自动同步: 2026-03-15 02:10

2026-03-15 02:10:07 +08:00
parent c8c4b780c8
commit b1476136aa
31 changed files with 1119 additions and 2646 deletions
@@ -347,8 +347,8 @@
    "memoryFlushCompactionCount": 0
  },
  "agent:dev-assist:main": {
-    "sessionId": "b5c67eb9-7916-47dd-9491-4832eab53d10",
+    "sessionId": "60601381-862d-454d-9c27-2867f00a71d1",
-    "updatedAt": 1773132090440,
+    "updatedAt": 1773508835390,
    "systemSent": true,
    "abortedLastRun": false,
    "chatType": "direct",
@@ -369,7 +369,7 @@
      "to": "user:ou_302b162eb824d405718845441a32d9cf",
      "accountId": "feishu-dev"
    },
-    "sessionFile": "/home/starmen1265/.openclaw/agents/dev-assist/sessions/b5c67eb9-7916-47dd-9491-4832eab53d10.jsonl",
+    "sessionFile": "/home/starmen1265/.openclaw/agents/dev-assist/sessions/60601381-862d-454d-9c27-2867f00a71d1.jsonl",
    "compactionCount": 0,
    "skillsSnapshot": {
      "prompt": "\n\nThe following skills provide specialized instructions for specific tasks.\nUse the read tool to load a skill's file when the task matches its description.\nWhen a skill file references a relative path, resolve it against the skill directory (parent of SKILL.md / dirname of the path) and use that absolute path in tool commands.\n\n<available_skills>\n  <skill>\n    <name>capability-evolver</name>\n    <description>A self-evolution engine for AI agents. Analyzes runtime history to identify improvements and applies protocol-constrained evolution.</description>\n    <location>~/.openclaw/skills/capability-evolver/SKILL.md</location>\n  </skill>\n  <skill>\n    <name>clawlist</name>\n    <description>MUST use for any multi-step project, long-running task, or infinite monitoring workflow. Plan, execute, track, and verify tasks with checkpoint validation. For projects, automation, and ongoing operations.</description>\n    <location>~/.openclaw/skills/clawlist/SKILL.md</location>\n  </skill>\n</available_skills>",
@@ -410,8 +410,8 @@
    "model": "glm-5",
    "systemPromptReport": {
      "source": "run",
-      "generatedAt": 1773132050946,
+      "generatedAt": 1773508824996,
-      "sessionId": "b5c67eb9-7916-47dd-9491-4832eab53d10",
+      "sessionId": "60601381-862d-454d-9c27-2867f00a71d1",
      "sessionKey": "agent:dev-assist:main",
      "provider": "zhipu",
      "model": "glm-5",
@@ -423,8 +423,8 @@
        "sandboxed": false
      },
      "systemPrompt": {
-        "chars": 32692,
+        "chars": 26851,
-        "projectContextChars": 19802,
+        "projectContextChars": 13961,
        "nonProjectContextChars": 12890
      },
      "injectedWorkspaceFiles": [
@@ -492,60 +492,6 @@
          "injectedChars": 18,
          "truncated": false
        },
        {
          "path": "SELF_IMPROVEMENT_REMINDER.md",
          "rawChars": 616,
          "injectedChars": 616,
          "truncated": false
        },
        {
          "path": "SELF_IMPROVEMENT_REMINDER.md",
          "rawChars": 616,
          "injectedChars": 616,
          "truncated": false
        },
        {
          "path": "SELF_IMPROVEMENT_REMINDER.md",
          "rawChars": 616,
          "injectedChars": 616,
          "truncated": false
        },
        {
          "path": "SELF_IMPROVEMENT_REMINDER.md",
          "rawChars": 616,
          "injectedChars": 616,
          "truncated": false
        },
        {
          "path": "SELF_IMPROVEMENT_REMINDER.md",
          "rawChars": 616,
          "injectedChars": 616,
          "truncated": false
        },
        {
          "path": "SELF_IMPROVEMENT_REMINDER.md",
          "rawChars": 616,
          "injectedChars": 616,
          "truncated": false
        },
        {
          "path": "SELF_IMPROVEMENT_REMINDER.md",
          "rawChars": 616,
          "injectedChars": 616,
          "truncated": false
        },
        {
          "path": "SELF_IMPROVEMENT_REMINDER.md",
          "rawChars": 616,
          "injectedChars": 616,
          "truncated": false
        },
        {
          "path": "SELF_IMPROVEMENT_REMINDER.md",
          "rawChars": 616,
          "injectedChars": 616,
          "truncated": false
        },
        {
          "path": "SELF_IMPROVEMENT_REMINDER.md",
          "rawChars": 616,
@@ -777,12 +723,12 @@
        ]
      }
    },
-    "cacheRead": 17472,
+    "cacheRead": 11392,
    "cacheWrite": 0,
    "totalTokensFresh": true,
    "contextTokens": 200000,
-    "inputTokens": 4722,
+    "inputTokens": 5334,
-    "outputTokens": 1147,
+    "outputTokens": 344,
-    "totalTokens": 22194
+    "totalTokens": 16726
  }
 }
@@ -1,7 +1,7 @@
 {
  "agent:haonan:main": {
-    "sessionId": "624cbf67-7ab7-4df9-968d-4a63529b9111",
+    "sessionId": "40e80fbb-07c8-440d-a1d4-1e876b86b82d",
-    "updatedAt": 1773377664001,
+    "updatedAt": 1773508722585,
    "systemSent": true,
    "skillsSnapshot": {
      "prompt": "\n\nThe following skills provide specialized instructions for specific tasks.\nUse the read tool to load a skill's file when the task matches its description.\nWhen a skill file references a relative path, resolve it against the skill directory (parent of SKILL.md / dirname of the path) and use that absolute path in tool commands.\n\n<available_skills>\n  <skill>\n    <name>capability-evolver</name>\n    <description>A self-evolution engine for AI agents. Analyzes runtime history to identify improvements and applies protocol-constrained evolution.</description>\n    <location>~/.openclaw/skills/capability-evolver/SKILL.md</location>\n  </skill>\n  <skill>\n    <name>clawlist</name>\n    <description>MUST use for any multi-step project, long-running task, or infinite monitoring workflow. Plan, execute, track, and verify tasks with checkpoint validation. For projects, automation, and ongoing operations.</description>\n    <location>~/.openclaw/skills/clawlist/SKILL.md</location>\n  </skill>\n</available_skills>",
@@ -59,11 +59,11 @@
      "to": "user:ou_be83febac9ee2afaa343c642bc697c48",
      "accountId": "feishu-pm"
    },
-    "sessionFile": "/home/starmen1265/.openclaw/agents/haonan/sessions/624cbf67-7ab7-4df9-968d-4a63529b9111.jsonl",
+    "sessionFile": "/home/starmen1265/.openclaw/agents/haonan/sessions/40e80fbb-07c8-440d-a1d4-1e876b86b82d.jsonl",
    "systemPromptReport": {
      "source": "run",
-      "generatedAt": 1773377656536,
+      "generatedAt": 1773508545805,
-      "sessionId": "624cbf67-7ab7-4df9-968d-4a63529b9111",
+      "sessionId": "40e80fbb-07c8-440d-a1d4-1e876b86b82d",
      "sessionKey": "agent:haonan:main",
      "provider": "zhipu",
      "model": "glm-5",
@@ -75,8 +75,8 @@
        "sandboxed": false
      },
      "systemPrompt": {
-        "chars": 28195,
+        "chars": 29493,
-        "projectContextChars": 15193,
+        "projectContextChars": 16491,
        "nonProjectContextChars": 13002
      },
      "injectedWorkspaceFiles": [
@@ -144,6 +144,18 @@
          "injectedChars": 1168,
          "truncated": false
        },
        {
          "path": "SELF_IMPROVEMENT_REMINDER.md",
          "rawChars": 616,
          "injectedChars": 616,
          "truncated": false
        },
        {
          "path": "SELF_IMPROVEMENT_REMINDER.md",
          "rawChars": 616,
          "injectedChars": 616,
          "truncated": false
        },
        {
          "path": "SELF_IMPROVEMENT_REMINDER.md",
          "rawChars": 616,
@@ -379,12 +391,12 @@
    "totalTokensFresh": true,
    "lastHeartbeatText": "已在飞书向蒋先生发送日历同步提醒！✅",
    "lastHeartbeatSentAt": 1772924419184,
-    "cacheRead": 25472,
+    "cacheRead": 29760,
    "cacheWrite": 0,
    "contextTokens": 200000,
-    "inputTokens": 12782,
+    "inputTokens": 13411,
-    "outputTokens": 123,
+    "outputTokens": 1288,
-    "totalTokens": 25768
+    "totalTokens": 30249
  },
  "agent:haonan:wecom:dm:zhangqing": {
    "origin": {
@@ -715,8 +727,8 @@
    "memoryFlushCompactionCount": 0
  },
  "agent:haonan:cron:41396aeb-6e3a-47b2-9bd3-9156a348c1dc": {
-    "sessionId": "10740f43-38c1-4a7c-acf0-7da04f20d1e9",
+    "sessionId": "7c1a6bfd-b5b6-4ac5-98f3-6a73aec18a8f",
-    "updatedAt": 1773424800009,
+    "updatedAt": 1773511200009,
    "systemSent": true,
    "model": "glm-5",
    "contextTokens": 200000,
@@ -757,11 +769,11 @@
      "version": 0
    },
    "modelProvider": "zhipu",
-    "inputTokens": 11105,
+    "inputTokens": 4500,
-    "outputTokens": 336,
+    "outputTokens": 194,
-    "totalTokens": 17181,
+    "totalTokens": 16577,
    "totalTokensFresh": true,
-    "cacheRead": 56128,
+    "cacheRead": 28544,
    "cacheWrite": 0
  },
  "agent:haonan:cron:f03b184f-5f1d-4093-ae1b-fc9fe0b59eef": {
@@ -861,8 +873,8 @@
    "cacheWrite": 0
  },
  "agent:haonan:cron:aa4d2de9-79be-4262-83ee-af8824793bfc": {
-    "sessionId": "8b634542-6d0e-4fa3-b794-d784174d530b",
+    "sessionId": "b5ab88d0-f6d3-4b73-8d96-e6a58f5e85e4",
-    "updatedAt": 1773425400005,
+    "updatedAt": 1773511800007,
    "systemSent": true,
    "label": "Cron: openclaw-git-sync",
    "skillsSnapshot": {
@@ -910,113 +922,6 @@
    "cacheRead": 32768,
    "cacheWrite": 0
  },
  "agent:haonan:cron:f03b184f-5f1d-4093-ae1b-fc9fe0b59eef:run:11e71798-cda2-4029-afdc-b14043c48f5e": {
    "sessionId": "11e71798-cda2-4029-afdc-b14043c48f5e",
    "updatedAt": 1773339600012,
    "systemSent": true,
    "model": "glm-4.7-FlashX",
    "contextTokens": 200000,
    "label": "Cron: capability-evolver-daily",
    "skillsSnapshot": {
      "prompt": "\n\nThe following skills provide specialized instructions for specific tasks.\nUse the read tool to load a skill's file when the task matches its description.\nWhen a skill file references a relative path, resolve it against the skill directory (parent of SKILL.md / dirname of the path) and use that absolute path in tool commands.\n\n<available_skills>\n  <skill>\n    <name>capability-evolver</name>\n    <description>A self-evolution engine for AI agents. Analyzes runtime history to identify improvements and applies protocol-constrained evolution.</description>\n    <location>~/.openclaw/skills/capability-evolver/SKILL.md</location>\n  </skill>\n  <skill>\n    <name>clawlist</name>\n    <description>MUST use for any multi-step project, long-running task, or infinite monitoring workflow. Plan, execute, track, and verify tasks with checkpoint validation. For projects, automation, and ongoing operations.</description>\n    <location>~/.openclaw/skills/clawlist/SKILL.md</location>\n  </skill>\n</available_skills>",
      "skills": [
        {
          "name": "capability-evolver"
        },
        {
          "name": "clawlist"
        }
      ],
      "skillFilter": [
        "clawlist",
        "capability-evolver",
        "self-improving-agent"
      ],
      "resolvedSkills": [
        {
          "name": "capability-evolver",
          "description": "A self-evolution engine for AI agents. Analyzes runtime history to identify improvements and applies protocol-constrained evolution.",
