Objective
+	page := 0
+
+	for {
+		if page > 0 {
+			select {
+			case <-ctx.Done():
+				return nil, ctx.Err()
+			case <-time.After(500 * time.Millisecond):
+			}
+		}
+		page++
+
+		path := fmt.Sprintf("/open-apis/okr/v2/cycles/%s/objectives", cycleID)
+		data, err := runtime.CallAPITyped("GET", path, queryParams, nil)
+		if err != nil {
+			return nil, wrapOkrNetworkErr(err, "failed to fetch objectives")
+		}
+
+		itemsRaw, _ := data["items"].([]interface{})
+		for _, item := range itemsRaw {
+			raw, err := json.Marshal(item)
+			if err != nil {
+				continue
+			}
+			var obj Objective
+			if err := json.Unmarshal(raw, &obj); err != nil {
+				continue
+			}
+			objectives = append(objectives, obj)
+		}
+
+		hasMore, pageToken := common.PaginationMeta(data)
+		if !hasMore || pageToken == "" {
+			break
+		}
+		queryParams["page_token"] = pageToken
+	}
+
+	// Sort objectives by position
+	sort.Slice(objectives, func(i, j int) bool {
+		pi := int32(0)
+		if objectives[i].Position != nil {
+			pi = *objectives[i].Position
+		}
+		pj := int32(0)
+		if objectives[j].Position != nil {
+			pj = *objectives[j].Position
+		}
+		return pi < pj
+	})
+
+	return objectives, nil
+}
+
+// fetchKeyResults fetches all key results for an objective.
+func fetchKeyResults(ctx context.Context, runtime *common.RuntimeContext, objectiveID string) ([]KeyResult, error) {
+	queryParams := map[string]interface{}{"page_size": "100"}
+	var keyResults []KeyResult
+	page := 0
+
+	for {
+		if page > 0 {
+			select {
+			case <-ctx.Done():
+				return nil, ctx.Err()
+			case <-time.After(500 * time.Millisecond):
+			}
+		}
+		page++
+
+		path := fmt.Sprintf("/open-apis/okr/v2/objectives/%s/key_results", objectiveID)
+		data, err := runtime.CallAPITyped("GET", path, queryParams, nil)
+		if err != nil {
+			return nil, wrapOkrNetworkErr(err, "failed to fetch key results")
+		}
+
+		itemsRaw, _ := data["items"].([]interface{})
+		for _, item := range itemsRaw {
+			raw, err := json.Marshal(item)
+			if err != nil {
+				continue
+			}
+			var kr KeyResult
+			if err := json.Unmarshal(raw, &kr); err != nil {
+				continue
+			}
+			keyResults = append(keyResults, kr)
+		}
+
+		hasMore, pageToken := common.PaginationMeta(data)
+		if !hasMore || pageToken == "" {
+			break
+		}
+		queryParams["page_token"] = pageToken
+	}
+
+	// Sort key results by position
+	sort.Slice(keyResults, func(i, j int) bool {
+		pi := int32(0)
+		if keyResults[i].Position != nil {
+			pi = *keyResults[i].Position
+		}
+		pj := int32(0)
+		if keyResults[j].Position != nil {
+			pj = *keyResults[j].Position
+		}
+		return pi < pj
+	})
+
+	return keyResults, nil
+}
+
+// buildReorderedIDs builds the complete ordered ID list from current items and reorder ops.
+// Positions are treated as 1-indexed placement keys stored in a map (safe for large values).
+// Items are first placed at user-specified positions, remaining items fill empty slots
+// in original order starting from position 1, and final output is sorted by position.
+func buildReorderedIDs[T reorderItem](items []T, ops []reorderOp, total int) ([]string, error) {
+	// Create a map of ID to current position
+	idToPos := make(map[string]int)
+	for i, item := range items {
+		idToPos[item.GetID()] = i
+	}
+
+	// Validate all ops IDs exist
+	for _, op := range ops {
+		if _, ok := idToPos[op.ID]; !ok {
+			return nil, errs.NewValidationError(errs.SubtypeInvalidArgument, "id %q not found in current list", op.ID).WithParam("--ops")
+		}
+	}
+
+	// Use map to store position -> ID (1-indexed, safe for large position values)
+	posToID := make(map[int]string)
+	used := make(map[string]bool)
+	for _, op := range ops {
+		posToID[int(op.Position)] = op.ID
+		used[op.ID] = true
+	}
+
+	// Collect unused items in original order
+	var unused []string
+	for _, item := range items {
+		id := item.GetID()
+		if !used[id] {
+			unused = append(unused, id)
+		}
+	}
+
+	// Fill empty slots starting from position 1, in original order
+	unusedIdx := 0
+	for pos := 1; unusedIdx < len(unused); pos++ {
+		if _, occupied := posToID[pos]; !occupied {
+			posToID[pos] = unused[unusedIdx]
+			unusedIdx++
+		}
+	}
+
+	// Collect all positions, sort them, and build result in position order
+	positions := make([]int, 0, len(posToID))
+	for pos := range posToID {
+		positions = append(positions, pos)
+	}
+	sort.Ints(positions)
+
+	result := make([]string, 0, len(positions))
+	for _, pos := range positions {
+		result = append(result, posToID[pos])
+	}
+
+	return result, nil
+}
+
+// GetID implements the interface for Objective.
+func (o Objective) GetID() string { return o.ID }
+
+// GetID implements the interface for KeyResult.
+func (k KeyResult) GetID() string { return k.ID }
+
+// OKRReorder adjusts the position of objectives or key results.
+var OKRReorder = common.Shortcut{
+	Service:     "okr",
+	Command:     "+reorder",
+	Description: "Adjust the position (order) of OKR objectives or key results",
+	Risk:        "write",
+	Scopes:      []string{"okr:okr.content:writeonly"},
+	AuthTypes:   []string{"user", "bot"},
+	HasFormat:   true,
+	Flags: []common.Flag{
+		{Name: "level", Desc: "level to reorder: objective | key-result, Required.", Enum: []string{"objective", "key-result"}},
+		{Name: "cycle-id", Desc: "OKR cycle ID (int64), Required."},
+		{Name: "objective-id", Desc: "objective ID (required when --level=key-result)"},
+		{Name: "ops", Desc: "JSON array of reorder operations: [{\"id\":\"...\",\"position\":1}], Required.", Input: []string{common.File, common.Stdin}},
+	},
+	Validate: func(ctx context.Context, runtime *common.RuntimeContext) error {
+		level := runtime.Str("level")
+		if strings.TrimSpace(level) == "" {
+			return errs.NewValidationError(errs.SubtypeInvalidArgument, "--level is required").WithParam("--level")
+		}
+		if level != "objective" && level != "key-result" {
+			return errs.NewValidationError(errs.SubtypeInvalidArgument, "--level must be one of: objective | key-result").WithParam("--level")
+		}
+
+		cycleID := runtime.Str("cycle-id")
+		if strings.TrimSpace(cycleID) == "" {
+			return errs.NewValidationError(errs.SubtypeInvalidArgument, "--cycle-id is required").WithParam("--cycle-id")
+		}
+		if id, err := strconv.ParseInt(cycleID, 10, 64); err != nil || id <= 0 {
+			return errs.NewValidationError(errs.SubtypeInvalidArgument, "--cycle-id must be a positive int64").WithParam("--cycle-id")
+		}
+
+		if level == "key-result" {
+			objID := runtime.Str("objective-id")
+			if objID == "" {
+				return errs.NewValidationError(errs.SubtypeInvalidArgument, "--objective-id is required when --level=key-result").WithParam("--objective-id")
+			}
+			if id, err := strconv.ParseInt(objID, 10, 64); err != nil || id <= 0 {
+				return errs.NewValidationError(errs.SubtypeInvalidArgument, "--objective-id must be a positive int64").WithParam("--objective-id")
+			}
+		}
+
+		opsStr := runtime.Str("ops")
+		if strings.TrimSpace(opsStr) == "" {
+			return errs.NewValidationError(errs.SubtypeInvalidArgument, "--ops is required").WithParam("--ops")
+		}
+		if err := common.RejectDangerousCharsTyped("--ops", opsStr); err != nil {
+			return err
+		}
+		if _, err := parseReorderOps(opsStr); err != nil {
+			return err
+		}
+
+		return nil
+	},
+	DryRun: func(ctx context.Context, runtime *common.RuntimeContext) *common.DryRunAPI {
+		level := runtime.Str("level")
+		cycleID := runtime.Str("cycle-id")
+		objectiveID := runtime.Str("objective-id")
+		ops, _ := parseReorderOps(runtime.Str("ops"))
+
+		apis := common.NewDryRunAPI()
+
+		if level == "objective" {
+			// First fetch objectives
+			listParams := map[string]interface{}{
+				"page_size": 100,
+			}
+			listPath := fmt.Sprintf("/open-apis/okr/v2/cycles/%s/objectives", cycleID)
+			apis = apis.
+				GET(listPath).
+				Params(listParams).
+				Desc("Fetch all objectives in the cycle to determine current order")
+
+			// Then reorder
+			reorderParams := map[string]interface{}{
+				"cycle_id": cycleID,
+			}
+			// Build sample body with placeholder IDs
+			objectiveIDs := make([]string, 0, len(ops))
+			for _, op := range ops {
+				objectiveIDs = append(objectiveIDs, op.ID)
+			}
+			reorderBody := map[string]interface{}{
+				"objective_ids": objectiveIDs,
+			}
+			reorderPath := fmt.Sprintf("/open-apis/okr/v2/cycles/%s/objectives_position", cycleID)
+			apis = apis.
+				PUT(reorderPath).
+				Params(reorderParams).
+				Body(reorderBody).
+				Desc("Update objective positions (full list sent, not just changes)")
+		} else {
+			// key-result level
+			listParams := map[string]interface{}{
+				"page_size": 100,
+			}
+			listPath := fmt.Sprintf("/open-apis/okr/v2/objectives/%s/key_results", objectiveID)
+			apis = apis.
+				GET(listPath).
+				Params(listParams).
+				Desc("Fetch all key results for the objective to determine current order")
+
+			reorderParams := map[string]interface{}{
+				"objective_id": objectiveID,
+			}
+			// Build sample body with placeholder IDs
+			keyResultIDs := make([]string, 0, len(ops))
+			for _, op := range ops {
+				keyResultIDs = append(keyResultIDs, op.ID)
+			}
+			reorderBody := map[string]interface{}{
+				"key_result_ids": keyResultIDs,
+			}
+			reorderPath := fmt.Sprintf("/open-apis/okr/v2/objectives/%s/key_results_position", objectiveID)
+			apis = apis.
+				PUT(reorderPath).
+				Params(reorderParams).
+				Body(reorderBody).
+				Desc("Update key result positions (full list sent, not just changes)")
+		}
+
+		return apis
+	},
+	Execute: func(ctx context.Context, runtime *common.RuntimeContext) error {
+		level := runtime.Str("level")
+		cycleID := runtime.Str("cycle-id")
+		objectiveID := runtime.Str("objective-id")
+		ops, err := parseReorderOps(runtime.Str("ops"))
+		if err != nil {
+			return err
+		}
+
+		var reorderedIDs []string
+		var total int
+
+		if level == "objective" {
+			objectives, err := fetchObjectives(ctx, runtime, cycleID)
+			if err != nil {
+				return err
+			}
+			total = len(objectives)
+
+			reorderedIDs, err = buildReorderedIDs(objectives, ops, total)
+			if err != nil {
+				return err
+			}
+
+			// Submit reorder
+			params := map[string]interface{}{
+				"cycle_id": cycleID,
+			}
+			body := map[string]interface{}{
+				"objective_ids": reorderedIDs,
+			}
+			path := fmt.Sprintf("/open-apis/okr/v2/cycles/%s/objectives_position", cycleID)
+			_, err = runtime.CallAPITyped("PUT", path, params, body)
+			if err != nil {
+				return wrapOkrNetworkErr(err, "failed to update objective positions")
+			}
+		} else {
+			// key-result level
+			keyResults, err := fetchKeyResults(ctx, runtime, objectiveID)
+			if err != nil {
+				return err
+			}
+			total = len(keyResults)
+
+			reorderedIDs, err = buildReorderedIDs(keyResults, ops, total)
+			if err != nil {
+				return err
+			}
+
+			// Submit reorder
+			params := map[string]interface{}{
+				"objective_id": objectiveID,
+			}
+			body := map[string]interface{}{
+				"key_result_ids": reorderedIDs,
+			}
+			path := fmt.Sprintf("/open-apis/okr/v2/objectives/%s/key_results_position", objectiveID)
+			_, err = runtime.CallAPITyped("PUT", path, params, body)
+			if err != nil {
+				return wrapOkrNetworkErr(err, "failed to update key result positions")
+			}
+		}
+
+		// Build response
+		result := map[string]interface{}{
+			"level":    level,
+			"cycle_id": cycleID,
+			"total":    total,
+			"ordered":  reorderedIDs,
+		}
+
+		runtime.OutFormat(result, nil, func(w io.Writer) {
+			fmt.Fprintf(w, "Successfully reordered %d %s(s)\n", total, level)
+			fmt.Fprintln(w, "New order:")
+			for i, id := range reorderedIDs {
+				fmt.Fprintf(w, "  Position %d: %s\n", i+1, id)
+			}
+		})
+
+		return nil
+	},
+}

shortcuts/okr/okr_reorder_test.go

@@ -0,0 +1,712 @@
+// Copyright (c) 2026 Lark Technologies Pte. Ltd.
+// SPDX-License-Identifier: MIT
+
+package okr
+
+import (
+	"bytes"
+	"encoding/json"
+	"errors"
+	"strings"
+	"testing"
+
+	"github.com/larksuite/cli/errs"
+	"github.com/larksuite/cli/internal/cmdutil"
+	"github.com/larksuite/cli/internal/core"
+	"github.com/larksuite/cli/internal/httpmock"
+	"github.com/spf13/cobra"
+)
+
+// testReorderItem implements reorderItem for testing.