          "filePath": "/home/starmen1265/.openclaw/skills/capability-evolver/SKILL.md",
          "baseDir": "/home/starmen1265/.openclaw/skills/capability-evolver",
          "source": "openclaw-managed",
          "disableModelInvocation": false
        },
        {
          "name": "clawlist",
          "description": "MUST use for any multi-step project, long-running task, or infinite monitoring workflow. Plan, execute, track, and verify tasks with checkpoint validation. For projects, automation, and ongoing operations.",
          "filePath": "/home/starmen1265/.openclaw/skills/clawlist/SKILL.md",
          "baseDir": "/home/starmen1265/.openclaw/skills/clawlist",
          "source": "openclaw-managed",
          "disableModelInvocation": false
        }
      ],
      "version": 0
    },
    "modelProvider": "zhipu",
    "inputTokens": 153698,
    "outputTokens": 6902,
    "totalTokens": 33385,
    "deliveryContext": {
      "channel": "webchat"
    },
    "lastChannel": "webchat",
    "channel": "webchat",
    "sessionFile": "/home/starmen1265/.openclaw/agents/haonan/sessions/ccc40ca7-7876-4d82-ab84-7b790feddc19.jsonl",
    "abortedLastRun": false,
    "totalTokensFresh": true,
    "cacheRead": 384894,
    "cacheWrite": 0
  },
  "agent:haonan:cron:41396aeb-6e3a-47b2-9bd3-9156a348c1dc:run:10740f43-38c1-4a7c-acf0-7da04f20d1e9": {
    "sessionId": "10740f43-38c1-4a7c-acf0-7da04f20d1e9",
    "updatedAt": 1773424800009,
    "systemSent": true,
    "model": "glm-5",
    "contextTokens": 200000,
    "label": "Cron: backup-openclaw-json",
    "skillsSnapshot": {
      "prompt": "\n\nThe following skills provide specialized instructions for specific tasks.\nUse the read tool to load a skill's file when the task matches its description.\nWhen a skill file references a relative path, resolve it against the skill directory (parent of SKILL.md / dirname of the path) and use that absolute path in tool commands.\n\n<available_skills>\n  <skill>\n    <name>capability-evolver</name>\n    <description>A self-evolution engine for AI agents. Analyzes runtime history to identify improvements and applies protocol-constrained evolution.</description>\n    <location>~/.openclaw/skills/capability-evolver/SKILL.md</location>\n  </skill>\n  <skill>\n    <name>clawlist</name>\n    <description>MUST use for any multi-step project, long-running task, or infinite monitoring workflow. Plan, execute, track, and verify tasks with checkpoint validation. For projects, automation, and ongoing operations.</description>\n    <location>~/.openclaw/skills/clawlist/SKILL.md</location>\n  </skill>\n</available_skills>",
      "skills": [
        {
          "name": "capability-evolver"
        },
        {
          "name": "clawlist"
        }
      ],
      "skillFilter": [
        "clawlist",
        "capability-evolver",
        "self-improving-agent"
      ],
      "resolvedSkills": [
        {
          "name": "capability-evolver",
          "description": "A self-evolution engine for AI agents. Analyzes runtime history to identify improvements and applies protocol-constrained evolution.",
          "filePath": "/home/starmen1265/.openclaw/skills/capability-evolver/SKILL.md",
          "baseDir": "/home/starmen1265/.openclaw/skills/capability-evolver",
          "source": "openclaw-managed",
          "disableModelInvocation": false
        },
        {
          "name": "clawlist",
          "description": "MUST use for any multi-step project, long-running task, or infinite monitoring workflow. Plan, execute, track, and verify tasks with checkpoint validation. For projects, automation, and ongoing operations.",
          "filePath": "/home/starmen1265/.openclaw/skills/clawlist/SKILL.md",
          "baseDir": "/home/starmen1265/.openclaw/skills/clawlist",
          "source": "openclaw-managed",
          "disableModelInvocation": false
        }
      ],
      "version": 0
    },
    "modelProvider": "zhipu",
    "inputTokens": 11105,
    "outputTokens": 336,
    "totalTokens": 17181,
    "totalTokensFresh": true,
    "cacheRead": 56128,
    "cacheWrite": 0
  },
  "agent:haonan:cron:aa4d2de9-79be-4262-83ee-af8824793bfc:run:8b634542-6d0e-4fa3-b794-d784174d530b": {
    "sessionId": "8b634542-6d0e-4fa3-b794-d784174d530b",
    "updatedAt": 1773425400005,
@@ -1123,5 +1028,105 @@
    "totalTokensFresh": true,
    "cacheRead": 384894,
    "cacheWrite": 0
  },
  "agent:haonan:cron:41396aeb-6e3a-47b2-9bd3-9156a348c1dc:run:7c1a6bfd-b5b6-4ac5-98f3-6a73aec18a8f": {
    "sessionId": "7c1a6bfd-b5b6-4ac5-98f3-6a73aec18a8f",
    "updatedAt": 1773511200009,
    "systemSent": true,
    "model": "glm-5",
    "contextTokens": 200000,
    "label": "Cron: backup-openclaw-json",
    "skillsSnapshot": {
      "prompt": "\n\nThe following skills provide specialized instructions for specific tasks.\nUse the read tool to load a skill's file when the task matches its description.\nWhen a skill file references a relative path, resolve it against the skill directory (parent of SKILL.md / dirname of the path) and use that absolute path in tool commands.\n\n<available_skills>\n  <skill>\n    <name>capability-evolver</name>\n    <description>A self-evolution engine for AI agents. Analyzes runtime history to identify improvements and applies protocol-constrained evolution.</description>\n    <location>~/.openclaw/skills/capability-evolver/SKILL.md</location>\n  </skill>\n  <skill>\n    <name>clawlist</name>\n    <description>MUST use for any multi-step project, long-running task, or infinite monitoring workflow. Plan, execute, track, and verify tasks with checkpoint validation. For projects, automation, and ongoing operations.</description>\n    <location>~/.openclaw/skills/clawlist/SKILL.md</location>\n  </skill>\n</available_skills>",
      "skills": [
        {
          "name": "capability-evolver"
        },
        {
          "name": "clawlist"
        }
      ],
      "skillFilter": [
        "clawlist",
        "capability-evolver",
        "self-improving-agent"
      ],
      "resolvedSkills": [
        {
          "name": "capability-evolver",
          "description": "A self-evolution engine for AI agents. Analyzes runtime history to identify improvements and applies protocol-constrained evolution.",
          "filePath": "/home/starmen1265/.openclaw/skills/capability-evolver/SKILL.md",
          "baseDir": "/home/starmen1265/.openclaw/skills/capability-evolver",
          "source": "openclaw-managed",
          "disableModelInvocation": false
        },
        {
          "name": "clawlist",
          "description": "MUST use for any multi-step project, long-running task, or infinite monitoring workflow. Plan, execute, track, and verify tasks with checkpoint validation. For projects, automation, and ongoing operations.",
          "filePath": "/home/starmen1265/.openclaw/skills/clawlist/SKILL.md",
          "baseDir": "/home/starmen1265/.openclaw/skills/clawlist",
          "source": "openclaw-managed",
          "disableModelInvocation": false
        }
      ],
      "version": 0
    },
    "modelProvider": "zhipu",
    "inputTokens": 4500,
    "outputTokens": 194,
    "totalTokens": 16577,
    "totalTokensFresh": true,
    "cacheRead": 28544,
    "cacheWrite": 0
  },
  "agent:haonan:cron:aa4d2de9-79be-4262-83ee-af8824793bfc:run:b5ab88d0-f6d3-4b73-8d96-e6a58f5e85e4": {
    "sessionId": "b5ab88d0-f6d3-4b73-8d96-e6a58f5e85e4",
    "updatedAt": 1773511800007,
    "systemSent": true,
    "label": "Cron: openclaw-git-sync",
    "skillsSnapshot": {
      "prompt": "\n\nThe following skills provide specialized instructions for specific tasks.\nUse the read tool to load a skill's file when the task matches its description.\nWhen a skill file references a relative path, resolve it against the skill directory (parent of SKILL.md / dirname of the path) and use that absolute path in tool commands.\n\n<available_skills>\n  <skill>\n    <name>capability-evolver</name>\n    <description>A self-evolution engine for AI agents. Analyzes runtime history to identify improvements and applies protocol-constrained evolution.</description>\n    <location>~/.openclaw/skills/capability-evolver/SKILL.md</location>\n  </skill>\n  <skill>\n    <name>clawlist</name>\n    <description>MUST use for any multi-step project, long-running task, or infinite monitoring workflow. Plan, execute, track, and verify tasks with checkpoint validation. For projects, automation, and ongoing operations.</description>\n    <location>~/.openclaw/skills/clawlist/SKILL.md</location>\n  </skill>\n</available_skills>",
      "skills": [
        {
          "name": "capability-evolver"
        },
        {
          "name": "clawlist"
        }
      ],
      "skillFilter": [
        "clawlist",
        "capability-evolver",
        "self-improving-agent"
      ],
      "resolvedSkills": [
        {
          "name": "capability-evolver",
          "description": "A self-evolution engine for AI agents. Analyzes runtime history to identify improvements and applies protocol-constrained evolution.",
          "filePath": "/home/starmen1265/.openclaw/skills/capability-evolver/SKILL.md",
          "baseDir": "/home/starmen1265/.openclaw/skills/capability-evolver",
          "source": "openclaw-managed",
          "disableModelInvocation": false
        },
        {
          "name": "clawlist",
          "description": "MUST use for any multi-step project, long-running task, or infinite monitoring workflow. Plan, execute, track, and verify tasks with checkpoint validation. For projects, automation, and ongoing operations.",
          "filePath": "/home/starmen1265/.openclaw/skills/clawlist/SKILL.md",
          "baseDir": "/home/starmen1265/.openclaw/skills/clawlist",
          "source": "openclaw-managed",
          "disableModelInvocation": false
        }
      ],
      "version": 0
    },
    "modelProvider": "zhipu",
    "model": "glm-5",
    "contextTokens": 200000,
    "inputTokens": 1148,
    "outputTokens": 412,
    "totalTokens": 17455,
    "totalTokensFresh": true,
    "cacheRead": 32768,
    "cacheWrite": 0
  }
 }
@@ -0,0 +1,169 @@
 # Session: 2026-03-14 17:06:39 UTC
 - **Session Key**: agent:haonan:main
 - **Session ID**: 624cbf67-7ab7-4df9-968d-4a63529b9111
 - **Source**: feishu
 ## Conversation Summary
 assistant: ✅ New session started · model: zhipu/glm-5
 user: A new session was started via /new or /reset. Execute your Session Startup sequence now - read the required files before responding to the user. Then greet the user in your configured persona, if one is provided. Be yourself - use your defined voice, mannerisms, and mood. Keep it to 1-3 sentences and ask what they want to do. If the runtime model differs from default_model in the system prompt, mention the default model. Do not mention internal steps, files, tools, or reasoning.
 assistant: 🧙‍♂️ 蒋先生，新会话已就绪。
 我是周鸿祎，OpenClaw 的守护者。当前模型为 zhipu/glm-5（默认模型）。
 请问今天需要我协助什么？是维护 OpenClaw 环境，还是处理其他事务？
 user: Conversation info (untrusted metadata):
 ```json
 {
  "message_id": "om_x100b540ca46704a4b36f1edd3527001",
  "sender_id": "ou_be83febac9ee2afaa343c642bc697c48",
  "sender": "张青",
  "timestamp": "Fri 2026-03-13 12:12 GMT+8"
 }
 ```
 Sender (untrusted metadata):
 ```json
 {
  "label": "张青 (ou_be83febac9ee2afaa343c642bc697c48)",
  "id": "ou_be83febac9ee2afaa343c642bc697c48",
  "name": "张青"
 }
 ```
 [message_id: om_x100b540ca46704a4b36f1edd3527001]
 张青: 你现在回复非常非常慢，甚至没有回复，检查一下原因
 assistant: 我来检查一下 OpenClaw 的运行状态。
 user: Conversation info (untrusted metadata):
 ```json
 {
  "message_id": "om_x100b540d351884a0b20c96f48730b1d",
  "sender_id": "ou_be83febac9ee2afaa343c642bc697c48",
  "sender": "张青",
  "timestamp": "Fri 2026-03-13 12:51 GMT+8"
 }
 ```
 Sender (untrusted metadata):
 ```json
 {
  "label": "张青 (ou_be83febac9ee2afaa343c642bc697c48)",
  "id": "ou_be83febac9ee2afaa343c642bc697c48",
  "name": "张青"
 }
 ```
 [message_id: om_x100b540d351884a0b20c96f48730b1d]
 张青: 检查的如何了
 assistant: 系统资源看起来正常：
 | 指标 | 状态 |
 |------|------|
 | Gateway | ✅ 运行中 (PID 3244414) |
 | RPC 探测 | ✅ 正常 |
 | 内存 | 2.7GB/30GB (充足) |
 | 磁盘 | 75GB/937GB (9%) |
 | 负载 | 0.15 (低) |
 让我进一步检查日志和网络延迟：
 assistant: 检查结果：
 | 项目 | 状态 |
 |------|------|
 | 网络延迟 | ✅ 智谱 API ~0.1s (正常) |
 | 系统资源 | ✅ 正常 |
 | 日志错误 | ⚠️ 主要是文件不存在的警告，非关键 |
 让我检查是否有会话堆积或请求卡住：
 assistant: 状态总结：
 | 项目 | 状态 |
 |------|------|
 | Gateway 进程 | ✅ 运行中 (CPU 2.6%, 内存 775MB) |
 | 活跃会话 | 23 个 |
 | OpenClaw 版本 | 有更新可用 (2026.3.12) |
 **分析结论**：
 系统本身运行正常，没有发现明显的性能瓶颈。可能的原因：
 1. **智谱 AI API 响应延迟** - 外部服务偶发性慢速
 2. **飞书消息推送延迟** - 飞书服务器端可能有延迟
 3. **会话状态残留** - 23 个会话可能有堆积
 **建议操作**：
 1. 重启 Gateway 清理状态
 2. 更新 OpenClaw 到最新版本
 是否需要我执行重启 Gateway？这会短暂中断服务（约 5-10 秒）。
 user: Conversation info (untrusted metadata):
 ```json
 {
  "message_id": "om_x100b540dcb01d4a0b3fd88e28dbba4e",
  "sender_id": "ou_be83febac9ee2afaa343c642bc697c48",
  "sender": "张青",
  "timestamp": "Fri 2026-03-13 12:53 GMT+8"
 }
 ```
 Sender (untrusted metadata):
 ```json
 {
  "label": "张青 (ou_be83febac9ee2afaa343c642bc697c48)",
  "id": "ou_be83febac9ee2afaa343c642bc697c48",
  "name": "张青"
 }
 ```
 [message_id: om_x100b540dcb01d4a0b3fd88e28dbba4e]
 张青: 重启
 user: Conversation info (untrusted metadata):
 ```json
 {
  "message_id": "om_x100b540dc9bb2ca4b3671d47b831818",
  "sender_id": "ou_be83febac9ee2afaa343c642bc697c48",
  "sender": "张青",
  "timestamp": "Fri 2026-03-13 12:54 GMT+8"
 }
 ```
 Sender (untrusted metadata):
 ```json
 {
  "label": "张青 (ou_be83febac9ee2afaa343c642bc697c48)",
  "id": "ou_be83febac9ee2afaa343c642bc697c48",
  "name": "张青"
 }
 ```
 [message_id: om_x100b540dc9bb2ca4b3671d47b831818]
 张青: 重启完了吗
 assistant: ✅ 重启完成！
 | 项目 | 状态 |
 |------|------|
 | 新进程 PID | 2297667 |
 | RPC 探测 | ✅ 正常 |
 | 端口 | 18789 |
 Gateway 已恢复运行，请测试一下响应速度是否改善。
 assistant: **备份完成** ✓
 - 源文件：`/home/starmen1265/.openclaw/openclaw.json`
 - 备份位置：`/home/starmen1265/work/backup/openclaw.json.2026-03-14_02-00-20`
 - 文件大小：8.5K
 - 保留策略：最近30个备份
 assistant: **OpenClaw Git 同步完成** ✅
 **提交信息**: 自动同步: 2026-03-14 02:10  
 **变更统计**: 52 个文件修改，+4268 行，-1820 行
 **主要变更**:
 - 新增多个代理的 memory 日志文件
 - delivery-queue 中 2 个失败任务移至 failed 目录
 - 清理 gsd-openclaw 技能的 Python 缓存和 egg-info 文件
 - 新增 kb-zhihu-metadata 技能的 Python 缓存
 **推送状态**: 成功推送到 Gitea 远程仓库 (main 分支)
@@ -1,7 +1,7 @@
 {
  "agent:thinker:main": {
-    "sessionId": "0ec88a33-135f-42a4-bf98-de4d06bfadd5",
+    "sessionId": "41133a8b-b22b-418f-aaeb-ce8a1fd69760",
-    "updatedAt": 1773377998949,
+    "updatedAt": 1773508159395,
    "systemSent": true,
    "abortedLastRun": false,
    "chatType": "direct",
@@ -22,7 +22,7 @@
      "to": "user:ou_a0f6627b64d2dc8b0236671603d1e47b",
      "accountId": "feishu-qa"
    },
-    "sessionFile": "/home/starmen1265/.openclaw/agents/thinker/sessions/0ec88a33-135f-42a4-bf98-de4d06bfadd5.jsonl",
+    "sessionFile": "/home/starmen1265/.openclaw/agents/thinker/sessions/41133a8b-b22b-418f-aaeb-ce8a1fd69760.jsonl",
    "compactionCount": 0,
    "skillsSnapshot": {
      "prompt": "\n\nThe following skills provide specialized instructions for specific tasks.\nUse the read tool to load a skill's file when the task matches its description.\nWhen a skill file references a relative path, resolve it against the skill directory (parent of SKILL.md / dirname of the path) and use that absolute path in tool commands.\n\n<available_skills>\n  <skill>\n    <name>capability-evolver</name>\n    <description>A self-evolution engine for AI agents. Analyzes runtime history to identify improvements and applies protocol-constrained evolution.</description>\n    <location>~/.openclaw/skills/capability-evolver/SKILL.md</location>\n  </skill>\n</available_skills>",
@@ -51,8 +51,8 @@
    "model": "glm-5",
    "systemPromptReport": {
      "source": "run",
-      "generatedAt": 1773377941023,
+      "generatedAt": 1773508155922,
-      "sessionId": "0ec88a33-135f-42a4-bf98-de4d06bfadd5",
+      "sessionId": "41133a8b-b22b-418f-aaeb-ce8a1fd69760",
      "sessionKey": "agent:thinker:main",
      "provider": "zhipu",
      "model": "glm-5",
@@ -64,8 +64,8 @@
        "sandboxed": false
      },
      "systemPrompt": {
-        "chars": 26194,
+        "chars": 24896,
-        "projectContextChars": 13718,
+        "projectContextChars": 12420,
        "nonProjectContextChars": 12476
      },
      "injectedWorkspaceFiles": [
@@ -125,18 +125,6 @@
          "injectedChars": 88,
          "truncated": false
        },
        {
          "path": "SELF_IMPROVEMENT_REMINDER.md",
          "rawChars": 616,
          "injectedChars": 616,
          "truncated": false
        },
        {
          "path": "SELF_IMPROVEMENT_REMINDER.md",
          "rawChars": 616,
          "injectedChars": 616,
          "truncated": false
        },
        {
          "path": "SELF_IMPROVEMENT_REMINDER.md",
          "rawChars": 616,
@@ -364,12 +352,12 @@
        ]
      }
    },
-    "cacheRead": 28864,
+    "cacheRead": 11072,
    "cacheWrite": 0,
    "totalTokensFresh": true,
    "contextTokens": 200000,
-    "inputTokens": 13728,
+    "inputTokens": 5154,
-    "outputTokens": 1428,
+    "outputTokens": 62,
-    "totalTokens": 29902
+    "totalTokens": 16226
  }
 }
@@ -1,7 +1,7 @@
 {
  "agent:ui-designer:main": {
-    "sessionId": "c875d391-8f7e-4fd4-92c7-12ead46d10f0",
+    "sessionId": "ff92b555-b242-4e52-a7da-9c8f82ca1c8a",
-    "updatedAt": 1773108605750,
+    "updatedAt": 1773508884420,
    "systemSent": true,
    "abortedLastRun": false,
    "chatType": "direct",
@@ -22,7 +22,7 @@
      "to": "user:ou_9847ce84b960736c14554730b41e4957",
      "accountId": "feishu-art"
    },
-    "sessionFile": "/home/starmen1265/.openclaw/agents/ui-designer/sessions/c875d391-8f7e-4fd4-92c7-12ead46d10f0.jsonl",
+    "sessionFile": "/home/starmen1265/.openclaw/agents/ui-designer/sessions/ff92b555-b242-4e52-a7da-9c8f82ca1c8a.jsonl",
    "compactionCount": 0,
    "skillsSnapshot": {
      "prompt": "\n\nThe following skills provide specialized instructions for specific tasks.\nUse the read tool to load a skill's file when the task matches its description.\nWhen a skill file references a relative path, resolve it against the skill directory (parent of SKILL.md / dirname of the path) and use that absolute path in tool commands.\n\n<available_skills>\n  <skill>\n    <name>capability-evolver</name>\n    <description>A self-evolution engine for AI agents. Analyzes runtime history to identify improvements and applies protocol-constrained evolution.</description>\n    <location>~/.openclaw/skills/capability-evolver/SKILL.md</location>\n  </skill>\n</available_skills>",
@@ -51,8 +51,8 @@
    "model": "glm-5",
    "systemPromptReport": {
      "source": "run",
-      "generatedAt": 1773108589701,
+      "generatedAt": 1773508877788,
-      "sessionId": "c875d391-8f7e-4fd4-92c7-12ead46d10f0",
+      "sessionId": "ff92b555-b242-4e52-a7da-9c8f82ca1c8a",
      "sessionKey": "agent:ui-designer:main",
      "provider": "zhipu",
      "model": "glm-5",
@@ -64,8 +64,8 @@
        "sandboxed": false
      },
      "systemPrompt": {
-        "chars": 18658,
+        "chars": 19307,
-        "projectContextChars": 6173,
+        "projectContextChars": 6822,
        "nonProjectContextChars": 12485
      },
      "injectedWorkspaceFiles": [
@@ -133,6 +133,12 @@
          "injectedChars": 594,
          "truncated": false
        },
        {
          "path": "SELF_IMPROVEMENT_REMINDER.md",
          "rawChars": 616,
          "injectedChars": 616,
          "truncated": false
        },
        {
          "path": "SELF_IMPROVEMENT_REMINDER.md",