+type testReorderItem struct {
+	id string
+}
+
+func (t testReorderItem) GetID() string { return t.id }
+
+func reorderTestConfig(t *testing.T) *core.CliConfig {
+	t.Helper()
+	return &core.CliConfig{
+		AppID:     "test-okr-reorder",
+		AppSecret: "secret-okr-reorder",
+		Brand:     core.BrandFeishu,
+	}
+}
+
+func runReorderShortcut(t *testing.T, f *cmdutil.Factory, stdout *bytes.Buffer, args []string) error {
+	t.Helper()
+	parent := &cobra.Command{Use: "okr"}
+	OKRReorder.Mount(parent, f)
+	parent.SetArgs(args)
+	parent.SilenceErrors = true
+	parent.SilenceUsage = true
+	if stdout != nil {
+		stdout.Reset()
+	}
+	return parent.Execute()
+}
+
+// --- Validate tests ---
+
+func TestReorderValidate_MissingLevel(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, _ := cmdutil.TestFactory(t, reorderTestConfig(t))
+	err := runReorderShortcut(t, f, stdout, []string{
+		"+reorder",
+		"--cycle-id", "123",
+		"--ops", `[{"id":"1","position":1}]`,
+	})
+	// cobra Required:true reports flag name without "--" prefix
+	if err == nil || !strings.Contains(err.Error(), "level") {
+		t.Fatalf("expected --level required error, got: %v", err)
+	}
+}
+
+func TestReorderValidate_InvalidLevel(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, _ := cmdutil.TestFactory(t, reorderTestConfig(t))
+	err := runReorderShortcut(t, f, stdout, []string{
+		"+reorder",
+		"--level", "invalid",
+		"--cycle-id", "123",
+		"--ops", `[{"id":"1","position":1}]`,
+	})
+	if err == nil {
+		t.Fatal("expected error for invalid --level")
+	}
+	_, ok := errs.ProblemOf(err)
+	if !ok {
+		t.Fatalf("expected typed error, got: %v", err)
+	}
+	validationErr, ok := err.(*errs.ValidationError)
+	if !ok {
+		t.Fatalf("expected ValidationError, got: %T", err)
+	}
+	if validationErr.Param != "--level" {
+		t.Fatalf("expected param --level, got %q", validationErr.Param)
+	}
+}
+
+func TestReorderValidate_MissingCycleID(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, _ := cmdutil.TestFactory(t, reorderTestConfig(t))
+	err := runReorderShortcut(t, f, stdout, []string{
+		"+reorder",
+		"--level", "objective",
+		"--ops", `[{"id":"1","position":1}]`,
+	})
+	// cobra Required:true reports flag name without "--" prefix
+	if err == nil || !strings.Contains(err.Error(), "cycle-id") {
+		t.Fatalf("expected --cycle-id required error, got: %v", err)
+	}
+}
+
+func TestReorderValidate_MissingObjectiveIDForKRLevel(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, _ := cmdutil.TestFactory(t, reorderTestConfig(t))
+	err := runReorderShortcut(t, f, stdout, []string{
+		"+reorder",
+		"--level", "key-result",
+		"--cycle-id", "123",
+		"--ops", `[{"id":"1","position":1}]`,
+	})
+	if err == nil {
+		t.Fatal("expected error for missing --objective-id when --level=key-result")
+	}
+	_, ok := errs.ProblemOf(err)
+	if !ok {
+		t.Fatalf("expected typed error, got: %v", err)
+	}
+	validationErr, ok := err.(*errs.ValidationError)
+	if !ok {
+		t.Fatalf("expected ValidationError, got: %T", err)
+	}
+	if validationErr.Param != "--objective-id" {
+		t.Fatalf("expected param --objective-id, got %q", validationErr.Param)
+	}
+}
+
+func TestReorderValidate_MissingOps(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, _ := cmdutil.TestFactory(t, reorderTestConfig(t))
+	err := runReorderShortcut(t, f, stdout, []string{
+		"+reorder",
+		"--level", "objective",
+		"--cycle-id", "123",
+	})
+	// cobra Required:true reports flag name without "--" prefix
+	if err == nil || !strings.Contains(err.Error(), "ops") {
+		t.Fatalf("expected --ops required error, got: %v", err)
+	}
+}
+
+func TestReorderValidate_InvalidOpsJSON(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, _ := cmdutil.TestFactory(t, reorderTestConfig(t))
+	err := runReorderShortcut(t, f, stdout, []string{
+		"+reorder",
+		"--level", "objective",
+		"--cycle-id", "123",
+		"--ops", "not-json",
+	})
+	if err == nil {
+		t.Fatal("expected error for invalid --ops JSON")
+	}
+	prob, ok := errs.ProblemOf(err)
+	if !ok {
+		t.Fatalf("expected typed error, got: %v", err)
+	}
+	if prob.Category != errs.CategoryValidation {
+		t.Fatalf("expected CategoryValidation, got %q", prob.Category)
+	}
+	var validationErr *errs.ValidationError
+	if !errors.As(err, &validationErr) {
+		t.Fatalf("expected error to be *errs.ValidationError, got: %T", err)
+	}
+	if !errors.Is(err, validationErr) {
+		t.Fatal("errors.Is should find the ValidationError in the chain")
+	}
+	if validationErr.Param != "--ops" {
+		t.Fatalf("expected param --ops, got %q", validationErr.Param)
+	}
+}
+
+func TestReorderValidate_EmptyOpsArray(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, _ := cmdutil.TestFactory(t, reorderTestConfig(t))
+	err := runReorderShortcut(t, f, stdout, []string{
+		"+reorder",
+		"--level", "objective",
+		"--cycle-id", "123",
+		"--ops", "[]",
+	})
+	if err == nil {
+		t.Fatal("expected error for empty --ops array")
+	}
+	_, ok := errs.ProblemOf(err)
+	if !ok {
+		t.Fatalf("expected typed error, got: %v", err)
+	}
+	validationErr, ok := err.(*errs.ValidationError)
+	if !ok {
+		t.Fatalf("expected ValidationError, got: %T", err)
+	}
+	if validationErr.Param != "--ops" {
+		t.Fatalf("expected param --ops, got %q", validationErr.Param)
+	}
+}
+
+func TestReorderValidate_DuplicateID(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, _ := cmdutil.TestFactory(t, reorderTestConfig(t))
+	err := runReorderShortcut(t, f, stdout, []string{
+		"+reorder",
+		"--level", "objective",
+		"--cycle-id", "123",
+		"--ops", `[{"id":"1","position":1},{"id":"1","position":2}]`,
+	})
+	if err == nil {
+		t.Fatal("expected error for duplicate id in --ops")
+	}
+	prob, ok := errs.ProblemOf(err)
+	if !ok {
+		t.Fatalf("expected typed error, got: %v", err)
+	}
+	if prob.Category != errs.CategoryValidation {
+		t.Fatalf("expected CategoryValidation, got %q", prob.Category)
+	}
+	var validationErr *errs.ValidationError
+	if !errors.As(err, &validationErr) {
+		t.Fatalf("expected error to be *errs.ValidationError, got: %T", err)
+	}
+	if !errors.Is(err, validationErr) {
+		t.Fatal("errors.Is should find the ValidationError in the chain")
+	}
+	if validationErr.Param != "--ops" {
+		t.Fatalf("expected param --ops, got %q", validationErr.Param)
+	}
+	if !strings.Contains(err.Error(), "duplicate id") {
+		t.Fatalf("expected error to mention duplicate id, got: %v", err)
+	}
+}
+
+func TestReorderValidate_DuplicatePosition(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, _ := cmdutil.TestFactory(t, reorderTestConfig(t))
+	err := runReorderShortcut(t, f, stdout, []string{
+		"+reorder",
+		"--level", "objective",
+		"--cycle-id", "123",
+		"--ops", `[{"id":"1","position":1},{"id":"2","position":1}]`,
+	})
+	if err == nil {
+		t.Fatal("expected error for duplicate position in --ops")
+	}
+	prob, ok := errs.ProblemOf(err)
+	if !ok {
+		t.Fatalf("expected typed error, got: %v", err)
+	}
+	if prob.Category != errs.CategoryValidation {
+		t.Fatalf("expected CategoryValidation, got %q", prob.Category)
+	}
+	var validationErr *errs.ValidationError
+	if !errors.As(err, &validationErr) {
+		t.Fatalf("expected error to be *errs.ValidationError, got: %T", err)
+	}
+	if !errors.Is(err, validationErr) {
+		t.Fatal("errors.Is should find the ValidationError in the chain")
+	}
+	if validationErr.Param != "--ops" {
+		t.Fatalf("expected param --ops, got %q", validationErr.Param)
+	}
+	if !strings.Contains(err.Error(), "duplicate position") {
+		t.Fatalf("expected error to mention duplicate position, got: %v", err)
+	}
+}
+
+func TestReorderValidate_NegativePosition(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, _ := cmdutil.TestFactory(t, reorderTestConfig(t))
+	err := runReorderShortcut(t, f, stdout, []string{
+		"+reorder",
+		"--level", "objective",
+		"--cycle-id", "123",
+		"--ops", `[{"id":"1","position":0}]`,
+	})
+	if err == nil {
+		t.Fatal("expected error for position <= 0")
+	}
+	_, ok := errs.ProblemOf(err)
+	if !ok {
+		t.Fatalf("expected typed error, got: %v", err)
+	}
+	validationErr, ok := err.(*errs.ValidationError)
+	if !ok {
+		t.Fatalf("expected ValidationError, got: %T", err)
+	}
+	if validationErr.Param != "--ops" {
+		t.Fatalf("expected param --ops, got %q", validationErr.Param)
+	}
+}
+
+// --- DryRun tests ---
+
+func TestReorderDryRun_Objectives(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, _ := cmdutil.TestFactory(t, reorderTestConfig(t))
+	err := runReorderShortcut(t, f, stdout, []string{
+		"+reorder",
+		"--level", "objective",
+		"--cycle-id", "123",
+		"--ops", `[{"id":"1","position":2},{"id":"2","position":1}]`,
+		"--dry-run",
+	})
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	output := stdout.String()
+	if !strings.Contains(output, "/open-apis/okr/v2/cycles/123/objectives") {
+		t.Fatalf("dry-run output should contain objectives list API path, got: %s", output)
+	}
+	if !strings.Contains(output, "GET") {
+		t.Fatalf("dry-run output should contain GET method for list, got: %s", output)
+	}
+	if !strings.Contains(output, "/open-apis/okr/v2/cycles/123/objectives_position") {
+		t.Fatalf("dry-run output should contain position update API path, got: %s", output)
+	}
+	if !strings.Contains(output, "PUT") {
+		t.Fatalf("dry-run output should contain PUT method for update, got: %s", output)
+	}
+	if !strings.Contains(output, "objective_ids") {
+		t.Fatalf("dry-run output should contain objective_ids in body, got: %s", output)
+	}
+}
+
+func TestReorderDryRun_KeyResults(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, _ := cmdutil.TestFactory(t, reorderTestConfig(t))
+	err := runReorderShortcut(t, f, stdout, []string{
+		"+reorder",
+		"--level", "key-result",
+		"--cycle-id", "123",
+		"--objective-id", "456",
+		"--ops", `[{"id":"1","position":2},{"id":"2","position":1}]`,
+		"--dry-run",
+	})
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	output := stdout.String()
+	if !strings.Contains(output, "/open-apis/okr/v2/objectives/456/key_results") {
+		t.Fatalf("dry-run output should contain key_results list API path, got: %s", output)
+	}
+	if !strings.Contains(output, "/open-apis/okr/v2/objectives/456/key_results_position") {
+		t.Fatalf("dry-run output should contain key_results position update API path, got: %s", output)
+	}
+	if !strings.Contains(output, "key_result_ids") {
+		t.Fatalf("dry-run output should contain key_result_ids in body, got: %s", output)
+	}
+}
+
+// --- Execute tests ---
+
+func TestReorderExecute_Objectives_Success(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, reg := cmdutil.TestFactory(t, reorderTestConfig(t))
+	// Mock fetch objectives
+	reg.Register(&httpmock.Stub{
+		Method: "GET",
+		URL:    "/open-apis/okr/v2/cycles/123/objectives",
+		Body: map[string]interface{}{
+			"code": 0,
+			"msg":  "ok",
+			"data": map[string]interface{}{
+				"items": []interface{}{
+					map[string]interface{}{"id": "1", "position": 1, "cycle_id": "123", "owner": map[string]interface{}{"owner_type": "user"}},
+					map[string]interface{}{"id": "2", "position": 2, "cycle_id": "123", "owner": map[string]interface{}{"owner_type": "user"}},
+					map[string]interface{}{"id": "3", "position": 3, "cycle_id": "123", "owner": map[string]interface{}{"owner_type": "user"}},
+				},
+				"has_more": false,
+			},
+		},
+	})
+	// Mock reorder
+	reg.Register(&httpmock.Stub{
+		Method: "PUT",
+		URL:    "/open-apis/okr/v2/cycles/123/objectives_position",
+		Body: map[string]interface{}{
+			"code": 0,
+			"msg":  "ok",
+			"data": map[string]interface{}{
+				"items": []interface{}{
+					map[string]interface{}{"id": "3", "position": 1},
+					map[string]interface{}{"id": "1", "position": 2},
+					map[string]interface{}{"id": "2", "position": 3},
+				},
+			},
+		},
+	})
+	err := runReorderShortcut(t, f, stdout, []string{
+		"+reorder",
+		"--level", "objective",
+		"--cycle-id", "123",
+		"--ops", `[{"id":"3","position":1}]`,
+	})
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+
+	// Verify output
+	data := decodeEnvelope(t, stdout)
+	if data["level"] != "objective" {
+		t.Fatalf("expected level=objective, got %v", data["level"])
+	}
+	if data["cycle_id"] != "123" {
+		t.Fatalf("expected cycle_id=123, got %v", data["cycle_id"])
+	}
+	ordered, _ := data["ordered"].([]interface{})
+	if len(ordered) != 3 {
+		t.Fatalf("expected 3 items in ordered list, got %d", len(ordered))
+	}
+	// First should be 3, then 1, then 2
+	if ordered[0] != "3" || ordered[1] != "1" || ordered[2] != "2" {
+		t.Fatalf("expected ordered [3,1,2], got %v", ordered)
+	}
+}
+
+func TestReorderExecute_KeyResults_Success(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, reg := cmdutil.TestFactory(t, reorderTestConfig(t))
+	// Mock fetch key results
+	reg.Register(&httpmock.Stub{
+		Method: "GET",
+		URL:    "/open-apis/okr/v2/objectives/456/key_results",
+		Body: map[string]interface{}{
+			"code": 0,
+			"msg":  "ok",
+			"data": map[string]interface{}{
+				"items": []interface{}{
+					map[string]interface{}{"id": "kr1", "position": 1, "objective_id": "456", "owner": map[string]interface{}{"owner_type": "user"}},
+					map[string]interface{}{"id": "kr2", "position": 2, "objective_id": "456", "owner": map[string]interface{}{"owner_type": "user"}},
+				},
+				"has_more": false,
+			},
+		},
+	})
+	// Mock reorder
+	reg.Register(&httpmock.Stub{
+		Method: "PUT",
+		URL:    "/open-apis/okr/v2/objectives/456/key_results_position",
+		Body: map[string]interface{}{
+			"code": 0,
+			"msg":  "ok",
+			"data": map[string]interface{}{
+				"items": []interface{}{
+					map[string]interface{}{"id": "kr2", "position": 1},
+					map[string]interface{}{"id": "kr1", "position": 2},
+				},
+			},
+		},
+	})
+	err := runReorderShortcut(t, f, stdout, []string{
+		"+reorder",
+		"--level", "key-result",
+		"--cycle-id", "123",
+		"--objective-id", "456",
+		"--ops", `[{"id":"kr2","position":1}]`,
+	})
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+
+	// Verify output
+	data := decodeEnvelope(t, stdout)
+	if data["level"] != "key-result" {
+		t.Fatalf("expected level=key-result, got %v", data["level"])
+	}
+	if data["cycle_id"] != "123" {
+		t.Fatalf("expected cycle_id=123, got %v", data["cycle_id"])
+	}
+	ordered, _ := data["ordered"].([]interface{})
+	if len(ordered) != 2 {
+		t.Fatalf("expected 2 items in ordered list, got %d", len(ordered))
+	}
+	if ordered[0] != "kr2" || ordered[1] != "kr1" {
+		t.Fatalf("expected ordered [kr2,kr1], got %v", ordered)
+	}
+}
+