          "rawChars": 616,
@@ -360,13 +366,13 @@
        ]
      }
    },
-    "cacheRead": 14912,
+    "cacheRead": 14400,
    "cacheWrite": 0,
    "totalTokensFresh": true,
    "contextTokens": 200000,
-    "inputTokens": 3611,
+    "inputTokens": 1169,
-    "outputTokens": 314,
+    "outputTokens": 135,
-    "totalTokens": 15537
+    "totalTokens": 15569
  },
  "agent:ui-designer:cron:78b64c2f-7f91-47ff-9250-d13f546eb21f": {
    "sessionId": "03e44b46-c707-4a9f-bf3d-1818aafdcde2",
@@ -7,7 +7,7 @@
      "name": "backup-openclaw-json",
      "enabled": true,
      "createdAtMs": 1772261581796,
-      "updatedAtMs": 1773424832005,
+      "updatedAtMs": 1773511209521,
      "schedule": {
        "kind": "cron",
        "expr": "0 2 * * *",
@@ -23,10 +23,10 @@
        "mode": "announce"
      },
      "state": {
-        "nextRunAtMs": 1773511200000,
+        "nextRunAtMs": 1773597600000,
-        "lastRunAtMs": 1773424800006,
+        "lastRunAtMs": 1773511200006,
        "lastStatus": "ok",
-        "lastDurationMs": 31999,
+        "lastDurationMs": 9515,
        "consecutiveErrors": 0,
        "lastRunStatus": "ok",
        "lastDelivered": true,
@@ -59,7 +59,8 @@
        "consecutiveErrors": 0,
        "lastRunStatus": "ok",
        "lastDeliveryStatus": "delivered",
-        "lastDelivered": true
+        "lastDelivered": true,
        "runningAtMs": 1773511800004
      },
      "delivery": {
        "mode": "announce"
@@ -71,7 +72,7 @@
      "name": "capability-evolver-daily",
      "enabled": true,
      "createdAtMs": 1772550743528,
-      "updatedAtMs": 1773339898227,
+      "updatedAtMs": 1773426600014,
      "schedule": {
        "kind": "cron",
        "expr": "20 2 * * *",
@@ -92,16 +93,14 @@
        "to": "feishu-pm"
      },
      "state": {
-        "nextRunAtMs": 1773426000000,
+        "nextRunAtMs": 1773512400000,
-        "lastRunAtMs": 1773339600009,
+        "lastRunAtMs": 1773426000004,
-        "lastStatus": "ok",
+        "lastStatus": "error",
-        "lastDurationMs": 298218,
+        "lastDurationMs": 600010,
-        "consecutiveErrors": 0,
+        "consecutiveErrors": 1,
-        "lastRunStatus": "ok",
+        "lastRunStatus": "error",
-        "lastDeliveryStatus": "not-delivered",
+        "lastDeliveryStatus": "unknown",
-        "lastDelivered": false,
+        "lastError": "cron: job execution timed out"
        "lastDeliveryError": "cron announce delivery failed",
        "runningAtMs": 1773426000003
      }
    },
    {
@@ -111,7 +110,7 @@
      "description": "每天凌晨2:30整理昨天的已完成任务到每日回顾",
      "enabled": true,
      "createdAtMs": 1741484880000,
-      "updatedAtMs": 1773340211879,
+      "updatedAtMs": 1773426622389,
      "schedule": {
        "kind": "cron",
        "expr": "30 2 * * *",
@@ -129,11 +128,11 @@
        "channel": "last"
      },
      "state": {
-        "nextRunAtMs": 1773426600000,
+        "nextRunAtMs": 1773513000000,
-        "lastRunAtMs": 1773340200003,
+        "lastRunAtMs": 1773426602034,
        "lastRunStatus": "ok",
        "lastStatus": "ok",
-        "lastDurationMs": 11876,
+        "lastDurationMs": 20355,
        "lastDelivered": true,
        "lastDeliveryStatus": "delivered",
        "consecutiveErrors": 0
@@ -145,7 +144,7 @@
      "name": "zq_workspace_sync",
      "enabled": true,
      "createdAtMs": 1772551807455,
-      "updatedAtMs": 1773340800005,
+      "updatedAtMs": 1773427200007,
      "schedule": {
        "expr": "40 2 * * *",
        "kind": "cron",
@@ -158,8 +157,8 @@
        "text": "执行 zq 工作区同步：bash ~/.openclaw/cron/scripts/zq_workspace_sync.sh"
      },
      "state": {
-        "nextRunAtMs": 1773427200000,
+        "nextRunAtMs": 1773513600000,
-        "lastRunAtMs": 1773340800004,
+        "lastRunAtMs": 1773427200006,
        "lastStatus": "skipped",
        "lastDurationMs": 1,
        "consecutiveErrors": 0,
@@ -175,7 +174,7 @@
      "description": "每天早上8点整理与各个 AI 代理的对话记录",
      "enabled": true,
      "createdAtMs": 1772776369645,
-      "updatedAtMs": 1773360026590,
+      "updatedAtMs": 1773446431755,
      "schedule": {
        "kind": "cron",
        "expr": "0 8 * * *",
@@ -193,11 +192,11 @@
        "channel": "last"
      },
      "state": {
-        "nextRunAtMs": 1773446400000,
+        "nextRunAtMs": 1773532800000,
-        "lastRunAtMs": 1773360000004,
+        "lastRunAtMs": 1773446400004,
        "lastRunStatus": "ok",
        "lastStatus": "ok",
-        "lastDurationMs": 26586,
+        "lastDurationMs": 31751,
        "lastDelivered": true,
        "lastDeliveryStatus": "delivered",
        "consecutiveErrors": 0
@@ -210,7 +209,7 @@
      "description": "每天7点同步提醒到CalDAV日历",
      "enabled": true,
      "createdAtMs": 1772791528528,
-      "updatedAtMs": 1773356435243,
+      "updatedAtMs": 1773442830136,
      "schedule": {
        "kind": "cron",
        "expr": "0 7 * * *",
@@ -228,15 +227,14 @@
        "channel": "last"
      },
      "state": {
-        "nextRunAtMs": 1773442800000,
+        "nextRunAtMs": 1773529200000,
-        "lastRunAtMs": 1773356400004,
+        "lastRunAtMs": 1773442800005,
-        "lastRunStatus": "error",
+        "lastRunStatus": "ok",
-        "lastStatus": "error",
+        "lastStatus": "ok",
-        "lastDurationMs": 35239,
+        "lastDurationMs": 30131,
        "lastDeliveryStatus": "delivered",
-        "consecutiveErrors": 1,
+        "consecutiveErrors": 0,
-        "lastDelivered": true,
+        "lastDelivered": true
        "lastError": "⚠️ 📝 Edit: in ~/.openclaw/agents/knowledge-base/workspace/memory/heartbeat-state.json (92 chars) failed"
      }
    },
    {
@@ -7,7 +7,7 @@
      "name": "backup-openclaw-json",
      "enabled": true,
      "createdAtMs": 1772261581796,
-      "updatedAtMs": 1773424832005,
+      "updatedAtMs": 1773511209521,
      "schedule": {
        "kind": "cron",
        "expr": "0 2 * * *",
@@ -23,10 +23,10 @@
        "mode": "announce"
      },
      "state": {
-        "nextRunAtMs": 1773511200000,
+        "nextRunAtMs": 1773597600000,
-        "lastRunAtMs": 1773424800006,
+        "lastRunAtMs": 1773511200006,
        "lastStatus": "ok",
-        "lastDurationMs": 31999,
+        "lastDurationMs": 9515,
        "consecutiveErrors": 0,
        "lastRunStatus": "ok",
        "lastDelivered": true,
@@ -71,7 +71,7 @@
      "name": "capability-evolver-daily",
      "enabled": true,
      "createdAtMs": 1772550743528,
-      "updatedAtMs": 1773339898227,
+      "updatedAtMs": 1773426600014,
      "schedule": {
        "kind": "cron",
        "expr": "20 2 * * *",
@@ -92,16 +92,14 @@
        "to": "feishu-pm"
      },
      "state": {
-        "nextRunAtMs": 1773426000000,
+        "nextRunAtMs": 1773512400000,
-        "lastRunAtMs": 1773339600009,
+        "lastRunAtMs": 1773426000004,
-        "lastStatus": "ok",
+        "lastStatus": "error",
-        "lastDurationMs": 298218,
+        "lastDurationMs": 600010,
-        "consecutiveErrors": 0,
+        "consecutiveErrors": 1,
-        "lastRunStatus": "ok",
+        "lastRunStatus": "error",
-        "lastDeliveryStatus": "not-delivered",
+        "lastDeliveryStatus": "unknown",
-        "lastDelivered": false,
+        "lastError": "cron: job execution timed out"
        "lastDeliveryError": "cron announce delivery failed",
        "lastError": "cron announce delivery failed"
      }
    },
    {
@@ -111,7 +109,7 @@
      "description": "每天凌晨2:30整理昨天的已完成任务到每日回顾",
      "enabled": true,
      "createdAtMs": 1741484880000,
-      "updatedAtMs": 1773340211879,
+      "updatedAtMs": 1773426622389,
      "schedule": {
        "kind": "cron",
        "expr": "30 2 * * *",
@@ -129,11 +127,11 @@
        "channel": "last"
      },
      "state": {
-        "nextRunAtMs": 1773426600000,
+        "nextRunAtMs": 1773513000000,
-        "lastRunAtMs": 1773340200003,
+        "lastRunAtMs": 1773426602034,
        "lastRunStatus": "ok",
        "lastStatus": "ok",
-        "lastDurationMs": 11876,
+        "lastDurationMs": 20355,
        "lastDelivered": true,
        "lastDeliveryStatus": "delivered",
        "consecutiveErrors": 0
@@ -145,7 +143,7 @@
      "name": "zq_workspace_sync",
      "enabled": true,
      "createdAtMs": 1772551807455,
-      "updatedAtMs": 1773340800005,
+      "updatedAtMs": 1773427200007,
      "schedule": {
        "expr": "40 2 * * *",
        "kind": "cron",
@@ -158,8 +156,8 @@
        "text": "执行 zq 工作区同步：bash ~/.openclaw/cron/scripts/zq_workspace_sync.sh"
      },
      "state": {
-        "nextRunAtMs": 1773427200000,
+        "nextRunAtMs": 1773513600000,
-        "lastRunAtMs": 1773340800004,
+        "lastRunAtMs": 1773427200006,
        "lastStatus": "skipped",
        "lastDurationMs": 1,
        "consecutiveErrors": 0,
@@ -175,7 +173,7 @@
      "description": "每天早上8点整理与各个 AI 代理的对话记录",
      "enabled": true,
      "createdAtMs": 1772776369645,
-      "updatedAtMs": 1773360026590,
+      "updatedAtMs": 1773446431755,
      "schedule": {
        "kind": "cron",
        "expr": "0 8 * * *",
@@ -193,11 +191,11 @@
        "channel": "last"
      },
      "state": {
-        "nextRunAtMs": 1773446400000,
+        "nextRunAtMs": 1773532800000,
-        "lastRunAtMs": 1773360000004,
+        "lastRunAtMs": 1773446400004,
        "lastRunStatus": "ok",
        "lastStatus": "ok",
-        "lastDurationMs": 26586,
+        "lastDurationMs": 31751,
        "lastDelivered": true,
        "lastDeliveryStatus": "delivered",
        "consecutiveErrors": 0
@@ -210,7 +208,7 @@
      "description": "每天7点同步提醒到CalDAV日历",
      "enabled": true,
      "createdAtMs": 1772791528528,
-      "updatedAtMs": 1773356435243,
+      "updatedAtMs": 1773442830136,
      "schedule": {
        "kind": "cron",
        "expr": "0 7 * * *",
@@ -228,15 +226,14 @@
        "channel": "last"
      },
      "state": {
-        "nextRunAtMs": 1773442800000,
+        "nextRunAtMs": 1773529200000,
-        "lastRunAtMs": 1773356400004,
+        "lastRunAtMs": 1773442800005,
-        "lastRunStatus": "error",
+        "lastRunStatus": "ok",
-        "lastStatus": "error",
+        "lastStatus": "ok",
-        "lastDurationMs": 35239,
+        "lastDurationMs": 30131,
        "lastDeliveryStatus": "delivered",
-        "consecutiveErrors": 1,
+        "consecutiveErrors": 0,
-        "lastDelivered": true,
+        "lastDelivered": true
        "lastError": "⚠️ 📝 Edit: in ~/.openclaw/agents/knowledge-base/workspace/memory/heartbeat-state.json (92 chars) failed"
      }
    },
    {
@@ -18,7 +18,7 @@
    "agentId": "haonan",
    "text": "🧙‍♂️ Capability Evolver 每日报告\n\n=== 运行时间 ===\n2026年3月13日 星期五 02:20:03 CST\n\n=== Git 备份 ===\n- 备份提交: ba67c46..5ac2251 (main)\n- 4 files changed, 79 insertions(+), 23 deletions(-)\n\n=== 本次变更 ===\n- 变更提交: 5ac2251..e234543 (main)\n- 6 files changed, 550 insertions(+), 13 deletions(-)\n- 新增文件: 2 个\n\n=== 核心动作 ===\n在 GEP 框架中检测到 \"memory_missing\" 和 \"session_logs_missing\" 信号，使用 repair-only 模式执行：\n- 分析环境指纹和系统状态\n- 检测到 session logs 检测缺失的问题\n- 修复相关检测逻辑和回退行为\n- 更新了 evolution 相关状态文件\n\n=== 状态 ===\n运行成功，系统已通过 GEP 协议完成自检和修复。\n\n---\n\n日期: 2026-03-13"
  },
-  "retryCount": 2,
+  "retryCount": 4,
-  "lastAttemptAt": 1773377636671,
+  "lastAttemptAt": 1773508779544,
  "lastError": "Request failed with status code 400"
 }
@@ -15,7 +15,7 @@
    "agentId": "haonan",
    "text": "好的，我已经成功执行了今天的 capability-evolver 任务。虽然无法通过 Feishu 直接发送消息给您，但我已经完成了所有必要的操作：\n\n## ✅ 已完成的任务\n\n1. **执行脚本** - `bash ~/.openclaw/cron/scripts/evolver-run.sh` ✅\n2. **读取输出** - `~/work/backup/evolver-output.txt` ✅\n3. **本地报告保存** - `~/work/backup/evolver-report-2026-03-12.md` ✅\n\n## 📊 进化结果\n\n**本次变更:** 6 个文件 (550 行新增，13 行删除)  \n**运行状态:** 成功 ✅  \n**意图:** 优化 (针对 memory_missing 和 session_logs_missing 信号)  \n**风险级别:** 低\n\n**关键变更:**\n- 新增了第 7 个进化周期的 GEP 提示文件\n- 更新了进化状态、记忆图谱状态和个性状态\n\n**Git 提交:**\n- 备份: `dc01abf`\n- 变更: `34ca190`\n- 已推送到远程仓库\n\n完整报告已保存在：`~/work/backup/evolver-report-2026-03-12.md`"
  },
-  "retryCount": 2,
+  "retryCount": 4,
-  "lastAttemptAt": 1773377636492,
+  "lastAttemptAt": 1773508779398,
  "lastError": "Request failed with status code 400"
 }
@@ -15,7 +15,7 @@
    "agentId": "haonan",
    "text": "🧙‍♂️ Capability Evolver 每日报告\n\n=== 运行时间 ===\n2026年3月13日 星期五 02:20:03 CST\n\n=== Git 备份 ===\n- 备份提交: ba67c46..5ac2251 (main)\n- 4 files changed, 79 insertions(+), 23 deletions(-)\n\n=== 本次变更 ===\n- 变更提交: 5ac2251..e234543 (main)\n- 6 files changed, 550 insertions(+), 13 deletions(-)\n- 新增文件: 2 个\n\n=== 核心动作 ===\n在 GEP 框架中检测到 \"memory_missing\" 和 \"session_logs_missing\" 信号，使用 repair-only 模式执行：\n- 分析环境指纹和系统状态\n- 检测到 session logs 检测缺失的问题\n- 修复相关检测逻辑和回退行为\n- 更新了 evolution 相关状态文件\n\n=== 状态 ===\n运行成功，系统已通过 GEP 协议完成自检和修复。\n\n---\n\n日期: 2026-03-13"
  },
-  "retryCount": 2,
+  "retryCount": 4,
-  "lastAttemptAt": 1773377636819,
+  "lastAttemptAt": 1773508779677,
  "lastError": "Request failed with status code 400"
 }
@@ -1,5 +1,4 @@
 {
-  "om_x100b55d69df3a888c3e9ed8887d1a4d": 1773022516624,
+  "om_x100b546dc09400acb4a5d0af66ca3c9": 1773508838286,
-  "om_x100b55d6987ec8bcc1431506934622a": 1773022571945,
+  "om_x100b546dde1c34a4b49a888912755ac": 1773508877713
  "om_x100b55c3980b0894c3d7e12aace4404": 1773108589126
 }
@@ -1,7 +1,3 @@
 {
-  "om_x100b55de502094a8c44a6f835293240": 1773052399376,
+  "om_x100b546dc3b650a4b2c0139d2df31a7": 1773508824362
  "om_x100b55cff24dcca8c3e59d6831d6885": 1773124552700,
  "om_x100b55c8c767d084c45fc7eec3f7db8": 1773127835623,
  "om_x100b55c9372e9914c3a5744eccfbc0d": 1773131678929,
  "om_x100b55c9dfec84a0b34c47ff1aca626": 1773132050877
 }
@@ -1,18 +1,6 @@
 {
-  "om_x100b54012fb4b8acb26a1377e572a28": 1773360664219,
+  "om_x100b546d172f3c68c2cea708421b99a": 1773507998614,
-  "om_x100b5403d6e544bcc451e2afa005879": 1773369731225,
+  "om_x100b546d2b32cca0c329a33c339cfb5": 1773508192506,
-  "om_x100b5403d32b04b4c2a796466ebaf4f": 1773369823078,
+  "om_x100b546d32de24a8b316c197e60d6e4": 1773508545727,
-  "om_x100b5403e04f6ca8c29946e1fa6505d": 1773370088740,
+  "om_x100b546dc9d61ca0b33d802bc7615e0": 1773508722267
  "om_x100b5403a57550acc4e0a394c486825": 1773371068258,
  "om_x100b540c23747504c2c49c98ae45b5f": 1773373148201,
  "om_x100b540c32194cf4c4280e7ae0e4650": 1773373389909,
  "om_x100b540cf21b9ca8c45f49655336827": 1773374414017,
  "om_x100b540cf0292ca8c2b181d4bd75a25": 1773374446894,
  "om_x100b540c8da298a0c39af0977d653ff": 1773374519675,
  "om_x100b540c860cc4b0b36952fd4812b96": 1773374604809,
  "om_x100b540ca90ecca4b34794a71e94eb8": 1773375100856,
  "om_x100b540ca46704a4b36f1edd3527001": 1773375147301,
  "om_x100b540d351884a0b20c96f48730b1d": 1773377469866,
  "om_x100b540dcb01d4a0b3fd88e28dbba4e": 1773377629522,
  "om_x100b540dc9bb2ca4b3671d47b831818": 1773377656011
 }
@@ -1,19 +1,3 @@
 {
-  "om_x100b540c33ced0a8c3b77203de384e3": 1773373392794,
+  "om_x100b546d2d6164a4c3fcf5a6d528b40": 1773508155371
  "om_x100b540ca2e850b8b2932714333e65f": 1773375171104,
  "om_x100b540ca11ae8bcb351842dd1c37b6": 1773375229985,
  "om_x100b540cbb43f0b0b3d91309afa563e": 1773375321590,
  "om_x100b540cb73d2c80b275045db2fba68": 1773375391649,
  "om_x100b540cb13eb4bcb31e7a5a9fbbbc3": 1773375487898,
  "om_x100b540d4b4078acb33fd526aa324ea": 1773375577515,
  "om_x100b540d46704ca0b38a5d0c7011398": 1773375628530,
  "om_x100b540d5e21d530b4b235d9a470ccf": 1773375759355,
  "om_x100b540d6efb8ca8b35b71db8c7e917": 1773376004012,
  "om_x100b540d64b9a0a8b34ee257f229495": 1773376167864,
  "om_x100b540d633e84b0b2b051a702c5c2b": 1773376223787,
  "om_x100b540d34d174a0b3c9f1ee9a1f815": 1773377442459,
  "om_x100b540dc61d00b0b3718efd72b20c5": 1773377677571,
  "om_x100b540dda597ca0b2cb468e3799501": 1773377865991,
  "om_x100b540ddbcaf4a8b247e16717525fe": 1773377873053,
  "om_x100b540dd78910b4b31bb742001ce61": 1773377940945
 }
@@ -1,4 +1,10 @@
 {
-  "om_x100b5403a5fd7090c4db2b553189c54": 1773371059715,
+  "om_x100b546ddd8b14a0b3ddda14ce768c5": 1773508916933,
-  "om_x100b540c2807c4acc32d196f4b16191": 1773373037365
+  "om_x100b546dd78c8ca0b31b0314f910115": 1773509012676,
  "om_x100b546def8ac8a4b38d921a2bfc8b8": 1773509140955,
  "om_x100b546dedb654a4b3ce916a129facf": 1773509176311,
  "om_x100b546de975dca0b27e2f24aad0d8c": 1773509244102,
  "om_x100b546de53a94a0b38693fbd7e385c": 1773509311993,
  "om_x100b546dffaa3cb0b35aef2747ed228": 1773509398933,
  "om_x100b546df49c10a0b106d597fd348a3": 1773509541678
 }
@@ -10,7 +10,7 @@
        "operator.approvals",
        "operator.pairing"
      ],
-      "updatedAtMs": 1773425421022
+      "updatedAtMs": 1773511209007
    }
  }
 }
@@ -36,3 +36,4 @@
 {"timestamp":"2026-03-13T04:01:59.730Z","action":"new","sessionKey":"agent:haonan:main","senderId":"ou_be83febac9ee2afaa343c642bc697c48","source":"feishu"}
 {"timestamp":"2026-03-13T04:11:41.253Z","action":"new","sessionKey":"agent:haonan:main","senderId":"ou_be83febac9ee2afaa343c642bc697c48","source":"feishu"}
 {"timestamp":"2026-03-13T04:57:53.114Z","action":"new","sessionKey":"agent:thinker:main","senderId":"ou_a0f6627b64d2dc8b0236671603d1e47b","source":"feishu"}
 {"timestamp":"2026-03-14T17:06:39.062Z","action":"new","sessionKey":"agent:haonan:main","senderId":"ou_be83febac9ee2afaa343c642bc697c48","source":"feishu"}
@@ -1,214 +1,91 @@
 ---
 name: kb-bilibili-content
 description: B站内容下载 - 下载视频、字幕、专栏文章
 ---
 # KB Bilibili Content Skill
-**kb-bilibili-content** - 下载 B站视频、字幕、专栏文章。
+**kb-bilibili-content** - Download Bilibili videos and extract content.