+func TestReorderExecute_PositionOutOfRange(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, reg := cmdutil.TestFactory(t, reorderTestConfig(t))
+	// Mock fetch objectives (only 2 items)
+	reg.Register(&httpmock.Stub{
+		Method: "GET",
+		URL:    "/open-apis/okr/v2/cycles/123/objectives",
+		Body: map[string]interface{}{
+			"code": 0,
+			"msg":  "ok",
+			"data": map[string]interface{}{
+				"items": []interface{}{
+					map[string]interface{}{"id": "1", "position": 1, "cycle_id": "123", "owner": map[string]interface{}{"owner_type": "user"}},
+					map[string]interface{}{"id": "2", "position": 2, "cycle_id": "123", "owner": map[string]interface{}{"owner_type": "user"}},
+				},
+				"has_more": false,
+			},
+		},
+	})
+	reg.Register(&httpmock.Stub{
+		Method: "PUT",
+		URL:    "/open-apis/okr/v2/cycles/123/objectives_position",
+		Body: map[string]interface{}{
+			"code": 0,
+			"msg":  "ok",
+		},
+		BodyFilter: func(body []byte) bool {
+			var data map[string]interface{}
+			if err := json.Unmarshal(body, &data); err != nil {
+				return false
+			}
+			ids, ok := data["objective_ids"].([]interface{})
+			if !ok || len(ids) != 2 {
+				return false
+			}
+			// position 5 should be clamped to position 2 (last), so order is [2, 1]
+			return ids[0] == "2" && ids[1] == "1"
+		},
+	})
+	err := runReorderShortcut(t, f, stdout, []string{
+		"+reorder",
+		"--level", "objective",
+		"--cycle-id", "123",
+		"--ops", `[{"id":"1","position":5}]`, // position 5 exceeds total of 2, should clamp to last
+	})
+	if err != nil {
+		t.Fatalf("unexpected error for out-of-range position (should clamp): %v", err)
+	}
+}
+
+func TestReorderExecute_IDNotFound(t *testing.T) {
+	t.Parallel()
+	f, stdout, _, reg := cmdutil.TestFactory(t, reorderTestConfig(t))
+	// Mock fetch objectives
+	reg.Register(&httpmock.Stub{
+		Method: "GET",
+		URL:    "/open-apis/okr/v2/cycles/123/objectives",
+		Body: map[string]interface{}{
+			"code": 0,
+			"msg":  "ok",
+			"data": map[string]interface{}{
+				"items": []interface{}{
+					map[string]interface{}{"id": "1", "position": 1, "cycle_id": "123", "owner": map[string]interface{}{"owner_type": "user"}},
+					map[string]interface{}{"id": "2", "position": 2, "cycle_id": "123", "owner": map[string]interface{}{"owner_type": "user"}},
+				},
+				"has_more": false,
+			},
+		},
+	})
+	err := runReorderShortcut(t, f, stdout, []string{
+		"+reorder",
+		"--level", "objective",
+		"--cycle-id", "123",
+		"--ops", `[{"id":"999","position":1}]`, // ID 999 doesn't exist
+	})
+	if err == nil {
+		t.Fatal("expected error for non-existent ID")
+	}
+	prob, ok := errs.ProblemOf(err)
+	if !ok {
+		t.Fatalf("expected typed error, got: %v", err)
+	}
+	if prob.Category != errs.CategoryValidation {
+		t.Fatalf("expected CategoryValidation, got %q", prob.Category)
+	}
+	var validationErr *errs.ValidationError
+	if !errors.As(err, &validationErr) {
+		t.Fatalf("expected error to be *errs.ValidationError, got: %T", err)
+	}
+	if !errors.Is(err, validationErr) {
+		t.Fatal("errors.Is should find the ValidationError in the chain")
+	}
+	if validationErr.Param != "--ops" {
+		t.Fatalf("expected param --ops, got %q", validationErr.Param)
+	}
+	if !strings.Contains(err.Error(), "not found") {
+		t.Fatalf("expected error to mention not found, got: %v", err)
+	}
+}
+
+// --- Unit tests for helper functions ---
+
+func TestParseReorderOps_Valid(t *testing.T) {
+	t.Parallel()
+	ops, err := parseReorderOps(`[{"id":"1","position":2},{"id":"2","position":1}]`)
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	if len(ops) != 2 {
+		t.Fatalf("expected 2 ops, got %d", len(ops))
+	}
+	if ops[0].ID != "1" || ops[0].Position != 2 {
+		t.Fatalf("expected op[0] = {1,2}, got %+v", ops[0])
+	}
+	if ops[1].ID != "2" || ops[1].Position != 1 {
+		t.Fatalf("expected op[1] = {2,1}, got %+v", ops[1])
+	}
+}
+
+func TestBuildReorderedIDs(t *testing.T) {
+	t.Parallel()
+	items := []Objective{
+		{ID: "1"},
+		{ID: "2"},
+		{ID: "3"},
+		{ID: "4"},
+	}
+	ops := []reorderOp{
+		{ID: "4", Position: 1},
+		{ID: "2", Position: 3},
+	}
+	result, err := buildReorderedIDs(items, ops, 4)
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	// Expected: 4 at pos1, 1 at pos2 (unchanged), 2 at pos3, 3 at pos4
+	expected := []string{"4", "1", "2", "3"}
+	if len(result) != len(expected) {
+		t.Fatalf("expected %d items, got %d", len(expected), len(result))
+	}
+	for i, id := range expected {
+		if result[i] != id {
+			t.Fatalf("expected result[%d] = %q, got %q", i, id, result[i])
+		}
+	}
+}
+
+func TestBuildReorderedIDs_SingleClampToEnd(t *testing.T) {
+	t.Parallel()
+	items := []testReorderItem{
+		{id: "1"}, {id: "2"}, {id: "3"}, {id: "4"},
+	}
+	ops := []reorderOp{
+		{ID: "1", Position: 99}, // position 99 exceeds total of 4, should clamp to last
+	}
+	result, err := buildReorderedIDs(items, ops, 4)
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	// Expected: 2 at pos1, 3 at pos2, 4 at pos3, 1 at pos4 (clamped)
+	expected := []string{"2", "3", "4", "1"}
+	for i, id := range expected {
+		if result[i] != id {
+			t.Fatalf("expected result[%d] = %q, got %q", i, id, result[i])
+		}
+	}
+}
+
+func TestBuildReorderedIDs_MultipleClampToEnd(t *testing.T) {
+	t.Parallel()
+	items := []testReorderItem{
+		{id: "1"}, {id: "2"}, {id: "3"}, {id: "4"}, {id: "5"},
+	}
+	ops := []reorderOp{
+		{ID: "1", Position: 10}, // position 10 exceeds total of 5
+		{ID: "2", Position: 20}, // position 20 exceeds total of 5
+	}
+	result, err := buildReorderedIDs(items, ops, 5)
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	// Expected: 3 at pos1, 4 at pos2, 5 at pos3, 1 at pos4 (clamped, pos10 < pos20), 2 at pos5 (clamped)
+	expected := []string{"3", "4", "5", "1", "2"}
+	for i, id := range expected {
+		if result[i] != id {
+			t.Fatalf("expected result[%d] = %q, got %q", i, id, result[i])
+		}
+	}
+}
+
+func TestBuildReorderedIDs_MixedClamp(t *testing.T) {
+	t.Parallel()
+	items := []testReorderItem{
+		{id: "1"}, {id: "2"}, {id: "3"}, {id: "4"}, {id: "5"},
+	}
+	ops := []reorderOp{
+		{ID: "5", Position: 1},  // normal position
+		{ID: "1", Position: 99}, // clamped to end
+		{ID: "2", Position: 50}, // clamped to end, but position 50 < 99, so comes before 1
+	}
+	result, err := buildReorderedIDs(items, ops, 5)
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	// Expected: 5 at pos1, 3 at pos2, 4 at pos3, 2 at pos4 (clamped pos50), 1 at pos5 (clamped pos99)
+	expected := []string{"5", "3", "4", "2", "1"}
+	for i, id := range expected {
+		if result[i] != id {
+			t.Fatalf("expected result[%d] = %q, got %q", i, id, result[i])
+		}
+	}
+}
+
+func TestBuildReorderedIDs_LargePositionSafe(t *testing.T) {
+	t.Parallel()
+	items := []testReorderItem{
+		{id: "1"}, {id: "2"}, {id: "3"},
+	}
+	// Very large position should not cause memory issues with map-based implementation
+	ops := []reorderOp{
+		{ID: "1", Position: 100000000}, // 10^8, would be dangerous with slice
+	}
+	result, err := buildReorderedIDs(items, ops, 3)
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	// Expected: 2 at pos1, 3 at pos2, 1 at pos3 (clamped to end)
+	expected := []string{"2", "3", "1"}
+	for i, id := range expected {
+		if result[i] != id {
+			t.Fatalf("expected result[%d] = %q, got %q", i, id, result[i])
+		}
+	}
+}

shortcuts/okr/okr_weight.go

@@ -0,0 +1,490 @@
+// Copyright (c) 2026 Lark Technologies Pte. Ltd.
+// SPDX-License-Identifier: MIT
+
+package okr
+
+import (
+	"context"
+	"encoding/json"
+	"fmt"
+	"io"
+	"math"
+	"sort"
+	"strconv"
+	"strings"
+
+	"github.com/larksuite/cli/errs"
+	"github.com/larksuite/cli/shortcuts/common"
+)
+
+// weightItem is the interface for items that have ID and weight.
+type weightItem interface {
+	GetID() string
+	GetWeight() float64
+}
+
+// weightOp represents a single weight assignment.
+type weightOp struct {
+	ID     string  `json:"id"`
+	Weight float64 `json:"weight"`
+}
+
+// parseWeightOps parses and validates the --weights JSON array.
+func parseWeightOps(weightsStr string) ([]weightOp, error) {
+	var ops []weightOp
+	if err := json.Unmarshal([]byte(weightsStr), &ops); err != nil {
+		return nil, errs.NewValidationError(errs.SubtypeInvalidArgument, "--weights must be valid JSON array: %s", err).WithParam("--weights").WithCause(err)
+	}
+	if len(ops) == 0 {
+		return nil, errs.NewValidationError(errs.SubtypeInvalidArgument, "--weights must contain at least one weight assignment").WithParam("--weights")
+	}
+
+	seen := make(map[string]bool)
+	var sum float64
+	for i, op := range ops {
+		if strings.TrimSpace(op.ID) == "" {
+			return nil, errs.NewValidationError(errs.SubtypeInvalidArgument, "weights[%d].id is required and cannot be empty", i).WithParam("--weights")
+		}
+		if op.Weight < 0 {
+			return nil, errs.NewValidationError(errs.SubtypeInvalidArgument, "weights[%d].weight must be non-negative", i).WithParam("--weights")
+		}
+		if op.Weight > 1 {
+			return nil, errs.NewValidationError(errs.SubtypeInvalidArgument, "weights[%d].weight must be <= 1", i).WithParam("--weights")
+		}
+		// Check for at most 3 decimal places
+		if math.Round(op.Weight*1000)/1000 != op.Weight {
+			return nil, errs.NewValidationError(errs.SubtypeInvalidArgument, "weights[%d].weight must have at most 3 decimal places", i).WithParam("--weights")
+		}
+		if seen[op.ID] {
+			return nil, errs.NewValidationError(errs.SubtypeInvalidArgument, "duplicate id %q in --weights", op.ID).WithParam("--weights")
+		}
+		seen[op.ID] = true
+		sum += op.Weight
+	}
+
+	// Sum must be <= 1
+	if sum > 1+1e-9 {
+		return nil, errs.NewValidationError(errs.SubtypeInvalidArgument, "sum of weights must be <= 1, got %.6f", sum).WithParam("--weights")
+	}
+
+	return ops, nil
+}
+
+// formatWeight formats a fixed-point weight value as a json.Number with exactly 3 decimal places.
+// This ensures precise JSON serialization and avoids float64 precision issues.
+func formatWeight(fp int64) json.Number {
+	return json.Number(fmt.Sprintf("%d.%03d", fp/1000, fp%1000))
+}
+
+// normalizeWeights normalizes weights using fixed-point arithmetic (×1000).
+//   - Specified weights are used as-is (already validated to 3 decimal places).
+//   - Remaining weight (1 - sum_specified) is distributed to unspecified items
+//     proportionally based on their original weights.
+//   - Fixed-point arithmetic ensures exact sum = 1, with residual added to the last item.
+//   - Weights are returned as json.Number to avoid float64 precision issues in JSON serialization.
+func normalizeWeights[T weightItem](
+	items []T,
+	ops []weightOp,
+) ([]map[string]interface{}, error) {
+	const scale = 1000 // fixed-point scale for 3 decimal places
+
+	// Build map of specified weights (as fixed-point integers)
+	specified := make(map[string]int64)
+	var specifiedSum int64
+	for _, op := range ops {
+		fp := int64(math.Round(op.Weight * scale))
+		specified[op.ID] = fp
+		specifiedSum += fp
+	}
+
+	// Calculate remaining weight to distribute (as fixed-point)
+	remaining := scale - specifiedSum
+	if remaining < 0 {
+		return nil, errs.NewInternalError(errs.SubtypeUnknown, "weight calculation error: remaining weight is negative")
+	}
+
+	// Collect unspecified items and their original weights
+	type itemWithWeight struct {
+		item T
+		fp   int64 // original weight as fixed-point
+	}
+	var unspecified []itemWithWeight
+	var originalUnspecifiedSum int64
+
+	for _, item := range items {
+		id := item.GetID()
+		if _, ok := specified[id]; ok {
+			continue
+		}
+		origWeight := item.GetWeight()
+		if origWeight < 0 {
+			origWeight = 0
+		}
+		fp := int64(math.Round(origWeight * scale))
+		unspecified = append(unspecified, itemWithWeight{item: item, fp: fp})
+		originalUnspecifiedSum += fp
+	}
+
+	// Distribute remaining weight proportionally
+	result := make([]map[string]interface{}, 0, len(items))
+	var resultSum int64
+
+	// First add specified items in original order
+	for _, item := range items {
+		id := item.GetID()
+		if fp, ok := specified[id]; ok {
+			result = append(result, map[string]interface{}{
+				"id":     id,
+				"weight": formatWeight(fp),
+			})
+			resultSum += fp
+		}
+	}
+
+	// Then distribute to unspecified items
+	if len(unspecified) > 0 && remaining > 0 {
+		if originalUnspecifiedSum == 0 {
+			// All original weights are zero, distribute evenly
+			perItem := remaining / int64(len(unspecified))
+			residual := remaining - perItem*int64(len(unspecified))
+
+			for i, uw := range unspecified {
+				fp := perItem
+				// Add residual to the last unspecified item
+				if i == len(unspecified)-1 {
+					fp += residual
+				}
+				result = append(result, map[string]interface{}{
+					"id":     uw.item.GetID(),
+					"weight": formatWeight(fp),
+				})
+				resultSum += fp
+			}
+		} else {
+			// Distribute proportionally based on original weights
+			var distributed int64
+			for i, uw := range unspecified {
+				var fp int64
+				if i == len(unspecified)-1 {
+					// Last item gets the remainder to ensure exact sum
+					fp = remaining - distributed
+				} else {
+					// Proportional distribution
+					fp = int64(float64(remaining) * float64(uw.fp) / float64(originalUnspecifiedSum))
+					distributed += fp
+				}
+				result = append(result, map[string]interface{}{
+					"id":     uw.item.GetID(),
+					"weight": formatWeight(fp),
+				})
+				resultSum += fp
+			}
+		}
+	} else if remaining > 0 {
+		// All items were specified, add residual to the last item
+		if len(result) > 0 {
+			lastIdx := len(result) - 1
+			// Parse current weight as fixed-point and add residual
+			var lastFP int64
+			if lastWeight, ok := result[lastIdx]["weight"].(json.Number); ok {
+				if f, err := lastWeight.Float64(); err == nil {
+					lastFP = int64(math.Round(f * scale))
+				}
+			}
+			result[lastIdx]["weight"] = formatWeight(lastFP + remaining)
+			resultSum += remaining
+		}
+	}
+
+	// Verify sum is exactly 1.0
+	if resultSum != scale {
+		return nil, errs.NewInternalError(errs.SubtypeUnknown,
+			"weight normalization error: sum is %.6f, expected 1.0", float64(resultSum)/scale)
+	}
+
+	return result, nil
+}
+
+// GetWeight implements the interface for Objective.
+func (o Objective) GetWeight() float64 {
+	if o.Weight == nil {
+		return 0
+	}
+	return *o.Weight
+}
+
+// GetWeight implements the interface for KeyResult.
+func (k KeyResult) GetWeight() float64 {
+	if k.Weight == nil {
+		return 0
+	}
+	return *k.Weight
+}
+
+// OKRWeight adjusts the weight of objectives or key results.
+var OKRWeight = common.Shortcut{
+	Service:     "okr",
+	Command:     "+weight",
+	Description: "Adjust the weight of OKR objectives or key results",
+	Risk:        "write",
+	Scopes:      []string{"okr:okr.content:writeonly"},
+	AuthTypes:   []string{"user", "bot"},
+	HasFormat:   true,
+	Flags: []common.Flag{
+		{Name: "level", Desc: "level to adjust: objective | key-result", Enum: []string{"objective", "key-result"}, Required: true},
+		{Name: "cycle-id", Desc: "OKR cycle ID (int64)", Required: true},
+		{Name: "objective-id", Desc: "objective ID (required when --level=key-result)"},
+		{Name: "weights", Desc: "JSON array of weight assignments: [{\"id\":\"...\",\"weight\":0.5}]", Input: []string{common.File, common.Stdin}, Required: true},
+	},
+	Validate: func(ctx context.Context, runtime *common.RuntimeContext) error {
+		level := runtime.Str("level")
+		if level != "objective" && level != "key-result" {
+			return errs.NewValidationError(errs.SubtypeInvalidArgument, "--level must be one of: objective | key-result").WithParam("--level")
+		}
+
+		cycleID := runtime.Str("cycle-id")
+		if id, err := strconv.ParseInt(cycleID, 10, 64); err != nil || id <= 0 {
+			return errs.NewValidationError(errs.SubtypeInvalidArgument, "--cycle-id must be a positive int64").WithParam("--cycle-id")
+		}
+
+		if level == "key-result" {
+			objID := runtime.Str("objective-id")
+			if objID == "" {
+				return errs.NewValidationError(errs.SubtypeInvalidArgument, "--objective-id is required when --level=key-result").WithParam("--objective-id")
+			}
+			if id, err := strconv.ParseInt(objID, 10, 64); err != nil || id <= 0 {
+				return errs.NewValidationError(errs.SubtypeInvalidArgument, "--objective-id must be a positive int64").WithParam("--objective-id")
+			}
+		}
+
+		weightsStr := runtime.Str("weights")
+		if err := common.RejectDangerousCharsTyped("--weights", weightsStr); err != nil {
+			return err
+		}
+		if _, err := parseWeightOps(weightsStr); err != nil {
+			return err
+		}
+
+		return nil
+	},
+	DryRun: func(ctx context.Context, runtime *common.RuntimeContext) *common.DryRunAPI {
+		level := runtime.Str("level")
+		cycleID := runtime.Str("cycle-id")
+		objectiveID := runtime.Str("objective-id")
+		ops, _ := parseWeightOps(runtime.Str("weights"))
+
+		apis := common.NewDryRunAPI()
+
+		if level == "objective" {
+			// First fetch objectives
+			listParams := map[string]interface{}{
+				"page_size": 100,
+			}
+			listPath := fmt.Sprintf("/open-apis/okr/v2/cycles/%s/objectives", cycleID)
+			apis = apis.
+				GET(listPath).
+				Params(listParams).
+				Desc("Fetch all objectives in the cycle to get current weights for normalization")
+
+			// Then update weights
+			weightParams := map[string]interface{}{
+				"cycle_id": cycleID,
+			}
+			// Build sample body
+			objectiveWeights := make([]map[string]interface{}, 0, len(ops))
+			for _, op := range ops {
+				objectiveWeights = append(objectiveWeights, map[string]interface{}{
+					"objective_id": op.ID,
+					"weight":       op.Weight,
+				})
+			}
+			weightBody := map[string]interface{}{
+				"objective_weights": objectiveWeights,
+			}
+			weightPath := fmt.Sprintf("/open-apis/okr/v2/cycles/%s/objectives_weight", cycleID)
+			apis = apis.
+				PUT(weightPath).
+				Params(weightParams).
+				Body(weightBody).
+				Desc("Update objective weights (full list sent after normalization)")
+		} else {
+			// key-result level
+			listParams := map[string]interface{}{
+				"page_size": 100,
+			}
+			listPath := fmt.Sprintf("/open-apis/okr/v2/objectives/%s/key_results", objectiveID)
+			apis = apis.
+				GET(listPath).
+				Params(listParams).
+				Desc("Fetch all key results for the objective to get current weights for normalization")
+
+			weightParams := map[string]interface{}{
+				"objective_id": objectiveID,
+			}
+			// Build sample body
+			keyResultWeights := make([]map[string]interface{}, 0, len(ops))
+			for _, op := range ops {
+				keyResultWeights = append(keyResultWeights, map[string]interface{}{
+					"key_result_id": op.ID,
+					"weight":        op.Weight,
+				})
+			}
+			weightBody := map[string]interface{}{
+				"key_result_weights": keyResultWeights,
+			}
+			weightPath := fmt.Sprintf("/open-apis/okr/v2/objectives/%s/key_results_weight", objectiveID)
+			apis = apis.
+				PUT(weightPath).
+				Params(weightParams).
+				Body(weightBody).
+				Desc("Update key result weights (full list sent after normalization)")
+		}
+
+		return apis
+	},
+	Execute: func(ctx context.Context, runtime *common.RuntimeContext) error {
+		level := runtime.Str("level")
+		cycleID := runtime.Str("cycle-id")
+		objectiveID := runtime.Str("objective-id")
+		ops, err := parseWeightOps(runtime.Str("weights"))
+		if err != nil {
+			return err
+		}
+
+		var normalizedWeights []map[string]interface{}
+		var total int
+
+		if level == "objective" {
+			objectives, err := fetchObjectives(ctx, runtime, cycleID)
+			if err != nil {
+				return err
+			}
+			total = len(objectives)
+
+			// Validate all specified IDs exist
+			objIDs := make(map[string]bool)
+			for _, obj := range objectives {
+				objIDs[obj.ID] = true
+			}
+			for _, op := range ops {
+				if !objIDs[op.ID] {
+					return errs.NewValidationError(errs.SubtypeInvalidArgument, "objective id %q not found in cycle", op.ID).WithParam("--weights")
+				}
+			}
+
+			normalizedWeights, err = normalizeWeights(objectives, ops)
+			if err != nil {
+				return err
+			}
+
+			// Build position map for sorting
+			posMap := make(map[string]int32)
+			for _, obj := range objectives {
+				if obj.Position != nil {
+					posMap[obj.ID] = *obj.Position
+				}
+			}
+
+			// Submit weight update
+			params := map[string]interface{}{
+				"cycle_id": cycleID,
+			}
+			objectiveWeights := make([]map[string]interface{}, 0, len(normalizedWeights))
+			for _, w := range normalizedWeights {
+				objectiveWeights = append(objectiveWeights, map[string]interface{}{
+					"objective_id": w["id"],
+					"weight":       w["weight"],
+				})
+			}
+			// Sort by position to match API requirements
+			sort.Slice(objectiveWeights, func(i, j int) bool {
+				idI := objectiveWeights[i]["objective_id"].(string)
+				idJ := objectiveWeights[j]["objective_id"].(string)
+				return posMap[idI] < posMap[idJ]
+			})
+			body := map[string]interface{}{
+				"objective_weights": objectiveWeights,
+			}
+			path := fmt.Sprintf("/open-apis/okr/v2/cycles/%s/objectives_weight", cycleID)
+			_, err = runtime.CallAPITyped("PUT", path, params, body)
+			if err != nil {
+				return wrapOkrNetworkErr(err, "failed to update objective weights")
+			}
+		} else {
+			// key-result level
+			keyResults, err := fetchKeyResults(ctx, runtime, objectiveID)
+			if err != nil {
+				return err
+			}
+			total = len(keyResults)
+
+			// Validate all specified IDs exist
+			krIDs := make(map[string]bool)
+			for _, kr := range keyResults {
+				krIDs[kr.ID] = true
+			}
+			for _, op := range ops {
+				if !krIDs[op.ID] {
+					return errs.NewValidationError(errs.SubtypeInvalidArgument, "key_result id %q not found in objective", op.ID).WithParam("--weights")
+				}
+			}
+
+			normalizedWeights, err = normalizeWeights(keyResults, ops)
+			if err != nil {
+				return err
+			}
+
+			// Build position map for sorting
+			posMap := make(map[string]int32)
+			for _, kr := range keyResults {
+				if kr.Position != nil {
+					posMap[kr.ID] = *kr.Position
+				}
+			}
+
+			// Submit weight update
+			params := map[string]interface{}{
+				"objective_id": objectiveID,
+			}
+			keyResultWeights := make([]map[string]interface{}, 0, len(normalizedWeights))
+			for _, w := range normalizedWeights {
+				keyResultWeights = append(keyResultWeights, map[string]interface{}{
+					"key_result_id": w["id"],
+					"weight":        w["weight"],
+				})
+			}
+			// Sort by position to match API requirements
+			sort.Slice(keyResultWeights, func(i, j int) bool {
+				idI := keyResultWeights[i]["key_result_id"].(string)
+				idJ := keyResultWeights[j]["key_result_id"].(string)
+				return posMap[idI] < posMap[idJ]
+			})
+			body := map[string]interface{}{
+				"key_result_weights": keyResultWeights,
+			}
+			path := fmt.Sprintf("/open-apis/okr/v2/objectives/%s/key_results_weight", objectiveID)
+			_, err = runtime.CallAPITyped("PUT", path, params, body)
+			if err != nil {
+				return wrapOkrNetworkErr(err, "failed to update key result weights")
+			}
+		}
+
+		// Build response
+		result := map[string]interface{}{
+			"level":    level,
+			"cycle_id": cycleID,
+			"total":    total,
+			"weights":  normalizedWeights,
+		}
+
+		runtime.OutFormat(result, nil, func(w io.Writer) {
+			fmt.Fprintf(w, "Successfully updated weights for %d %s(s)\n", total, level)
+			fmt.Fprintln(w, "Weights:")
+			for _, weightEntry := range normalizedWeights {
+				fmt.Fprintf(w, "  %s: %v\n", weightEntry["id"], weightEntry["weight"])
+			}
+		})
+
+		return nil
+	},
+}

shortcuts/okr/okr_weight_test.go

@@ -0,0 +1,747 @@
+// Copyright (c) 2026 Lark Technologies Pte. Ltd.
+// SPDX-License-Identifier: MIT
+
+package okr
+
+import (
+	"bytes"
+	"encoding/json"
+	"errors"
+	"math"
+	"strings"
+	"testing"
+
+	"github.com/larksuite/cli/errs"
+	"github.com/larksuite/cli/internal/cmdutil"
+	"github.com/larksuite/cli/internal/core"
+	"github.com/larksuite/cli/internal/httpmock"
+	"github.com/spf13/cobra"
+)
+
+// getWeightFloat extracts a float64 weight from either float64 or json.Number.
+func getWeightFloat(v interface{}) float64 {
+	switch val := v.(type) {
+	case float64:
+		return val
+	case json.Number:
+		f, _ := val.Float64()
+		return f
+	default:
+		return 0
+	}
+}
+
+func weightTestConfig(t *testing.T) *core.CliConfig {
+	t.Helper()
+	t.Setenv("LARKSUITE_CLI_CONFIG_DIR", t.TempDir())
+	return &core.CliConfig{
+		AppID:     "test-okr-weight",
+		AppSecret: "secret-okr-weight",
+		Brand:     core.BrandFeishu,
+	}
+}
+
+func runWeightShortcut(t *testing.T, f *cmdutil.Factory, stdout *bytes.Buffer, args []string) error {
+	t.Helper()
+	parent := &cobra.Command{Use: "okr"}
+	OKRWeight.Mount(parent, f)
+	parent.SetArgs(args)
+	parent.SilenceErrors = true
+	parent.SilenceUsage = true
+	if stdout != nil {
+		stdout.Reset()
+	}
+	return parent.Execute()
+}
+
+// --- Validate tests ---
+
+func TestWeightValidate_MissingLevel(t *testing.T) {
+	f, stdout, _, _ := cmdutil.TestFactory(t, weightTestConfig(t))
+	err := runWeightShortcut(t, f, stdout, []string{
+		"+weight",
+		"--cycle-id", "123",
+		"--weights", `[{"id":"1","weight":0.5}]`,
+	})
+	// cobra Required:true reports flag name without "--" prefix
+	if err == nil || !strings.Contains(err.Error(), "level") {
+		t.Fatalf("expected --level required error, got: %v", err)
+	}
+}
+
+func TestWeightValidate_InvalidLevel(t *testing.T) {
+	f, stdout, _, _ := cmdutil.TestFactory(t, weightTestConfig(t))
+	err := runWeightShortcut(t, f, stdout, []string{
+		"+weight",
+		"--level", "invalid",
+		"--cycle-id", "123",
+		"--weights", `[{"id":"1","weight":0.5}]`,
+	})
+	if err == nil {
+		t.Fatal("expected error for invalid --level")
+	}
+	prob, ok := errs.ProblemOf(err)
+	if !ok {
+		t.Fatalf("expected typed error, got: %v", err)
+	}
+	if prob.Category != errs.CategoryValidation {
+		t.Fatalf("expected CategoryValidation, got %q", prob.Category)
+	}
+	var validationErr *errs.ValidationError
+	if !errors.As(err, &validationErr) {
+		t.Fatalf("expected error to be *errs.ValidationError, got: %T", err)
+	}
+	if !errors.Is(err, validationErr) {
+		t.Fatal("errors.Is should find the ValidationError in the chain")
+	}
+	if validationErr.Param != "--level" {
+		t.Fatalf("expected param --level, got %q", validationErr.Param)
+	}
+}
+
+func TestWeightValidate_MissingCycleID(t *testing.T) {
+	f, stdout, _, _ := cmdutil.TestFactory(t, weightTestConfig(t))
+	err := runWeightShortcut(t, f, stdout, []string{
+		"+weight",
+		"--level", "objective",
+		"--weights", `[{"id":"1","weight":0.5}]`,
+	})
+	// cobra Required:true reports flag name without "--" prefix
+	if err == nil || !strings.Contains(err.Error(), "cycle-id") {
+		t.Fatalf("expected --cycle-id required error, got: %v", err)
+	}
+}
+
+func TestWeightValidate_MissingObjectiveIDForKRLevel(t *testing.T) {
+	f, stdout, _, _ := cmdutil.TestFactory(t, weightTestConfig(t))
+	err := runWeightShortcut(t, f, stdout, []string{
+		"+weight",
+		"--level", "key-result",
+		"--cycle-id", "123",
+		"--weights", `[{"id":"1","weight":0.5}]`,
+	})
+	if err == nil {
+		t.Fatal("expected error for missing --objective-id when --level=key-result")
+	}
+	prob, ok := errs.ProblemOf(err)
+	if !ok {
+		t.Fatalf("expected typed error, got: %v", err)
+	}
+	if prob.Category != errs.CategoryValidation {
+		t.Fatalf("expected CategoryValidation, got %q", prob.Category)
+	}
+	var validationErr *errs.ValidationError
+	if !errors.As(err, &validationErr) {
+		t.Fatalf("expected error to be *errs.ValidationError, got: %T", err)
+	}
+	if !errors.Is(err, validationErr) {
+		t.Fatal("errors.Is should find the ValidationError in the chain")
+	}
+	if validationErr.Param != "--objective-id" {
+		t.Fatalf("expected param --objective-id, got %q", validationErr.Param)
+	}
+}
+
+func TestWeightValidate_MissingWeights(t *testing.T) {
+	f, stdout, _, _ := cmdutil.TestFactory(t, weightTestConfig(t))
+	err := runWeightShortcut(t, f, stdout, []string{
+		"+weight",
+		"--level", "objective",
+		"--cycle-id", "123",
+	})
+	// cobra Required:true reports flag name without "--" prefix
+	if err == nil || !strings.Contains(err.Error(), "weights") {
+		t.Fatalf("expected --weights required error, got: %v", err)
+	}
+}
+
+func TestWeightValidate_InvalidWeightsJSON(t *testing.T) {
+	f, stdout, _, _ := cmdutil.TestFactory(t, weightTestConfig(t))
+	err := runWeightShortcut(t, f, stdout, []string{
+		"+weight",
+		"--level", "objective",
+		"--cycle-id", "123",
+		"--weights", "not-json",
+	})
+	if err == nil {
+		t.Fatal("expected error for invalid --weights JSON")
+	}
+	prob, ok := errs.ProblemOf(err)
+	if !ok {
+		t.Fatalf("expected typed error, got: %v", err)