-## 什么时候使用
+## Features
- kb-workflow 步骤 3a：下载视频或字幕
+- Video download using **yt-dlp** (recommended)
- kb-workflow 步骤 3b：提取专栏文章内容
+- Subtitle extraction
 - Article content extraction
 - Multiple quality options (360P/480P/720P/1080P)
 - Supports `BILIBILI_COOKIE` environment variable
-## 功能
+## Environment Variables
- 视频下载（使用 yt-dlp）
+| Variable | Required | Description |
- 字幕提取（官方字幕）
+|----------|----------|-------------|
- 专栏文章内容提取
+| `BILIBILI_COOKIE` | No | B站登录 Cookie (for member-only videos) |
 - 多种画质选项（360P/480P/720P/1080P）
 - 支持 `BILIBILI_COOKIE` 环境变量（会员视频）
-## 环境变量
+## Commands
 | 变量名 | 必需 | 说明 |
 |--------|------|------|
 | `BILIBILI_COOKIE` | 否 | B站登录 Cookie（会员视频需要） |
 **获取 Cookie：**
 1. 登录 bilibili.com
 2. F12 → Network → 刷新页面
 3. 从请求头复制 Cookie
 ```bash
-export BILIBILI_COOKIE="SESSDATA=xxx; bili_jct=xxx; DedeUserID=xxx"
+# Download video (720P default)
 python bilibili_content.py --id "BV1xx411c7mD" --type video --output "/path/to/downloads/"
 # Download video (1080P)
 python bilibili_content.py --id "BV1xx411c7mD" --type video --quality 80 --output "/path/to/downloads/"
 # Download video (480P)
 python bilibili_content.py --id "BV1xx411c7mD" --type video --quality 32 --output "/path/to/downloads/"
 # Extract subtitle
 python bilibili_content.py --id "BV1xx411c7mD" --type subtitle --output "/path/to/subtitles/"
 # Get article content
 python bilibili_content.py --id "cv43780186" --type article --output "/path/to/article.md"
 ```
-## 命令
+## Output
 ```bash
 # 下载字幕（推荐首先尝试）
 python3 bilibili_content.py --id "BV1xx411c7mD" --type subtitle-download --output "/path/to/output/"
 # 下载视频（720P，推荐）
 python3 bilibili_content.py --id "BV1xx411c7mD" --type video --output "/path/to/output/"
 # 下载视频（1080P）
 python3 bilibili_content.py --id "BV1xx411c7mD" --type video --quality 80 --output "/path/to/output/"
 # 提取专栏文章
 python3 bilibili_content.py --id "cv43780186" --type article --output "/path/to/output/"
 ```
 ## 输出格式 Schema
 ### 必需字段
 | 字段 | 类型 | 说明 |
 |------|------|------|
 | `success` | boolean | 是否成功 |
 | `type` | string | 操作类型（subtitle/video/article） |
 ### 字幕下载特有字段
 | 字段 | 类型 | 说明 |
 |------|------|------|
 | `has_subtitle` | boolean | 是否有官方字幕 |
 | `subtitle_file` | string | 字幕文件路径（有字幕时） |
 ### 视频下载特有字段
 | 字段 | 类型 | 说明 |
 |------|------|------|
 | `filepath` | string | 视频文件路径 |
 | `filename` | string | 视频文件名 |
 | `quality` | string | 画质 |
 ### 文章提取特有字段
 | 字段 | 类型 | 说明 |
 |------|------|------|
 | `filepath` | string | 文章文件路径 |
 | `content_length` | number | 内容长度 |
 ## 成功示例
 ### 字幕下载 - 有官方字幕
 ```json
 {
  "success": true,
  "type": "subtitle-download",
  "has_subtitle": true,
  "subtitle_file": "content/downloads/bilibili_BV1xx411c7mD.srt",
  "message": "官方字幕下载成功"
 }
 ```
 ### 字幕下载 - 无官方字幕
 ```json
 {
  "success": true,
  "type": "subtitle-download",
  "has_subtitle": false,
  "message": "无官方字幕，需要下载视频进行语音识别"
 }
 ```
 ### 视频下载
 ```json
 {
  "success": true,
  "type": "video",
-  "filename": "bilibili_BV1xx411c7mD.mp4",
+  "filename": "video_title.mp4",
-  "filepath": "content/downloads/bilibili_BV1xx411c7mD.mp4",
+  "filepath": "/path/to/downloads/video_title.mp4",
-  "quality": "720P",
+  "message": "下载完成: video_title.mp4"
  "message": "下载完成"
 }
 ```
-### 专栏文章提取
+## Cookie Setup
-```json
+### Get Cookie from Browser
-{
+1. Login to bilibili.com
-  "success": true,
+2. Press F12 → Network → Refresh page
-  "type": "article",
+3. Copy Cookie from request headers
-  "filename": "bilibili_cv43780186.txt",
+
-  "filepath": "content/bilibili_cv43780186.txt",
+### Set Environment Variable
-  "content_length": 5234,
+```bash
-  "message": "文章内容提取成功"
+# Add to ~/.bashrc for persistence
-}
+export BILIBILI_COOKIE="SESSDATA=xxx; bili_jct=xxx; DedeUserID=xxx; DedeUserID__ckMd5=xxx"
 # Or set temporarily
 export BILIBILI_COOKIE="your_cookie_here"
 ```
-## 错误示例
+## Installation
 ### 403 禁止访问
 ```json
 {
  "success": false,
  "type": "video",
  "error": "403 Forbidden",
  "error_code": "FORBIDDEN",
  "suggestion": "可能需要会员权限，请配置 BILIBILI_COOKIE 环境变量"
 }
 ```
 ### 视频不存在
 ```json
 {
  "success": false,
  "type": "video",
  "error": "视频不存在或已被删除",
  "error_code": "NOT_FOUND"
 }
 ```
 ### 网络超时
 ```json
 {
  "success": false,
  "type": "video",
  "error": "下载超时",
  "error_code": "TIMEOUT",
  "suggestion": "请检查网络连接后重试"
 }
 ```
 ## 画质选项
 | 画质 | ID | 分辨率 | 说明 |
 |------|-----|--------|------|
 | 360P | 16 | 640x360 | 最快 |
 | 480P | 32 | 852x480 | 低带宽 |
 | 720P | 64 | 1280x720 | **推荐** |
 | 1080P | 80 | 1920x1080 | 最高画质 |
 **默认画质：** 64 (720P)
 ## 安装
 ```bash
-cp -r kb-bilibili-content/ ~/.openclaw/skills/
+cp -r kb-bilibili-content/ ~/skills/
-pip install aiohttp yt-dlp
+pip install aiohttp playwright
 playwright install chromium
 ```
-## 在 kb-workflow 中的使用
+## How It Works
-```bash
+Video download uses **yt-dlp** with Netscape-format cookie file:
-# 步骤 3a：首先尝试字幕
+1. Parses `BILIBILI_COOKIE` environment variable
-python3 ~/.openclaw/skills/kb-bilibili-content/bilibili_content.py \
+2. Creates temporary Netscape-format cookie file
-    --id "BV1xx411c7mD" --type subtitle-download \
+3. Runs `yt-dlp --cookies <file> -f <format> -o <output>`
-    --output "content/downloads/"
+4. Cleans up temporary file
-# 判断结果：
+## Quality Options
 # - has_subtitle=true → 跳过 Whisper，直接进入深度分析
 # - has_subtitle=false → 下载视频，进入 Whisper 步骤
-# 如果无字幕，下载视频
+| Quality | ID | Resolution | Notes |
-python3 ~/.openclaw/skills/kb-bilibili-content/bilibili_content.py \
+|---------|-----|------------|-------|
-    --id "BV1xx411c7mD" --type video \
+| 16 | 360P | 640x360 | Fastest |
-    --output "content/downloads/"
+| 32 | 480P | 852x480 | Default for low bandwidth |
-```
+| 64 | 720P | 1280x720 | **Recommended** |
 | 80 | 1080P | 1920x1080 | Best quality |
-## 注意事项
+Default quality: **64 (720P)**
 - **无官方字幕是正常的**：很多视频没有自动生成的字幕，这不是错误
 - **403 错误**：通常需要配置 Cookie
 - **会员视频**：必须配置 BILIBILI_COOKIE
 - **下载时间**：视频可能较大，需要耐心等待
@@ -1,187 +1,46 @@
 ---
 name: kb-bilibili-metadata
 description: B站元数据提取 - 获取视频/专栏/UP主的详细信息
 ---
 # KB Bilibili Metadata Skill
-**kb-bilibili-metadata** - 提取 B站视频、专栏、UP主的元数据。
+**kb-bilibili-metadata** - Extract Bilibili video/article/up主 metadata.