+	}
+	if prob.Category != errs.CategoryValidation {
+		t.Fatalf("expected CategoryValidation, got %q", prob.Category)
+	}
+	var validationErr *errs.ValidationError
+	if !errors.As(err, &validationErr) {
+		t.Fatalf("expected error to be *errs.ValidationError, got: %T", err)
+	}
+	if !errors.Is(err, validationErr) {
+		t.Fatal("errors.Is should find the ValidationError in the chain")
+	}
+	if validationErr.Param != "--weights" {
+		t.Fatalf("expected param --weights, got %q", validationErr.Param)
+	}
+}
+
+func TestWeightValidate_EmptyWeightsArray(t *testing.T) {
+	f, stdout, _, _ := cmdutil.TestFactory(t, weightTestConfig(t))
+	err := runWeightShortcut(t, f, stdout, []string{
+		"+weight",
+		"--level", "objective",
+		"--cycle-id", "123",
+		"--weights", "[]",
+	})
+	if err == nil {
+		t.Fatal("expected error for empty --weights array")
+	}
+	prob, ok := errs.ProblemOf(err)
+	if !ok {
+		t.Fatalf("expected typed error, got: %v", err)
+	}
+	if prob.Category != errs.CategoryValidation {
+		t.Fatalf("expected CategoryValidation, got %q", prob.Category)
+	}
+	var validationErr *errs.ValidationError
+	if !errors.As(err, &validationErr) {
+		t.Fatalf("expected error to be *errs.ValidationError, got: %T", err)
+	}
+	if !errors.Is(err, validationErr) {
+		t.Fatal("errors.Is should find the ValidationError in the chain")
+	}
+	if validationErr.Param != "--weights" {
+		t.Fatalf("expected param --weights, got %q", validationErr.Param)
+	}
+}
+
+func TestWeightValidate_NegativeWeight(t *testing.T) {
+	f, stdout, _, _ := cmdutil.TestFactory(t, weightTestConfig(t))
+	err := runWeightShortcut(t, f, stdout, []string{
+		"+weight",
+		"--level", "objective",
+		"--cycle-id", "123",
+		"--weights", `[{"id":"1","weight":-0.1}]`,
+	})
+	if err == nil {
+		t.Fatal("expected error for negative weight")
+	}
+	prob, ok := errs.ProblemOf(err)
+	if !ok {
+		t.Fatalf("expected typed error, got: %v", err)
+	}
+	if prob.Category != errs.CategoryValidation {
+		t.Fatalf("expected CategoryValidation, got %q", prob.Category)
+	}
+	var validationErr *errs.ValidationError
+	if !errors.As(err, &validationErr) {
+		t.Fatalf("expected error to be *errs.ValidationError, got: %T", err)
+	}
+	if !errors.Is(err, validationErr) {
+		t.Fatal("errors.Is should find the ValidationError in the chain")
+	}
+	if validationErr.Param != "--weights" {
+		t.Fatalf("expected param --weights, got %q", validationErr.Param)
+	}
+	if !strings.Contains(err.Error(), "non-negative") {
+		t.Fatalf("expected error to mention non-negative, got: %v", err)
+	}
+}
+
+func TestWeightValidate_WeightGreaterThanOne(t *testing.T) {
+	f, stdout, _, _ := cmdutil.TestFactory(t, weightTestConfig(t))
+	err := runWeightShortcut(t, f, stdout, []string{
+		"+weight",
+		"--level", "objective",
+		"--cycle-id", "123",
+		"--weights", `[{"id":"1","weight":1.5}]`,
+	})
+	if err == nil {
+		t.Fatal("expected error for weight > 1")
+	}
+	prob, ok := errs.ProblemOf(err)
+	if !ok {
+		t.Fatalf("expected typed error, got: %v", err)
+	}
+	if prob.Category != errs.CategoryValidation {
+		t.Fatalf("expected CategoryValidation, got %q", prob.Category)
+	}
+	var validationErr *errs.ValidationError
+	if !errors.As(err, &validationErr) {
+		t.Fatalf("expected error to be *errs.ValidationError, got: %T", err)
+	}
+	if !errors.Is(err, validationErr) {
+		t.Fatal("errors.Is should find the ValidationError in the chain")
+	}
+	if validationErr.Param != "--weights" {
+		t.Fatalf("expected param --weights, got %q", validationErr.Param)
+	}
+}
+
+func TestWeightValidate_TooManyDecimalPlaces(t *testing.T) {
+	f, stdout, _, _ := cmdutil.TestFactory(t, weightTestConfig(t))
+	err := runWeightShortcut(t, f, stdout, []string{
+		"+weight",
+		"--level", "objective",
+		"--cycle-id", "123",
+		"--weights", `[{"id":"1","weight":0.1234}]`,
+	})
+	if err == nil {
+		t.Fatal("expected error for weight with more than 3 decimal places")
+	}
+	prob, ok := errs.ProblemOf(err)
+	if !ok {
+		t.Fatalf("expected typed error, got: %v", err)
+	}
+	if prob.Category != errs.CategoryValidation {
+		t.Fatalf("expected CategoryValidation, got %q", prob.Category)
+	}
+	var validationErr *errs.ValidationError
+	if !errors.As(err, &validationErr) {
+		t.Fatalf("expected error to be *errs.ValidationError, got: %T", err)
+	}
+	if !errors.Is(err, validationErr) {
+		t.Fatal("errors.Is should find the ValidationError in the chain")
+	}
+	if validationErr.Param != "--weights" {
+		t.Fatalf("expected param --weights, got %q", validationErr.Param)
+	}
+	if !strings.Contains(err.Error(), "3 decimal places") {
+		t.Fatalf("expected error to mention 3 decimal places, got: %v", err)
+	}
+}
+
+func TestWeightValidate_SumGreaterThanOne(t *testing.T) {
+	f, stdout, _, _ := cmdutil.TestFactory(t, weightTestConfig(t))
+	err := runWeightShortcut(t, f, stdout, []string{
+		"+weight",
+		"--level", "objective",
+		"--cycle-id", "123",
+		"--weights", `[{"id":"1","weight":0.6},{"id":"2","weight":0.5}]`,
+	})
+	if err == nil {
+		t.Fatal("expected error for sum of weights > 1")
+	}
+	prob, ok := errs.ProblemOf(err)
+	if !ok {
+		t.Fatalf("expected typed error, got: %v", err)
+	}
+	if prob.Category != errs.CategoryValidation {
+		t.Fatalf("expected CategoryValidation, got %q", prob.Category)
+	}
+	var validationErr *errs.ValidationError
+	if !errors.As(err, &validationErr) {
+		t.Fatalf("expected error to be *errs.ValidationError, got: %T", err)
+	}
+	if !errors.Is(err, validationErr) {
+		t.Fatal("errors.Is should find the ValidationError in the chain")
+	}
+	if validationErr.Param != "--weights" {
+		t.Fatalf("expected param --weights, got %q", validationErr.Param)
+	}
+	if !strings.Contains(err.Error(), "sum of weights") {
+		t.Fatalf("expected error to mention sum of weights, got: %v", err)
+	}
+}
+
+func TestWeightValidate_DuplicateID(t *testing.T) {
+	f, stdout, _, _ := cmdutil.TestFactory(t, weightTestConfig(t))
+	err := runWeightShortcut(t, f, stdout, []string{
+		"+weight",
+		"--level", "objective",
+		"--cycle-id", "123",
+		"--weights", `[{"id":"1","weight":0.3},{"id":"1","weight":0.4}]`,
+	})
+	if err == nil {
+		t.Fatal("expected error for duplicate id in --weights")
+	}
+	prob, ok := errs.ProblemOf(err)
+	if !ok {
+		t.Fatalf("expected typed error, got: %v", err)
+	}
+	if prob.Category != errs.CategoryValidation {
+		t.Fatalf("expected CategoryValidation, got %q", prob.Category)
+	}
+	var validationErr *errs.ValidationError
+	if !errors.As(err, &validationErr) {
+		t.Fatalf("expected error to be *errs.ValidationError, got: %T", err)
+	}
+	if !errors.Is(err, validationErr) {
+		t.Fatal("errors.Is should find the ValidationError in the chain")
+	}
+	if validationErr.Param != "--weights" {
+		t.Fatalf("expected param --weights, got %q", validationErr.Param)
+	}
+	if !strings.Contains(err.Error(), "duplicate id") {
+		t.Fatalf("expected error to mention duplicate id, got: %v", err)
+	}
+}
+
+// --- DryRun tests ---
+
+func TestWeightDryRun_Objectives(t *testing.T) {
+	f, stdout, _, _ := cmdutil.TestFactory(t, weightTestConfig(t))
+	err := runWeightShortcut(t, f, stdout, []string{
+		"+weight",
+		"--level", "objective",
+		"--cycle-id", "123",
+		"--weights", `[{"id":"1","weight":0.5},{"id":"2","weight":0.5}]`,
+		"--dry-run",
+	})
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	output := stdout.String()
+	if !strings.Contains(output, "/open-apis/okr/v2/cycles/123/objectives") {
+		t.Fatalf("dry-run output should contain objectives list API path, got: %s", output)
+	}
+	if !strings.Contains(output, "GET") {
+		t.Fatalf("dry-run output should contain GET method for list, got: %s", output)
+	}
+	if !strings.Contains(output, "/open-apis/okr/v2/cycles/123/objectives_weight") {
+		t.Fatalf("dry-run output should contain weight update API path, got: %s", output)
+	}
+	if !strings.Contains(output, "PUT") {
+		t.Fatalf("dry-run output should contain PUT method for update, got: %s", output)
+	}
+	if !strings.Contains(output, "objective_weights") {
+		t.Fatalf("dry-run output should contain objective_weights in body, got: %s", output)
+	}
+}
+
+func TestWeightDryRun_KeyResults(t *testing.T) {
+	f, stdout, _, _ := cmdutil.TestFactory(t, weightTestConfig(t))
+	err := runWeightShortcut(t, f, stdout, []string{
+		"+weight",
+		"--level", "key-result",
+		"--cycle-id", "123",
+		"--objective-id", "456",
+		"--weights", `[{"id":"kr1","weight":0.5},{"id":"kr2","weight":0.5}]`,
+		"--dry-run",
+	})
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	output := stdout.String()
+	if !strings.Contains(output, "/open-apis/okr/v2/objectives/456/key_results") {
+		t.Fatalf("dry-run output should contain key_results list API path, got: %s", output)
+	}
+	if !strings.Contains(output, "/open-apis/okr/v2/objectives/456/key_results_weight") {
+		t.Fatalf("dry-run output should contain key_results weight update API path, got: %s", output)
+	}
+	if !strings.Contains(output, "key_result_weights") {
+		t.Fatalf("dry-run output should contain key_result_weights in body, got: %s", output)
+	}
+}
+
+// --- Execute tests ---
+
+func TestWeightExecute_Objectives_Success(t *testing.T) {
+	f, stdout, _, reg := cmdutil.TestFactory(t, weightTestConfig(t))
+	// Mock fetch objectives
+	w1 := 0.5
+	w2 := 0.5
+	reg.Register(&httpmock.Stub{
+		Method: "GET",
+		URL:    "/open-apis/okr/v2/cycles/123/objectives",
+		Body: map[string]interface{}{
+			"code": 0,
+			"msg":  "ok",
+			"data": map[string]interface{}{
+				"items": []interface{}{
+					map[string]interface{}{"id": "1", "weight": &w1, "cycle_id": "123", "owner": map[string]interface{}{"owner_type": "user"}},
+					map[string]interface{}{"id": "2", "weight": &w2, "cycle_id": "123", "owner": map[string]interface{}{"owner_type": "user"}},
+				},
+				"has_more": false,
+			},
+		},
+	})
+	// Mock weight update
+	reg.Register(&httpmock.Stub{
+		Method: "PUT",
+		URL:    "/open-apis/okr/v2/cycles/123/objectives_weight",
+		Body: map[string]interface{}{
+			"code": 0,
+			"msg":  "ok",
+			"data": map[string]interface{}{
+				"items": []interface{}{
+					map[string]interface{}{"id": "1", "weight": 0.7},
+					map[string]interface{}{"id": "2", "weight": 0.3},
+				},
+			},
+		},
+	})
+	err := runWeightShortcut(t, f, stdout, []string{
+		"+weight",
+		"--level", "objective",
+		"--cycle-id", "123",
+		"--weights", `[{"id":"1","weight":0.7}]`,
+	})
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+
+	// Verify output
+	data := decodeEnvelope(t, stdout)
+	if data["level"] != "objective" {
+		t.Fatalf("expected level=objective, got %v", data["level"])
+	}
+	if data["cycle_id"] != "123" {
+		t.Fatalf("expected cycle_id=123, got %v", data["cycle_id"])
+	}
+	weights, _ := data["weights"].([]interface{})
+	if len(weights) != 2 {
+		t.Fatalf("expected 2 items in weights list, got %d", len(weights))
+	}
+
+	// Verify sum is exactly 1.0
+	var sum float64
+	for _, w := range weights {
+		wm, _ := w.(map[string]interface{})
+		weightVal := getWeightFloat(wm["weight"])
+		sum += weightVal
+	}
+	if math.Abs(sum-1.0) > 1e-9 {
+		t.Fatalf("expected sum of weights = 1.0, got %.10f", sum)
+	}
+}
+
+func TestWeightExecute_KeyResults_Success(t *testing.T) {
+	f, stdout, _, reg := cmdutil.TestFactory(t, weightTestConfig(t))
+	// Mock fetch key results
+	w1 := 0.3
+	w2 := 0.3
+	w3 := 0.4
+	reg.Register(&httpmock.Stub{
+		Method: "GET",
+		URL:    "/open-apis/okr/v2/objectives/456/key_results",
+		Body: map[string]interface{}{
+			"code": 0,
+			"msg":  "ok",
+			"data": map[string]interface{}{
+				"items": []interface{}{
+					map[string]interface{}{"id": "kr1", "weight": &w1, "objective_id": "456", "owner": map[string]interface{}{"owner_type": "user"}},
+					map[string]interface{}{"id": "kr2", "weight": &w2, "objective_id": "456", "owner": map[string]interface{}{"owner_type": "user"}},
+					map[string]interface{}{"id": "kr3", "weight": &w3, "objective_id": "456", "owner": map[string]interface{}{"owner_type": "user"}},
+				},
+				"has_more": false,
+			},
+		},
+	})
+	// Mock weight update
+	reg.Register(&httpmock.Stub{
+		Method: "PUT",
+		URL:    "/open-apis/okr/v2/objectives/456/key_results_weight",
+		Body: map[string]interface{}{
+			"code": 0,
+			"msg":  "ok",
+			"data": map[string]interface{}{
+				"items": []interface{}{},
+			},
+		},
+	})
+	err := runWeightShortcut(t, f, stdout, []string{
+		"+weight",
+		"--level", "key-result",
+		"--cycle-id", "123",
+		"--objective-id", "456",
+		"--weights", `[{"id":"kr1","weight":0.5}]`,
+	})
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+
+	// Verify output
+	data := decodeEnvelope(t, stdout)
+	if data["level"] != "key-result" {
+		t.Fatalf("expected level=key-result, got %v", data["level"])
+	}
+	if data["cycle_id"] != "123" {
+		t.Fatalf("expected cycle_id=123, got %v", data["cycle_id"])
+	}
+	weights, _ := data["weights"].([]interface{})
+
+	// Verify sum is exactly 1.0
+	var sum float64
+	for _, w := range weights {
+		wm, _ := w.(map[string]interface{})
+		weightVal := getWeightFloat(wm["weight"])
+		sum += weightVal
+	}
+	if math.Abs(sum-1.0) > 1e-9 {
+		t.Fatalf("expected sum of weights = 1.0, got %.10f", sum)
+	}
+}
+
+func TestWeightExecute_IDNotFound(t *testing.T) {
+	f, stdout, _, reg := cmdutil.TestFactory(t, weightTestConfig(t))
+	// Mock fetch objectives
+	w1 := 0.5
+	reg.Register(&httpmock.Stub{
+		Method: "GET",
+		URL:    "/open-apis/okr/v2/cycles/123/objectives",
+		Body: map[string]interface{}{
+			"code": 0,
+			"msg":  "ok",
+			"data": map[string]interface{}{
+				"items": []interface{}{
+					map[string]interface{}{"id": "1", "weight": &w1, "cycle_id": "123", "owner": map[string]interface{}{"owner_type": "user"}},
+				},
+				"has_more": false,
+			},
+		},
+	})
+	err := runWeightShortcut(t, f, stdout, []string{
+		"+weight",
+		"--level", "objective",
+		"--cycle-id", "123",
+		"--weights", `[{"id":"999","weight":0.5}]`,
+	})
+	if err == nil {
+		t.Fatal("expected error for non-existent ID")
+	}
+	prob, ok := errs.ProblemOf(err)
+	if !ok {
+		t.Fatalf("expected typed error, got: %v", err)
+	}
+	if prob.Category != errs.CategoryValidation {
+		t.Fatalf("expected CategoryValidation, got %q", prob.Category)
+	}
+	var validationErr *errs.ValidationError
+	if !errors.As(err, &validationErr) {
+		t.Fatalf("expected error to be *errs.ValidationError, got: %T", err)
+	}
+	if !errors.Is(err, validationErr) {
+		t.Fatal("errors.Is should find the ValidationError in the chain")
+	}
+	if validationErr.Param != "--weights" {
+		t.Fatalf("expected param --weights, got %q", validationErr.Param)
+	}
+	if !strings.Contains(err.Error(), "not found") {
+		t.Fatalf("expected error to mention not found, got: %v", err)
+	}
+}
+
+// --- Unit tests for helper functions ---
+
+func TestParseWeightOps_Valid(t *testing.T) {
+	ops, err := parseWeightOps(`[{"id":"1","weight":0.3},{"id":"2","weight":0.7}]`)
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	if len(ops) != 2 {
+		t.Fatalf("expected 2 ops, got %d", len(ops))
+	}
+	if ops[0].ID != "1" || math.Abs(ops[0].Weight-0.3) > 1e-9 {
+		t.Fatalf("expected op[0] = {1,0.3}, got %+v", ops[0])
+	}
+	if ops[1].ID != "2" || math.Abs(ops[1].Weight-0.7) > 1e-9 {
+		t.Fatalf("expected op[1] = {2,0.7}, got %+v", ops[1])
+	}
+}
+
+func TestNormalizeWeights_AllSpecified(t *testing.T) {
+	w1 := 0.0
+	w2 := 0.0
+	items := []Objective{
+		{ID: "1", Weight: &w1},
+		{ID: "2", Weight: &w2},
+	}
+	ops := []weightOp{
+		{ID: "1", Weight: 0.3},
+		{ID: "2", Weight: 0.7},
+	}
+	result, err := normalizeWeights(items, ops)
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	if len(result) != 2 {
+		t.Fatalf("expected 2 results, got %d", len(result))
+	}
+	// Sum should be exactly 1.0
+	var sum float64
+	for _, r := range result {
+		sum += getWeightFloat(r["weight"])
+	}
+	if math.Abs(sum-1.0) > 1e-9 {
+		t.Fatalf("expected sum = 1.0, got %.10f", sum)
+	}
+}
+
+func TestNormalizeWeights_PartialSpecified_Proportional(t *testing.T) {
+	w1 := 0.5
+	w2 := 0.3
+	w3 := 0.2
+	items := []Objective{
+		{ID: "1", Weight: &w1},
+		{ID: "2", Weight: &w2},
+		{ID: "3", Weight: &w3},
+	}
+	ops := []weightOp{
+		{ID: "1", Weight: 0.4}, // Specify 0.4 for item 1
+	}
+	result, err := normalizeWeights(items, ops)
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	if len(result) != 3 {
+		t.Fatalf("expected 3 results, got %d", len(result))
+	}
+	// Sum should be exactly 1.0
+	var sum float64
+	for _, r := range result {
+		sum += getWeightFloat(r["weight"])
+	}
+	if math.Abs(sum-1.0) > 1e-9 {
+		t.Fatalf("expected sum = 1.0, got %.10f", sum)
+	}
+	// Item 1 should have weight 0.4
+	var item1Weight float64
+	for _, r := range result {
+		if r["id"] == "1" {
+			item1Weight = getWeightFloat(r["weight"])
+			break
+		}
+	}
+	if math.Abs(item1Weight-0.4) > 1e-9 {
+		t.Fatalf("expected item 1 weight = 0.4, got %.10f", item1Weight)
+	}
+}
+
+func TestNormalizeWeights_ZeroOriginalWeights(t *testing.T) {
+	w1 := 0.0
+	w2 := 0.0
+	items := []Objective{
+		{ID: "1", Weight: &w1},
+		{ID: "2", Weight: &w2},
+	}
+	ops := []weightOp{
+		{ID: "1", Weight: 0.5},
+	}
+	result, err := normalizeWeights(items, ops)
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	if len(result) != 2 {
+		t.Fatalf("expected 2 results, got %d", len(result))
+	}
+	// Sum should be exactly 1.0
+	var sum float64
+	for _, r := range result {
+		sum += getWeightFloat(r["weight"])
+	}
+	if math.Abs(sum-1.0) > 1e-9 {
+		t.Fatalf("expected sum = 1.0, got %.10f", sum)
+	}
+	// When original weights are zero, remaining should be distributed evenly
+	var item2Weight float64
+	for _, r := range result {
+		if r["id"] == "2" {
+			item2Weight = getWeightFloat(r["weight"])
+			break
+		}
+	}
+	if math.Abs(item2Weight-0.5) > 1e-9 {
+		t.Fatalf("expected item 2 weight = 0.5 (even distribution), got %.10f", item2Weight)
+	}
+}

shortcuts/okr/shortcuts.go

@@ -18,5 +18,9 @@ func Shortcuts() []common.Shortcut {
 		OKRUpdateProgressRecord,
 		OKRDeleteProgressRecord,
 		OKRUploadImage,
+		OKRBatchCreate,
+		OKRReorder,
+		OKRWeight,
+		OKRIndicatorUpdate,
 	}
 }

shortcuts/slides/shortcuts.go

@@ -11,5 +11,6 @@ func Shortcuts() []common.Shortcut {
 		SlidesCreate,
 		SlidesMediaUpload,
 		SlidesReplaceSlide,
+		SlidesScreenshot,
 	}
 }

shortcuts/slides/slides_screenshot.go

@@ -0,0 +1,537 @@
+// Copyright (c) 2026 Lark Technologies Pte. Ltd.
+// SPDX-License-Identifier: MIT
+
+package slides
+
+import (
+	"bytes"
+	"context"
+	"encoding/base64"
+	"fmt"
+	"os"
+	"path/filepath"
+	"regexp"
+	"strings"
+
+	larkcore "github.com/larksuite/oapi-sdk-go/v3/core"
+
+	"github.com/larksuite/cli/errs"
+	"github.com/larksuite/cli/extension/fileio"
+	"github.com/larksuite/cli/internal/util"
+	"github.com/larksuite/cli/internal/validate"
+	"github.com/larksuite/cli/shortcuts/common"
+)
+
+const defaultSlidesScreenshotDir = ".lark-slides/screenshots"
+
+var unsafeScreenshotFileCharRegex = regexp.MustCompile(`[^A-Za-z0-9._-]+`)
+
+// SlidesScreenshot fetches server-rendered slide screenshots and writes them to
+// local files. The raw API returns Base64 image payloads; this shortcut keeps
+// those payloads out of stdout so agents only see small file metadata.
+var SlidesScreenshot = common.Shortcut{
+	Service:     "slides",
+	Command:     "+screenshot",
+	Description: "Save slide screenshots to local files without printing Base64 image data",
+	Risk:        "read",
+	Scopes:      []string{"slides:presentation:screenshot"},
+	// wiki:node:read is required only when --presentation is a wiki URL.
+	ConditionalScopes: []string{"wiki:node:read"},
+	AuthTypes:         []string{"user", "bot"},
+	Flags: []common.Flag{
+		{Name: "presentation", Desc: "xml_presentation_id, slides URL, or wiki URL that resolves to slides; list mode only"},
+		{Name: "slide-id", Type: "string_array", Desc: "slide page identifier (repeat for multiple slides)"},
+		{Name: "slide-number", Type: "int_array", Desc: "slide page number (repeat for multiple slides)"},
+		{Name: "content", Desc: "slide XML content to render directly instead of fetching existing slides", Input: []string{common.File, common.Stdin}},
+		{Name: "output-dir", Default: defaultSlidesScreenshotDir, Desc: "relative directory for saved screenshots"},
+		{Name: "output-name", Desc: "file name stem for --content render output"},
+	},
+	Validate: func(ctx context.Context, runtime *common.RuntimeContext) error {
+		renderMode := runtime.Changed("content")
+		if renderMode {
+			if strings.TrimSpace(runtime.Str("content")) == "" {
+				return slidesScreenshotFlagErrorf("--content cannot be empty")
+			}
+			if len(normalizeSlideIDs(runtime.StrArray("slide-id"))) > 0 || len(runtime.IntArray("slide-number")) > 0 {
+				return slidesScreenshotFlagErrorf("--content cannot be used with --slide-id or --slide-number")
+			}
+			if runtime.Changed("presentation") {
+				return slidesScreenshotFlagErrorf("--presentation cannot be used with --content")
+			}
+		} else {
+			ref, err := parsePresentationRef(runtime.Str("presentation"))
+			if err != nil {
+				return err
+			}
+			if ref.Kind == "wiki" {
+				if err := runtime.EnsureScopes([]string{"wiki:node:read"}); err != nil {
+					return err
+				}
+			}
+			if _, err := normalizeSlideNumbers(runtime.IntArray("slide-number")); err != nil {
+				return err
+			}
+			if !hasSlideScreenshotSelector(runtime) {
+				return slidesScreenshotFlagErrorf("--slide-id or --slide-number is required")
+			}
+		}
+		if _, err := validateScreenshotOutputDir(runtime, runtime.Str("output-dir")); err != nil {
+			return err
+		}
+		return nil
+	},
+	DryRun: func(ctx context.Context, runtime *common.RuntimeContext) *common.DryRunAPI {
+		if runtime.Changed("content") {
+			return dryRunRenderScreenshot(runtime)
+		}
+		ref, err := parsePresentationRef(runtime.Str("presentation"))
+		if err != nil {
+			return common.NewDryRunAPI().Set("error", err.Error())
+		}
+		slideIDs := normalizeSlideIDs(runtime.StrArray("slide-id"))
+		slideNumbers, err := normalizeSlideNumbers(runtime.IntArray("slide-number"))
+		if err != nil {
+			return common.NewDryRunAPI().Set("error", err.Error())
+		}
+		if len(slideIDs) == 0 && len(slideNumbers) == 0 {
+			return common.NewDryRunAPI().Set("error", "--slide-id or --slide-number is required")
+		}
+
+		presentationID := ref.Token
+		dry := common.NewDryRunAPI()
+		if ref.Kind == "wiki" {
+			presentationID = "<resolved_slides_token>"
+			dry.Desc("2-step orchestration: resolve wiki → fetch slide screenshot(s)").
+				GET("/open-apis/wiki/v2/spaces/get_node").
+				Desc("[1] Resolve wiki node to slides presentation").
+				Params(map[string]interface{}{"token": ref.Token})
+		} else {
+			dry.Desc(fmt.Sprintf("Fetch %d slide screenshot(s) and save files under %s", len(slideIDs)+len(slideNumbers), runtime.Str("output-dir")))
+		}
+		body := map[string]interface{}{}
+		if len(slideIDs) > 0 {
+			body["slide_ids"] = slideIDs
+		}
+		if len(slideNumbers) > 0 {
+			body["slide_numbers"] = slideNumbers
+		}
+		dry.POST(fmt.Sprintf(
+			"/open-apis/slides_ai/v1/xml_presentations/%s/slide_images",
+			validate.EncodePathSegment(presentationID),
+		)).
+			Body(body)
+		return dry.Set("output_dir", runtime.Str("output-dir")).Set("base64_output", "suppressed; decoded to local files during execution")
+	},
+	Execute: func(ctx context.Context, runtime *common.RuntimeContext) error {
+		if runtime.Changed("content") {
+			return executeRenderScreenshot(runtime)
+		}
+		ref, err := parsePresentationRef(runtime.Str("presentation"))
+		if err != nil {
+			return err
+		}
+		presentationID, err := resolvePresentationID(runtime, ref)
+		if err != nil {
+			return err
+		}
+
+		slideIDs := normalizeSlideIDs(runtime.StrArray("slide-id"))
+		slideNumbers, err := normalizeSlideNumbers(runtime.IntArray("slide-number"))
+		if err != nil {
+			return err
+		}
+		if len(slideIDs) == 0 && len(slideNumbers) == 0 {
+			return slidesScreenshotFlagErrorf("--slide-id or --slide-number is required")
+		}
+		outputDir := runtime.Str("output-dir")
+		safeOutputDir, err := ensureScreenshotOutputDir(runtime, outputDir)
+		if err != nil {
+			return err
+		}
+
+		url := fmt.Sprintf(
+			"/open-apis/slides_ai/v1/xml_presentations/%s/slide_images",
+			validate.EncodePathSegment(presentationID),
+		)
+		query := larkcore.QueryParams{}
+		body := map[string]interface{}{}
+		if len(slideIDs) > 0 {
+			body["slide_ids"] = slideIDs
+		}
+		if len(slideNumbers) > 0 {
+			body["slide_numbers"] = slideNumbers
+		}
+		data, err := doSlidesScreenshotAPIJSONWithLogID(runtime, "POST", url, query, body)
+		if err != nil {
+			return enrichSlidesScreenshotSelectorError(err, slideNumbers)
+		}
+
+		saved, err := saveSlideScreenshots(runtime, data, safeOutputDir, presentationID)
+		if err != nil {
+			return err
+		}
+		runtime.Out(map[string]interface{}{
+			"xml_presentation_id": presentationID,
+			"output_dir":          outputDir,
+			"screenshots":         saved,
+		}, nil)
+		return nil
+	},
+}
+
+func dryRunRenderScreenshot(runtime *common.RuntimeContext) *common.DryRunAPI {
+	content := runtime.Str("content")
+	if strings.TrimSpace(content) == "" {
+		return common.NewDryRunAPI().Set("error", "--content cannot be empty")
+	}
+	if len(normalizeSlideIDs(runtime.StrArray("slide-id"))) > 0 || len(runtime.IntArray("slide-number")) > 0 {
+		return common.NewDryRunAPI().Set("error", "--content cannot be used with --slide-id or --slide-number")
+	}
+	if runtime.Changed("presentation") {
+		return common.NewDryRunAPI().Set("error", "--presentation cannot be used with --content")
+	}
+	dry := common.NewDryRunAPI().Desc("Render slide XML content to a screenshot file")
+	dry.POST("/open-apis/slides_ai/v1/slide_image/render").
+		Body(map[string]interface{}{
+			"content": fmt.Sprintf("<xml omitted; length=%d>", len(content)),
+		})
+	return dry.Set("output_dir", runtime.Str("output-dir")).Set("base64_output", "suppressed; decoded to local file during execution")
+}
+
+func executeRenderScreenshot(runtime *common.RuntimeContext) error {
+	content := runtime.Str("content")
+	if strings.TrimSpace(content) == "" {
+		return slidesScreenshotFlagErrorf("--content cannot be empty")
+	}
+	if len(normalizeSlideIDs(runtime.StrArray("slide-id"))) > 0 || len(runtime.IntArray("slide-number")) > 0 {
+		return slidesScreenshotFlagErrorf("--content cannot be used with --slide-id or --slide-number")
+	}
+	if runtime.Changed("presentation") {
+		return slidesScreenshotFlagErrorf("--presentation cannot be used with --content")
+	}
+	outputDir := runtime.Str("output-dir")
+	safeOutputDir, err := ensureScreenshotOutputDir(runtime, outputDir)
+	if err != nil {
+		return err
+	}
+
+	data, err := doSlidesScreenshotAPIJSONWithLogID(runtime, "POST", "/open-apis/slides_ai/v1/slide_image/render", larkcore.QueryParams{}, map[string]interface{}{
+		"content": content,
+	})
+	if err != nil {
+		return err
+	}
+	saved, err := saveRenderedSlideScreenshot(runtime, data, safeOutputDir, runtime.Str("output-name"))
+	if err != nil {
+		return err
+	}
+	runtime.Out(map[string]interface{}{
+		"output_dir":  outputDir,
+		"screenshots": saved,
+	}, nil)
+	return nil
+}
+
+func normalizeSlideIDs(values []string) []string {
+	out := make([]string, 0, len(values))
+	seen := map[string]struct{}{}
+	for _, v := range values {
+		s := strings.TrimSpace(v)
+		if s == "" {
+			continue
+		}
+		if _, ok := seen[s]; ok {
+			continue
+		}
+		seen[s] = struct{}{}
+		out = append(out, s)
+	}
+	return out
+}
+
+func normalizeSlideNumbers(values []int) ([]int, error) {
+	out := make([]int, 0, len(values))
+	seen := map[int]struct{}{}
+	for _, n := range values {
+		if n < 1 {
+			return nil, slidesScreenshotFlagErrorf("--slide-number must be a positive integer")
+		}
+		if _, ok := seen[n]; ok {
+			continue
+		}
+		seen[n] = struct{}{}
+		out = append(out, n)
+	}
+	return out, nil
+}
+
+func hasSlideScreenshotSelector(runtime *common.RuntimeContext) bool {
+	return len(normalizeSlideIDs(runtime.StrArray("slide-id"))) > 0 || len(runtime.IntArray("slide-number")) > 0
+}
+
+func slidesScreenshotFlagErrorf(format string, args ...interface{}) error {
+	return errs.NewValidationError(errs.SubtypeInvalidArgument, format, args...)