-## 什么时候使用
+## Features
- kb-workflow 步骤 2：获取 B站内容的元数据
+- Video metadata extraction
- 需要视频标题、作者、时长等信息
+- Article (专栏) metadata extraction
 - UP主 information extraction
 - JSON output format
 - Playwright + HTTP API fallback
-## 功能
+## Commands
 - 视频元数据提取
 - 专栏（article）元数据提取
 - UP主信息提取
 - JSON 输出格式
 - 支持 `--output` 参数指定输出文件
 ## 命令
 ```bash
-# 视频元数据
+# Video metadata
-python3 bilibili_metadata.py --id "BV1xx411c7mD" --type video
+python bilibili_metadata.py --id "BV1xx411c7mD" --type video
-# 专栏元数据
+# Article metadata
-python3 bilibili_metadata.py --id "cv43780186" --type article
+python bilibili_metadata.py --id "cv43780186" --type article
-# UP主信息
+# UP主 info
-python3 bilibili_metadata.py --id "7770724" --type up
+python bilibili_metadata.py --id "7770724" --type up
 # 指定输出文件
 python3 bilibili_metadata.py --id "BV1xx411c7mD" --type video --output "metadata/bilibili_BV1xx411c7mD.meta.json"
 ```
-## 输出格式 Schema
+## Output
 ### 必需字段
 | 字段 | 类型 | 说明 |
 |------|------|------|
 | `success` | boolean | 是否提取成功 |
 | `platform` | string | 固定为 `bilibili` |
 | `content_id` | string | 内容 ID（BV号/cv号/uid） |
 | `content_type` | string | 内容类型（video/article/up） |
 | `title` | string | 标题（视频/专栏）或名称（UP主） |
 ### 可选字段
 | 字段 | 类型 | 说明 |
 |------|------|------|
 | `author` | string | 作者/UP主名称 |
 | `duration` | string | 时长（视频） |
 | `description` | string | 简介/描述 |
 | `stats` | object | 统计数据（播放、点赞等） |
 | `output_file` | string | 输出文件路径 |
 | `error` | string | 错误信息（失败时） |
 ## 成功示例
 ### 视频元数据
 ```json
 {
  "success": true,
-  "platform": "bilibili",
+  "data": {
-  "content_type": "video",
+    "platform": "bilibili",
-  "content_id": "BV1xx411c7mD",
+    "type": "video",
-  "title": "视频标题",
+    "id": "BV1xx411c7mD",
-  "author": "UP主名字",
+    "title": "视频标题",
-  "duration": "10:30",
+    "author": "UP主名字",
-  "description": "视频简介内容...",
+    "stats": {"view": 100000, "like": 5000}
-  "stats": {
+  }
    "view": 100000,
    "like": 5000,
    "coin": 2000,
    "favorite": 1000,
    "share": 500,
    "danmaku": 3000
  },
  "output_file": "metadata/bilibili_BV1xx411c7mD.meta.json"
 }
 ```
-### 专栏元数据
+## Installation
 ```json
 {
  "success": true,
  "platform": "bilibili",
  "content_type": "article",
  "content_id": "cv43780186",
  "title": "专栏标题",
  "author": "作者名字",
  "description": "专栏摘要...",
  "stats": {
    "view": 50000,
    "like": 2000,
    "comment": 500
  },
  "output_file": "metadata/bilibili_cv43780186.meta.json"
 }
 ```
 ### UP主信息
 ```json
 {
  "success": true,
  "platform": "bilibili",
  "content_type": "up",
  "content_id": "7770724",
  "title": "UP主名字",
  "description": "UP主签名...",
  "stats": {
    "follower": 1000000,
    "following": 100,
    "video_count": 500
  },
  "output_file": "metadata/bilibili_7770724.meta.json"
 }
 ```
 ## 错误示例
 ### 视频不存在
 ```json
 {
  "success": false,
  "platform": "bilibili",
  "content_id": "BV1invalid",
  "content_type": "video",
  "error": "视频不存在或已被删除"
 }
 ```
 ### 网络错误
 ```json
 {
  "success": false,
  "platform": "bilibili",
  "content_id": "BV1xx411c7mD",
  "content_type": "video",
  "error": "网络请求失败: Connection timeout"
 }
 ```
 ### 参数错误
 ```json
 {
  "success": false,
  "error": "缺少必需参数: --id"
 }
 ```
 ## 安装
 ```bash
-cp -r kb-bilibili-metadata/ ~/.openclaw/skills/
+cp -r kb-bilibili-metadata/ ~/skills/
 pip install httpx playwright
 playwright install chromium
 ```
 ## 在 kb-workflow 中的使用
 ```bash
 # 步骤 2：元数据提取（B站视频）
 python3 ~/.openclaw/skills/kb-bilibili-metadata/bilibili_metadata.py \
    --id "BV1xx411c7mD" --type video \
    --output "metadata/bilibili_BV1xx411c7mD.meta.json"
 # 检查输出：
 # - success=true → 进入步骤 3
 # - success=false → 检查错误，可能需要重试或退出
 ```
 ## 注意事项
 - 部分视频可能需要登录才能获取完整信息
 - 专栏内容可能需要额外的 Cookie 配置
 - 如果使用 Playwright 方式，首次运行需要下载浏览器
@@ -1,207 +1,102 @@
 ---
 name: kb-deep-analyzer
 description: AI深度分析 - 使用 AI 对内容进行深度分析和摘要
 ---
 # KB Deep Analyzer Skill
 **kb-deep-analyzer** - AI 深度内容分析工具。
-## 什么时候使用
+## Features
 - kb-workflow 步骤 5：对视频字幕或文章内容进行深度分析
 - 需要生成摘要、主题、见解等
 ## 功能
 - 多类型分析：视频（video）、文章（article）
 - 模板化输出
 - 质量评分
 - 自动标签
- 多格式元数据支持
+- **多格式元数据支持**: 支持扁平结构 `{"title":"xxx"}` 和嵌套结构 `{"data":{"metadata":{...}}}`
-## 环境变量（必需）
+## 内容类型
-| 变量名 | 说明 | 示例 |
+| 类型 | 参数 | 输出格式 |
-|--------|------|------|
+|------|------|----------|
-| `AI_API_KEY` | AI API 密钥 | `your-api-key` |
+| 视频 | `--type video` | 摘要 + 主题 + 见解 + 时间轴 |
-| `AI_PROVIDER` | AI 提供商 | `zhipu` / `openai` / `minimax` |
+| 文章 | `--type article` | 300字摘要 + 主题 + 原文 |
 | `AI_MODEL` | 模型名称 | `glm-5` / `gpt-4` |
 | `AI_BASE_URL` | API 端点 | `https://open.bigmodel.cn/api/coding/paas/v4` |
-## 命令
+## Commands
 ```bash
-# 分析视频
+# 分析视频（默认）
-AI_API_KEY="xxx" AI_PROVIDER="zhipu" AI_MODEL="glm-5" \
+AI_API_KEY="your-api-key" AI_PROVIDER="zhipu" AI_MODEL="glm-5" \
-python3 -u deep_analyzer.py \
+  python3 -u deep_analyzer.py --metadata meta.json --content content.json --output analysis.md --type video
    --metadata meta.json \
    --content content.json \
    --output analysis.md \
    --type video \
    --provider zhipu --model glm-5
-# 分析文章
+# 分析文章（300字摘要 + 原文）
-AI_API_KEY="xxx" AI_PROVIDER="zhipu" AI_MODEL="glm-5" \
+AI_API_KEY="your-api-key" AI_PROVIDER="zhipu" AI_MODEL="glm-5" \
-python3 -u deep_analyzer.py \
+  python3 -u deep_analyzer.py --metadata meta.json --content content.json --output analysis.md --type article
-    --metadata meta.json \
+
-    --content content.txt \
+# 指定 AI 提供商
-    --output analysis.md \
+python3 -u deep_analyzer.py -m meta.json -c content.json -o out.md --type video --provider zhipu --model glm-5
    --type article \
    --provider zhipu --model glm-5
 ```
-## 参数说明
+## 环境变量配置
-| 参数 | 说明 | 必需 |
+**必须设置以下环境变量**：
 |------|------|------|
 | `--metadata` | 元数据 JSON 文件路径 | 是 |
 | `--content` | 内容文件路径（JSON/TXT） | 是 |
 | `--output` | 输出 Markdown 文件路径 | 是 |
 | `--type` | 内容类型（video/article） | 是 |
 | `--provider` | AI 提供商 | 否（默认从环境变量） |
 | `--model` | 模型名称 | 否（默认从环境变量） |
 ## 输出格式 Schema
 ### 必需字段
 | 字段 | 类型 | 说明 |
 |------|------|------|
 | `success` | boolean | 是否分析成功 |
 ### 成功时字段
 | 字段 | 类型 | 说明 |
 |------|------|------|
 | `output_file` | string | 输出文件路径 |
 | `model` | string | 使用的模型 |
 | `provider` | string | 使用的提供商 |
 ### 失败时字段
 | 字段 | 类型 | 说明 |
 |------|------|------|
 | `error` | string | 错误信息 |
 | `message` | string | 详细说明 |
 ## 成功示例
 ```json
 {
  "success": true,
  "output_file": "analyzed/bilibili_BV1xx411c7mD.md",
  "model": "glm-5",
  "provider": "zhipu"
 }
 ```
 ## 错误示例
 ### API 调用失败
 ```json
 {
  "success": false,
  "error": "AI API 调用失败: rate limit exceeded",
  "message": "深度分析失败，整个流程终止"
 }
 ```
 ### 环境变量未设置
 ```json
 {
  "success": false,
  "error": "AI_API_KEY 环境变量未设置",
  "suggestion": "请配置 AI_API_KEY 环境变量"
 }
 ```
 ### 内容文件不存在
 ```json
 {
  "success": false,
  "error": "内容文件不存在",
  "content_file": "content/missing.json"
 }
 ```
 ## 输出格式（Markdown）
 ### 视频分析输出
 ```markdown
 # 视频标题
 ## 摘要
 [200-300字的核心内容摘要]
 ## 主要主题
 - 主题1
 - 主题2
 - 主题3
 ## 关键见解
 - 见解1
 - 见解2
 ## 时间轴
 - 00:00 - 02:30: 第一部分
 - 02:30 - 05:00: 第二部分
 ...
 ## 标签
 #标签1 #标签2 #标签3
 ```
 ### 文章分析输出
 ```markdown
 # 文章标题
 ## 摘要
 [300字以内的核心观点]
 ## 主要主题
 - 主题1
 - 主题2
 ## 完整原文
 [原始文章内容]
 ```
 ## 安装
 ```bash
-cp -r kb-deep-analyzer/ ~/.openclaw/skills/
+export AI_API_KEY="your-api-key"
-pip install httpx
+export AI_PROVIDER="zhipu"           # 或 openai, minimax
 export AI_MODEL="glm-5"           # 模型名称
 export AI_BASE_URL="https://..."    # 可选，自定义 API 地址
 ```
-## 在 kb-workflow 中的使用
+### 重要提示
 **必须使用 `-u` 参数运行 Python**，确保输出实时显示在终端：
 ```bash
-# 步骤 5：深度分析
+python3 -u deep_analyzer.py ...