+}
+
+func validateScreenshotOutputDir(runtime *common.RuntimeContext, outputDir string) (string, error) {
+	if _, err := runtime.ResolveSavePath(filepath.Join(outputDir, "probe.png")); err != nil {
+		return "", slidesScreenshotFlagErrorf("--output-dir invalid: %v", err)
+	}
+	return outputDir, nil
+}
+
+func ensureScreenshotOutputDir(runtime *common.RuntimeContext, outputDir string) (string, error) {
+	return validateScreenshotOutputDir(runtime, outputDir)
+}
+
+func saveSlideScreenshots(runtime *common.RuntimeContext, data map[string]interface{}, outputDir string, presentationID string) ([]map[string]interface{}, error) {
+	items := common.GetSlice(data, "slide_images")
+	if len(items) == 0 {
+		return nil, slidesScreenshotAPIDataError(data, "slides screenshot returned no slide_images")
+	}
+	saved := make([]map[string]interface{}, 0, len(items))
+	for i, item := range items {
+		m, ok := item.(map[string]interface{})
+		if !ok {
+			return nil, slidesScreenshotAPIDataError(data, "slides screenshot returned invalid slide_images[%d]", i)
+		}
+		item, err := saveSlideScreenshotImage(runtime, m, outputDir, slideScreenshotListFileBase(presentationID, m, i), "")
+		if err != nil {
+			if isSlidesScreenshotPassthroughError(err) {
+				return nil, err
+			}
+			return nil, slidesScreenshotAPIDataError(data, "slides screenshot returned invalid slide_images[%d]: %v", i, err)
+		}
+		saved = append(saved, item)
+	}
+	return saved, nil
+}
+
+func saveRenderedSlideScreenshot(runtime *common.RuntimeContext, data map[string]interface{}, outputDir string, outputName string) ([]map[string]interface{}, error) {
+	item := common.GetMap(data, "slide_image")
+	if item == nil {
+		return nil, slidesScreenshotAPIDataError(data, "slides render screenshot returned no slide_image")
+	}
+	saved, err := saveSlideScreenshotImage(runtime, item, outputDir, outputName, "rendered-slide")
+	if err != nil {
+		if isSlidesScreenshotPassthroughError(err) {
+			return nil, err
+		}
+		return nil, slidesScreenshotAPIDataError(data, "slides render screenshot returned invalid slide_image: %v", err)
+	}
+	return []map[string]interface{}{saved}, nil
+}
+
+func saveSlideScreenshotImage(runtime *common.RuntimeContext, item map[string]interface{}, outputDir string, outputName string, fallbackName string) (map[string]interface{}, error) {
+	slideID := strings.TrimSpace(common.GetString(item, "slide_id"))
+	ext, label, err := slideScreenshotFormat(item)
+	if err != nil {
+		return nil, slidesScreenshotImageDataError(slideID, "%s", err)
+	}
+	encoded := strings.TrimSpace(common.GetString(item, "data"))
+	if encoded == "" {
+		return nil, slidesScreenshotImageDataError(slideID, "empty image data")
+	}
+	imageBytes, err := base64.StdEncoding.DecodeString(encoded)
+	if err != nil {
+		return nil, slidesScreenshotImageDataCauseError(slideID, err, "decode screenshot: %s", err)
+	}
+	fileBase := strings.TrimSpace(outputName)
+	if fileBase == "" {
+		fileBase = slideID
+	}
+	if fileBase == "" {
+		fileBase = fallbackName
+	}
+	path, err := writeUniqueScreenshotFile(runtime, outputDir, fileBase, ext, imageBytes)
+	if err != nil {
+		return nil, err
+	}
+	return map[string]interface{}{
+		"slide_id":     slideID,
+		"slide_number": slideScreenshotInt(item, "slide_number"),
+		"format":       label,
+		"path":         path,
+		"size":         len(imageBytes),
+	}, nil
+}
+
+func slideScreenshotListFileBase(presentationID string, item map[string]interface{}, index int) string {
+	presentationID = strings.TrimSpace(presentationID)
+	slideID := strings.TrimSpace(common.GetString(item, "slide_id"))
+	slideNumber := slideScreenshotInt(item, "slide_number")
+	if presentationID != "" {
+		switch {
+		case slideNumber > 0 && slideID != "":
+			return fmt.Sprintf("%s_p%03d_%s", presentationID, slideNumber, slideID)
+		case slideNumber > 0:
+			return fmt.Sprintf("%s_p%03d", presentationID, slideNumber)
+		case slideID != "":
+			return fmt.Sprintf("%s_%s", presentationID, slideID)
+		}
+	}
+	if slideID != "" {
+		return slideID
+	}
+	if slideNumber := slideScreenshotInt(item, "slide_number"); slideNumber > 0 {
+		return fmt.Sprintf("slide-%d", slideNumber)
+	}
+	return fmt.Sprintf("slide-%d", index+1)
+}
+
+func slideScreenshotFormat(item map[string]interface{}) (string, string, error) {
+	format := slideScreenshotInt(item, "format")
+	switch format {
+	case 1:
+		return "png", "png", nil
+	case 2:
+		return "jpg", "jpeg", nil
+	default:
+		return "", "", errs.NewAPIError(errs.SubtypeInvalidResponse, "unsupported screenshot format %d", format)
+	}
+}
+
+func slidesScreenshotImageDataError(slideID string, format string, args ...interface{}) error {
+	msg := fmt.Sprintf(format, args...)
+	if slideID != "" {
+		msg = fmt.Sprintf("%s for slide %s", msg, slideID)
+	}
+	return errs.NewAPIError(errs.SubtypeInvalidResponse, "%s", msg)
+}
+
+func slidesScreenshotImageDataCauseError(slideID string, cause error, format string, args ...interface{}) error {
+	msg := fmt.Sprintf(format, args...)
+	if slideID != "" {
+		msg = fmt.Sprintf("%s for slide %s", msg, slideID)
+	}
+	return errs.NewAPIError(errs.SubtypeInvalidResponse, "%s", msg).WithCause(cause)
+}
+
+func slideScreenshotInt(item map[string]interface{}, key string) int {
+	n, ok := util.ToFloat64(item[key])
+	if !ok {
+		return 0
+	}
+	return int(n)
+}
+
+func doSlidesScreenshotAPIJSONWithLogID(runtime *common.RuntimeContext, method string, apiPath string, query larkcore.QueryParams, body interface{}) (map[string]interface{}, error) {
+	req := &larkcore.ApiReq{
+		HttpMethod:  method,
+		ApiPath:     apiPath,
+		QueryParams: query,
+	}
+	if body != nil {
+		req.Body = body
+	}
+	resp, err := runtime.DoAPI(req)
+	if err != nil {
+		return nil, errs.WrapInternal(err)
+	}
+	data, err := runtime.ClassifyAPIResponse(resp)
+	if err != nil {
+		return data, err
+	}
+	if data == nil {
+		data = map[string]interface{}{}
+	}
+	if logID := strings.TrimSpace(resp.Header.Get("x-tt-logid")); logID != "" {
+		data["log_id"] = logID
+	}
+	return data, nil
+}
+
+func enrichSlidesScreenshotSelectorError(err error, slideNumbers []int) error {
+	if len(slideNumbers) == 0 {
+		return err
+	}
+	p, ok := errs.ProblemOf(err)
+	if !ok {
+		return err
+	}
+	if p.Hint == "" {
+		p.Hint = "slide_numbers was rejected by the server; verify the page number exists in this presentation, or retry with --slide-id."
+	}
+	return err
+}
+
+func slidesScreenshotAPIDataError(data map[string]interface{}, format string, args ...interface{}) error {
+	msg := fmt.Sprintf(format, args...)
+	err := errs.NewAPIError(errs.SubtypeInvalidResponse, "%s; raw_data=%v", msg, summarizeScreenshotAPIData(data))
+	if logID := strings.TrimSpace(common.GetString(data, "log_id")); logID != "" {
+		err = err.WithLogID(logID)
+	}
+	return err
+}
+
+func isSlidesScreenshotPassthroughError(err error) bool {
+	_, ok := errs.ProblemOf(err)
+	return ok
+}
+
+func summarizeScreenshotAPIData(v interface{}) interface{} {
+	switch x := v.(type) {
+	case map[string]interface{}:
+		out := make(map[string]interface{}, len(x))
+		for k, val := range x {
+			out[k] = summarizeScreenshotAPIData(val)
+		}
+		return out
+	case []interface{}:
+		out := make([]interface{}, 0, len(x))
+		for i, val := range x {
+			if i >= 20 {
+				out = append(out, fmt.Sprintf("<omitted %d more items>", len(x)-i))
+				break
+			}
+			out = append(out, summarizeScreenshotAPIData(val))
+		}
+		return out
+	case string:
+		if len(x) > 512 {
+			return fmt.Sprintf("<omitted string length=%d prefix=%q>", len(x), x[:64])
+		}
+		return x
+	default:
+		return x
+	}
+}
+
+func safeScreenshotFileBase(base string) string {
+	name := unsafeScreenshotFileCharRegex.ReplaceAllString(base, "_")
+	name = strings.Trim(name, "._-")
+	if name == "" {
+		name = "slide"
+	}
+	return name
+}
+
+func writeUniqueScreenshotFile(runtime *common.RuntimeContext, outputDir string, fileBase string, ext string, imageBytes []byte) (string, error) {
+	base := safeScreenshotFileBase(fileBase)
+	for i := 0; i < 1000; i++ {
+		candidateBase := base
+		if i > 0 {
+			candidateBase = fmt.Sprintf("%s_%d", base, i+1)
+		}
+		path := filepath.Join(outputDir, candidateBase+"."+ext)
+		if _, err := runtime.FileIO().Stat(path); err == nil {
+			continue
+		} else if !isScreenshotFileNotExist(err) {
+			return "", errs.NewInternalError(errs.SubtypeFileIO, "write screenshot %s: %v", path, err).WithCause(err)
+		}
+		if _, err := runtime.FileIO().Save(path, fileio.SaveOptions{}, bytes.NewReader(imageBytes)); err != nil {
+			return "", common.WrapSaveErrorTyped(err)
+		}
+		resolvedPath, err := runtime.ResolveSavePath(path)
+		if err != nil {
+			return "", errs.NewInternalError(errs.SubtypeFileIO, "resolve saved screenshot path %s: %v", path, err).WithCause(err)
+		}
+		return resolvedPath, nil
+	}
+	path := filepath.Join(outputDir, base+"."+ext)
+	return "", errs.NewInternalError(errs.SubtypeFileIO, "write screenshot %s: too many duplicate file names", path)
+}
+
+func isScreenshotFileNotExist(err error) bool {
+	return os.IsNotExist(err)
+}

shortcuts/slides/slides_screenshot_test.go

@@ -0,0 +1,506 @@
+// Copyright (c) 2026 Lark Technologies Pte. Ltd.
+// SPDX-License-Identifier: MIT
+
+package slides
+
+import (
+	"encoding/base64"
+	"encoding/json"
+	"os"
+	"path/filepath"
+	"strings"
+	"testing"
+
+	"github.com/larksuite/cli/errs"
+	"github.com/larksuite/cli/internal/cmdutil"
+	"github.com/larksuite/cli/internal/httpmock"
+)
+
+func TestSlidesScreenshotDeclaredScopes(t *testing.T) {
+	got := SlidesScreenshot.DeclaredScopesForIdentity("user")
+	want := []string{"slides:presentation:screenshot", "wiki:node:read"}
+	if len(got) != len(want) || got[0] != want[0] || got[1] != want[1] {
+		t.Fatalf("declared scopes = %#v, want %#v", got, want)
+	}
+}
+
+func TestSlidesScreenshotWritesFilesAndSuppressesBase64(t *testing.T) {
+	dir := t.TempDir()
+	withSlidesTestWorkingDir(t, dir)
+
+	imageBytes := []byte("png-bytes")
+	jpegBytes := []byte("jpeg-bytes")
+	f, stdout, _, reg := cmdutil.TestFactory(t, slidesTestConfig(t, ""))
+	stub := &httpmock.Stub{
+		Method: "POST",
+		URL:    "/open-apis/slides_ai/v1/xml_presentations/pres_abc/slide_images",
+		Body: map[string]interface{}{
+			"code": 0,
+			"data": map[string]interface{}{
+				"slide_images": []map[string]interface{}{
+					{
+						"slide_id": "slide_1",
+						"format":   1,
+						"data":     base64.StdEncoding.EncodeToString(imageBytes),
+					},
+					{
+						"slide_id":     "slide_2",
+						"slide_number": 2,
+						"format":       2,
+						"data":         base64.StdEncoding.EncodeToString(jpegBytes),
+					},
+				},
+			},
+		},
+	}
+	reg.Register(stub)
+
+	err := runSlidesShortcut(t, f, stdout, SlidesScreenshot, []string{
+		"+screenshot",
+		"--presentation", "pres_abc",
+		"--slide-id", "slide_1",
+		"--output-dir", "shots",
+		"--as", "user",
+	})
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+
+	path := filepath.Join(dir, "shots", "pres_abc_slide_1.png")
+	gotBytes, err := os.ReadFile(path)
+	if err != nil {
+		t.Fatalf("read screenshot: %v", err)
+	}
+	if string(gotBytes) != string(imageBytes) {
+		t.Fatalf("written bytes = %q, want %q", gotBytes, imageBytes)
+	}
+	jpegPath := filepath.Join(dir, "shots", "pres_abc_p002_slide_2.jpg")
+	gotJPEGBytes, err := os.ReadFile(jpegPath)
+	if err != nil {
+		t.Fatalf("read jpeg screenshot: %v", err)
+	}
+	if string(gotJPEGBytes) != string(jpegBytes) {
+		t.Fatalf("written jpeg bytes = %q, want %q", gotJPEGBytes, jpegBytes)
+	}
+	if strings.Contains(stdout.String(), base64.StdEncoding.EncodeToString(imageBytes)) {
+		t.Fatalf("stdout leaked base64 image data: %s", stdout.String())
+	}
+
+	data := decodeShortcutData(t, stdout)
+	if data["xml_presentation_id"] != "pres_abc" {
+		t.Fatalf("xml_presentation_id = %v", data["xml_presentation_id"])
+	}
+	items, ok := data["screenshots"].([]interface{})
+	if !ok || len(items) != 2 {
+		t.Fatalf("screenshots = %#v, want two items", data["screenshots"])
+	}
+	item, _ := items[0].(map[string]interface{})
+	if item["slide_id"] != "slide_1" {
+		t.Fatalf("slide_id = %v, want slide_1", item["slide_id"])
+	}
+	gotPath := item["path"].(string)
+	if !filepath.IsAbs(gotPath) {
+		t.Fatalf("path = %v, want absolute path", gotPath)
+	}
+	if !strings.HasSuffix(gotPath, filepath.Join("shots", "pres_abc_slide_1.png")) {
+		t.Fatalf("path = %v, want shots/pres_abc_slide_1.png suffix", item["path"])
+	}
+	item2, _ := items[1].(map[string]interface{})
+	if item2["format"] != "jpeg" {
+		t.Fatalf("format = %v, want jpeg", item2["format"])
+	}
+	gotPath2 := item2["path"].(string)
+	if !filepath.IsAbs(gotPath2) {
+		t.Fatalf("path = %v, want absolute path", gotPath2)
+	}
+	if !strings.HasSuffix(gotPath2, filepath.Join("shots", "pres_abc_p002_slide_2.jpg")) {
+		t.Fatalf("path = %v, want shots/pres_abc_p002_slide_2.jpg suffix", item2["path"])
+	}
+
+	var body struct {
+		SlideIDs []string `json:"slide_ids"`
+	}
+	if err := json.Unmarshal(stub.CapturedBody, &body); err != nil {
+		t.Fatalf("decode request body: %v", err)
+	}
+	if len(body.SlideIDs) != 1 || body.SlideIDs[0] != "slide_1" {
+		t.Fatalf("slide_ids = %#v, want [slide_1]", body.SlideIDs)
+	}
+}
+
+func TestSlidesScreenshotListBySlideNumber(t *testing.T) {
+	dir := t.TempDir()
+	withSlidesTestWorkingDir(t, dir)
+
+	f, stdout, _, reg := cmdutil.TestFactory(t, slidesTestConfig(t, ""))
+	stub := &httpmock.Stub{
+		Method: "POST",
+		URL:    "/open-apis/slides_ai/v1/xml_presentations/pres_abc/slide_images",
+		Body: map[string]interface{}{
+			"code": 0,
+			"data": map[string]interface{}{
+				"slide_images": []map[string]interface{}{
+					{
+						"slide_number": 2,
+						"format":       1,
+						"data":         base64.StdEncoding.EncodeToString([]byte("png-bytes")),
+					},
+				},
+			},
+		},
+	}
+	reg.Register(stub)
+
+	err := runSlidesShortcut(t, f, stdout, SlidesScreenshot, []string{
+		"+screenshot",
+		"--presentation", "pres_abc",
+		"--slide-number", "2",
+		"--as", "user",
+	})
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+
+	var body struct {
+		SlideNumbers []int `json:"slide_numbers"`
+	}
+	if err := json.Unmarshal(stub.CapturedBody, &body); err != nil {
+		t.Fatalf("decode request body: %v", err)
+	}
+	if len(body.SlideNumbers) != 1 || body.SlideNumbers[0] != 2 {
+		t.Fatalf("slide_numbers = %#v, want [2]", body.SlideNumbers)
+	}
+	path := filepath.Join(dir, defaultSlidesScreenshotDir, "pres_abc_p002.png")
+	if _, err := os.ReadFile(path); err != nil {
+		t.Fatalf("read screenshot without slide_id: %v", err)
+	}
+}
+
+func TestSlidesScreenshotAvoidsOverwritingExistingFile(t *testing.T) {
+	dir := t.TempDir()
+	withSlidesTestWorkingDir(t, dir)
+	outputDir := filepath.Join(dir, "shots")
+	if err := os.MkdirAll(outputDir, 0o755); err != nil {
+		t.Fatalf("create output dir: %v", err)
+	}
+	existingPath := filepath.Join(outputDir, "pres_abc_p002.png")
+	if err := os.WriteFile(existingPath, []byte("existing"), 0o644); err != nil {
+		t.Fatalf("write existing screenshot: %v", err)
+	}
+
+	imageBytes := []byte("new-png")
+	f, stdout, _, reg := cmdutil.TestFactory(t, slidesTestConfig(t, ""))
+	reg.Register(&httpmock.Stub{
+		Method: "POST",
+		URL:    "/open-apis/slides_ai/v1/xml_presentations/pres_abc/slide_images",
+		Body: map[string]interface{}{
+			"code": 0,
+			"data": map[string]interface{}{
+				"slide_images": []map[string]interface{}{
+					{
+						"slide_number": 2,
+						"format":       1,
+						"data":         base64.StdEncoding.EncodeToString(imageBytes),
+					},
+				},
+			},
+		},
+	})
+
+	err := runSlidesShortcut(t, f, stdout, SlidesScreenshot, []string{
+		"+screenshot",
+		"--presentation", "pres_abc",
+		"--slide-number", "2",
+		"--output-dir", "shots",
+		"--as", "user",
+	})
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	gotExisting, err := os.ReadFile(existingPath)
+	if err != nil {
+		t.Fatalf("read existing screenshot: %v", err)
+	}
+	if string(gotExisting) != "existing" {
+		t.Fatalf("existing screenshot = %q, want unchanged", gotExisting)
+	}
+	newPath := filepath.Join(outputDir, "pres_abc_p002_2.png")
+	gotNew, err := os.ReadFile(newPath)
+	if err != nil {
+		t.Fatalf("read deduplicated screenshot: %v", err)
+	}
+	if string(gotNew) != string(imageBytes) {
+		t.Fatalf("deduplicated screenshot = %q, want %q", gotNew, imageBytes)
+	}
+	data := decodeShortcutData(t, stdout)
+	items, ok := data["screenshots"].([]interface{})
+	if !ok || len(items) != 1 {
+		t.Fatalf("screenshots = %#v, want one item", data["screenshots"])
+	}
+	item, _ := items[0].(map[string]interface{})
+	if !strings.HasSuffix(item["path"].(string), filepath.Join("shots", "pres_abc_p002_2.png")) {
+		t.Fatalf("path = %v, want shots/pres_abc_p002_2.png suffix", item["path"])
+	}
+}
+
+func TestSlidesScreenshotListRequiresSelector(t *testing.T) {
+	f, stdout, _, _ := cmdutil.TestFactory(t, slidesTestConfig(t, ""))
+
+	err := runSlidesShortcut(t, f, stdout, SlidesScreenshot, []string{
+		"+screenshot",
+		"--presentation", "pres_abc",
+		"--as", "user",
+	})
+	if err == nil {
+		t.Fatal("expected error")
+	}
+	if !strings.Contains(err.Error(), "--slide-id or --slide-number is required") {
+		t.Fatalf("error = %v, want missing selector error", err)
+	}
+}
+
+func TestSlidesScreenshotRenderContentWritesFile(t *testing.T) {
+	dir := t.TempDir()
+	withSlidesTestWorkingDir(t, dir)
+
+	content := `<slide xmlns="http://www.larkoffice.com/sml/2.0"><data></data></slide>`
+	if err := os.WriteFile(filepath.Join(dir, "slide.xml"), []byte(content), 0o644); err != nil {
+		t.Fatalf("write input xml: %v", err)
+	}
+	imageBytes := []byte("rendered-png")
+	f, stdout, _, reg := cmdutil.TestFactory(t, slidesTestConfig(t, ""))
+	stub := &httpmock.Stub{
+		Method: "POST",
+		URL:    "/open-apis/slides_ai/v1/slide_image/render",
+		Body: map[string]interface{}{
+			"code": 0,
+			"data": map[string]interface{}{
+				"slide_image": map[string]interface{}{
+					"slide_id":     "render_slide",
+					"slide_number": 1,
+					"format":       1,
+					"data":         base64.StdEncoding.EncodeToString(imageBytes),
+				},
+			},
+		},
+	}
+	reg.Register(stub)
+
+	err := runSlidesShortcut(t, f, stdout, SlidesScreenshot, []string{
+		"+screenshot",
+		"--content", "@slide.xml",
+		"--output-dir", "shots",
+		"--output-name", "preview",
+		"--as", "user",
+	})
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+
+	path := filepath.Join(dir, "shots", "preview.png")
+	gotBytes, err := os.ReadFile(path)
+	if err != nil {
+		t.Fatalf("read rendered screenshot: %v", err)
+	}
+	if string(gotBytes) != string(imageBytes) {
+		t.Fatalf("written bytes = %q, want %q", gotBytes, imageBytes)
+	}
+	if strings.Contains(stdout.String(), base64.StdEncoding.EncodeToString(imageBytes)) {
+		t.Fatalf("stdout leaked base64 image data: %s", stdout.String())
+	}
+
+	var body struct {
+		Content string `json:"content"`
+	}
+	if err := json.Unmarshal(stub.CapturedBody, &body); err != nil {
+		t.Fatalf("decode request body: %v", err)
+	}
+	if body.Content != content {
+		t.Fatalf("content = %q, want input XML", body.Content)
+	}
+
+	data := decodeShortcutData(t, stdout)
+	items, ok := data["screenshots"].([]interface{})
+	if !ok || len(items) != 1 {
+		t.Fatalf("screenshots = %#v, want one item", data["screenshots"])
+	}
+	item, _ := items[0].(map[string]interface{})
+	if !strings.HasSuffix(item["path"].(string), filepath.Join("shots", "preview.png")) {
+		t.Fatalf("path = %v, want shots/preview.png suffix", item["path"])
+	}
+}
+
+func TestSlidesScreenshotRenderRejectsSlideSelectors(t *testing.T) {
+	f, stdout, _, _ := cmdutil.TestFactory(t, slidesTestConfig(t, ""))
+
+	err := runSlidesShortcut(t, f, stdout, SlidesScreenshot, []string{
+		"+screenshot",
+		"--content", `<slide xmlns="http://www.larkoffice.com/sml/2.0"><data></data></slide>`,
+		"--slide-id", "slide_1",
+		"--as", "user",
+	})
+	if err == nil {
+		t.Fatal("expected error")
+	}
+	if !strings.Contains(err.Error(), "--content cannot be used with --slide-id or --slide-number") {
+		t.Fatalf("error = %v, want content/slide selector conflict", err)
+	}
+}
+
+func TestSlidesScreenshotRenderRejectsListOnlyFlags(t *testing.T) {
+	f, stdout, _, _ := cmdutil.TestFactory(t, slidesTestConfig(t, ""))
+
+	err := runSlidesShortcut(t, f, stdout, SlidesScreenshot, []string{
+		"+screenshot",
+		"--content", `<slide xmlns="http://www.larkoffice.com/sml/2.0"><data></data></slide>`,
+		"--presentation", "pres_abc",
+		"--as", "user",
+	})
+	if err == nil {
+		t.Fatal("expected error")
+	}
+	if !strings.Contains(err.Error(), "--presentation cannot be used with --content") {
+		t.Fatalf("error = %v, want presentation/content conflict", err)
+	}
+}
+
+func TestSlidesScreenshotDryRunSelectsListOrRenderAPI(t *testing.T) {
+	t.Run("list", func(t *testing.T) {
+		f, stdout, _, _ := cmdutil.TestFactory(t, slidesTestConfig(t, ""))
+		err := runSlidesShortcut(t, f, stdout, SlidesScreenshot, []string{
+			"+screenshot",
+			"--presentation", "pres_abc",
+			"--slide-number", "2",
+			"--dry-run",
+			"--as", "user",
+		})
+		if err != nil {
+			t.Fatalf("unexpected error: %v", err)
+		}
+		out := stdout.String()
+		if !strings.Contains(out, "/xml_presentations/pres_abc/slide_images") {
+			t.Fatalf("dry-run missing list endpoint: %s", out)
+		}
+		if !strings.Contains(out, "slide_numbers") {
+			t.Fatalf("dry-run missing slide_numbers body: %s", out)
+		}
+	})
+
+	t.Run("render", func(t *testing.T) {
+		f, stdout, _, _ := cmdutil.TestFactory(t, slidesTestConfig(t, ""))
+		err := runSlidesShortcut(t, f, stdout, SlidesScreenshot, []string{
+			"+screenshot",
+			"--content", `<slide xmlns="http://www.larkoffice.com/sml/2.0"><data></data></slide>`,
+			"--dry-run",
+			"--as", "user",
+		})
+		if err != nil {
+			t.Fatalf("unexpected error: %v", err)
+		}
+		out := stdout.String()
+		if !strings.Contains(out, "/slide_image/render") {
+			t.Fatalf("dry-run missing render endpoint: %s", out)
+		}
+		if !strings.Contains(out, "base64_output") {
+			t.Fatalf("dry-run missing base64 suppression note: %s", out)
+		}
+	})
+}
+
+func TestSlidesScreenshotRejectsBadOutputDir(t *testing.T) {
+	f, stdout, _, _ := cmdutil.TestFactory(t, slidesTestConfig(t, ""))
+
+	err := runSlidesShortcut(t, f, stdout, SlidesScreenshot, []string{
+		"+screenshot",
+		"--presentation", "pres_abc",
+		"--slide-id", "slide_1",
+		"--output-dir", "../outside",
+		"--as", "user",
+	})
+	if err == nil {
+		t.Fatal("expected error for unsafe output dir")
+	}
+	if !strings.Contains(err.Error(), "--output-dir invalid") {
+		t.Fatalf("error = %v, want output-dir validation", err)
+	}
+}
+
+func TestSlidesScreenshotNoImagesErrorIncludesRawDataAndLogID(t *testing.T) {
+	dir := t.TempDir()
+	withSlidesTestWorkingDir(t, dir)
+
+	f, stdout, _, reg := cmdutil.TestFactory(t, slidesTestConfig(t, ""))
+	reg.Register(&httpmock.Stub{
+		Method: "POST",
+		URL:    "/open-apis/slides_ai/v1/xml_presentations/pres_abc/slide_images",
+		Headers: map[string][]string{
+			"Content-Type": {"application/json"},
+			"X-Tt-Logid":   {"log-123"},
+		},
+		Body: map[string]interface{}{
+			"code": 0,
+			"data": map[string]interface{}{
+				"unexpected": "shape",
+			},
+		},
+	})
+
+	err := runSlidesShortcut(t, f, stdout, SlidesScreenshot, []string{
+		"+screenshot",
+		"--presentation", "pres_abc",
+		"--slide-id", "pJJ",
+		"--as", "user",
+	})
+	if err == nil {
+		t.Fatal("expected error")
+	}
+	p, ok := errs.ProblemOf(err)
+	if !ok {
+		t.Fatalf("error type = %T, want typed problem", err)
+	}
+	if p.LogID != "log-123" {
+		t.Fatalf("log_id = %v, want log-123", p.LogID)
+	}
+	if !strings.Contains(p.Message, "unexpected:shape") {
+		t.Fatalf("message = %q, want raw_data summary", p.Message)
+	}
+}
+
+func TestSlidesScreenshotSlideNumberAPIErrorAddsHint(t *testing.T) {
+	dir := t.TempDir()
+	withSlidesTestWorkingDir(t, dir)
+
+	f, stdout, _, reg := cmdutil.TestFactory(t, slidesTestConfig(t, ""))
+	reg.Register(&httpmock.Stub{
+		Method: "POST",
+		URL:    "/open-apis/slides_ai/v1/xml_presentations/pres_abc/slide_images",
+		Headers: map[string][]string{
+			"Content-Type": {"application/json"},
+			"X-Tt-Logid":   {"log-slide-number"},
+		},
+		Body: map[string]interface{}{
+			"code": 99992402,
+			"msg":  "field validation failed",
+		},
+	})
+
+	err := runSlidesShortcut(t, f, stdout, SlidesScreenshot, []string{
+		"+screenshot",
+		"--presentation", "pres_abc",
+		"--slide-number", "25",
+		"--as", "user",
+	})
+	if err == nil {
+		t.Fatal("expected error")
+	}
+	p, ok := errs.ProblemOf(err)
+	if !ok {
+		t.Fatalf("error type = %T, want typed problem", err)
+	}
+	if p.LogID != "log-slide-number" {
+		t.Fatalf("log_id = %v, want log-slide-number", p.LogID)
+	}
+	if !strings.Contains(p.Hint, "--slide-id") {
+		t.Fatalf("hint = %q, want --slide-id guidance", p.Hint)
+	}
+}