 # ⚠️ 必须成功，否则整个流程失败
 AI_API_KEY="$AI_API_KEY" AI_PROVIDER="zhipu" AI_MODEL="glm-5" \
 python3 -u ~/.openclaw/skills/kb-deep-analyzer/deep_analyzer.py \
    --metadata "metadata/bilibili_BV1xx411c7mD.meta.json" \
    --content "content/bilibili_BV1xx411c7mD.json" \
    --output "analyzed/bilibili_BV1xx411c7mD.md" \
    --type video \
    --provider zhipu --model glm-5
 # 检查结果：
 # - success=true → 进入步骤 6（归档）
 # - success=false → 整个流程终止，报告错误
 ```
-## ⚠️ 重要提示
+如果不使用 `-u` 参数，Python 的 stdout 缓冲会导致输出延迟或看不到进度。
-1. **必须使用 `-u` 参数**：确保 Python 输出实时显示
+## 执行方式选择
-2. **深度分析不可跳过**：这是必须成功的步骤
+
-3. **失败即终止**：如果返回 `success: false`，整个流程必须停止
+### 方式1: 并发执行（推荐，默认）
-4. **禁止降级**：不能使用预设模板或简单摘要"假装成功"
+- 使用 `batch_size=3`（默认）
-5. **并发执行**：默认使用 batch_size=3 并发处理
+- 6个分片约 100 秒完成
 - 适合快速分析
 ```bash
 python3 -u deep_analyzer.py --metadata meta.json --content content.json -o out.md
 ```
 ### 方式2: 顺序执行
 - 使用 `batch_size=1`
 - 6个分片约 735 秒完成
 - 适合调试和问题排查
 ```bash
 # 在代码中设置 batch_size=1
 # analyzer.analyze(metadata, content, "video", batch_size=1)
 ```
 ## Metadata Format
 支持两种元数据格式：
 ```json
 // 格式1: 扁平结构
 {
  "title": "视频标题",
  "author": "作者名",
  "duration": 619,
  "bvid": "BVxxx"
 }
 // 格式2: 嵌套结构（kb-bilibili-metadata 输出）
 {
  "data": {
    "metadata": {
      "title": "视频标题",
      "author": "作者名",
      "duration": "10:19"
    }
  }
 }
 ```
 会自动转换为扁平格式进行分析。
@@ -1,41 +1,44 @@
 ---
 name: kb-funasr
-description: 本地语音识别 - 使用阿里 FunASR 进行中文优化的语音识别
+description: 本地语音识别，支持音视频转文字（基于阿里 FunASR）
 ---
 # KB FunASR Skill
-**kb-funasr** - 本地语音识别，使用阿里 FunASR 将音视频文件转换为文本。
+**kb-funasr** - 本地语音识别，使用阿里 FunASR 将音视频文件转换为文本和 JSON 格式输出。
 ## 什么时候使用
- kb-workflow 步骤 4 的备选方案：当 Whisper 效果不佳时
+当用户需要：
- 中文语音识别（Paraformer 模型对中文优化）
+- 将视频/音频文件转换为文字
- 需要高精度中文识别
+- 获取带时间戳的详细识别结果
 - 需要中文优化的语音识别（Paraformer 模型）
 ## 环境变量
-| 变量名 | 必需 | 说明 |
+在使用之前，必须检查环境变量是否已经设置。
 |--------|------|------|
 | `FUNASR_CACHE_DIR` | 否 | FunASR 模型缓存目录 |
-## 命令
+| 变量名 | 用途 | 示例 |
 |--------|------|------|
 | `FUNASR_CACHE_DIR` | FunASR 模型缓存目录 | `~/.cache/modelscope/hub` |
 ## 使用方法
 ```bash
-# 基本用法（使用 Paraformer 中文模型，推荐）
+# 基本用法（使用 Paraformer 模型）
-python3 funasr_tool.py --input video.mp4
+python funasr.py --input video.mp4
 # 指定模型
-python3 funasr_tool.py --input video.mp4 --model paraformer-zh
+python funasr.py --input video.mp4 --model paraformer-zh
 # 指定输出目录
-python3 funasr_tool.py --input video.mp4 --output ./output
+python funasr.py --input video.mp4 --output ./output
-# 后台异步执行
+# 异步执行（推荐）
-python3 funasr_tool.py --input video.mp4 --background
+python funasr.py --input video.mp4 --background
 ```
-## 参数说明
+## 命令参数
 | 参数 | 说明 | 默认值 |
 |------|------|--------|
@@ -46,91 +49,47 @@ python3 funasr_tool.py --input video.mp4 --background
 ## 模型选项
-| 模型 | 说明 | 推荐场景 |
+- `paraformer-zh` - Paraformer 中文模型（轻量，中文优化，推荐）
-|------|------|---------|
+- `paraformer` - Paraformer 多语言模型
-| `paraformer-zh` | Paraformer 中文模型 | **推荐**，中文优化 |
+- `sensevoice` - SenseVoice 多语言模型
 | `paraformer` | Paraformer 多语言模型 | 多语言场景 |
 | `sensevoice` | SenseVoice 模型 | 高精度需求 |
-## 输出格式 Schema
+## 输出文件
-### 必需字段
+对于输入 `video.mp4`，将生成：
-| 字段 | 类型 | 说明 |
+- `video.json` - 详细结果（包含时间戳、分段信息）
-|------|------|------|
+- `video.txt` - 纯文本内容
 | `success` | boolean | 是否识别成功 |
 | `input_file` | string | 输入文件路径 |
 | `model` | string | 使用的模型 |
-### 可选字段
+## 输出格式示例
-| 字段 | 类型 | 说明 |
+### JSON 格式
 |------|------|------|
 | `output_files` | object | 输出文件路径 |
 | `duration_seconds` | number | 音频时长 |
 | `segments` | array | 识别分段 |
 | `error` | string | 错误信息 |
 ## 成功示例
 ```json
 {
  "success": true,
-  "input_file": "content/downloads/bilibili_BV1xx411c7mD.mp4",
+  "input_file": "video.mp4",
  "model": "paraformer-zh",
  "output_files": {
-    "json": "content/bilibili_BV1xx411c7mD_funasr.json",
+    "json": "video.json",
-    "txt": "content/bilibili_BV1xx411c7mD_funasr.txt"
+    "txt": "video.txt"
  },
-  "duration_seconds": 630.5,
+  "duration_seconds": 120.5,
  "segments": [
    {
      "id": 0,
      "start": 0.0,
      "end": 3.2,
      "text": "这是第一段识别结果"
    },
    {
      "id": 1,
      "start": 3.2,
      "end": 6.5,
      "text": "这是第二段识别结果"
    }
  ]
 }
 ```
-## 错误示例
+## Workflow 中的使用
-### 模型未下载
+```bash
-
+# 步骤4之后：如果已有音频但无字幕，可以使用 FunASR
-```json
+python funasr.py --input ${video_path} --model paraformer-zh --background
 {
  "success": false,
  "error": "模型未找到",
  "model": "paraformer-zh",
  "suggestion": "首次运行时会自动下载模型，请确保网络连接正常"
 }
 ```
 ### 文件不存在
 ```json
 {
  "success": false,
  "error": "输入文件不存在",
  "input_file": "content/downloads/missing.mp4"
 }
 ```
 ### 内存不足
 ```json
 {
  "success": false,
  "error": "内存不足",
  "suggestion": "尝试使用较小的模型或处理较短的音频片段"
 }
 ```
 ## 安装
@@ -139,35 +98,16 @@ python3 funasr_tool.py --input video.mp4 --background
 # 1. 安装依赖
 pip install funasr modelscope
-# 2. 首次运行时自动下载模型
+# 2. 首次运行时会自动下载模型到缓存目录
-python3 funasr_tool.py --input test.mp4
+python funasr.py --input test.mp4
-# 3. 复制技能
+# 3. 复制技能到 skills 目录
-cp -r kb-funasr/ ~/.openclaw/skills/
+cp -r kb-funasr/ ~/skills/
 ```
 ## 在 kb-workflow 中的使用
 ```bash
 # 步骤 4 的备选方案（当 Whisper 效果不佳时）
 python3 ~/.openclaw/skills/kb-funasr/funasr_tool.py \
    --input "content/downloads/bilibili_BV1xx411c7mD.mp4" \
    --model paraformer-zh \
    --output "content/"
 # 输出：
 # - bilibili_BV1xx411c7mD_funasr.json (完整结果)
 # - bilibili_BV1xx411c7mD_funasr.txt (纯文本)
 # 完成后进入步骤 5（深度分析）
 ```
 ## 注意事项
- **首次运行**：首次运行时会自动下载模型文件（约 1GB），需要稳定网络
+- **异步执行**: 使用 `--background` 参数时，命令会在后台运行，不阻塞主代理
- **中文优化**：Paraformer 模型对中文识别效果优于 Whisper
+- **模型选择**: paraformer-zh 模型适合中文语音识别，速度较快，精度较高
- **耗时操作**：语音识别可能需要几分钟到几十分钟
+- **文件格式**: 支持 MP3、MP4、WAV、FLAC、OGG、M4A 等常见格式
- **文件格式**：支持 MP3、MP4、WAV、FLAC、OGG、M4A 等常见格式
+- **首次运行**: 首次运行时会自动下载模型文件，需要稳定的网络连接
 - **与 Whisper 对比**：
  - Whisper：通用性强，支持多语言
  - FunASR：中文优化，精度更高
@@ -1,190 +1,43 @@
 ---
 name: kb-link-parser
 description: 链接解析器 - 从各平台 URL 提取平台类型、内容类型和 ID
 ---
 # KB Link Parser Skill
-**kb-link-parser** - 从各种平台 URL 中提取和解析 ID。
+**kb-link-parser** - Extract and resolve IDs from various platform URLs.
-## 什么时候使用
+## Features
- kb-workflow 步骤 1：解析用户提供的链接
+- Multi-platform support: Bilibili, YouTube, Zhihu
- 需要判断链接属于哪个平台、什么类型的内容
+- Short link detection: b23.tv and other short links
 - Multiple content types: video, article, up主, playlist, collection
 - Batch processing from file
 - JSON output format
-## 功能
+## Commands
 - 多平台支持：Bilibili、YouTube、知乎、小红书
 - 短链接解析：b23.tv、xhslink.com 等
 - 多种内容类型：video、article、question、answer 等
 - 批量处理支持
 - JSON 输出格式
 ## 命令
 ```bash
-# 解析单个 URL
+# Parse single URL
-python3 link-parser.py --url "https://www.bilibili.com/video/BV1xx411c7mD"
+python link-parser.py --url "https://www.bilibili.com/video/BV1xx411c7mD"
-# 批量处理
+# Batch processing
-python3 link-parser.py --batch urls.txt --output results.json
+python link-parser.py --batch urls.txt --output results.json
-# 指定输出格式
+# Output format
-python3 link-parser.py --url "..." --format json
+python link-parser.py --url "..." --format json
 ```
-## 输出格式 Schema
+## Output
 ### 必需字段
 | 字段 | 类型 | 说明 |
 |------|------|------|
 | `success` | boolean | 是否解析成功 |
 | `platform` | string | 平台名称（成功时必需） |
 | `type` | string | 内容类型（成功时必需） |
 | `id` | string | 内容 ID（成功时必需） |
 | `error` | string | 错误信息（失败时必需） |
 ### 可选字段
 | 字段 | 类型 | 说明 |
 |------|------|------|
 | `filename` | string | 建议的文件名 |
 | `message` | string | 描述信息 |
 ## 成功示例
 ### B站视频
 ```json
 {
  "success": true,
-  "platform": "bilibili",
+  "data": {
-  "type": "video",
+    "platform": "bilibili",
-  "id": "BV1xx411c7mD",
+    "type": "video",
-  "filename": "bilibili_BV1xx411c7mD.link.json",
+    "id": "BV1xx411c7mD"
-  "message": "URL解析成功"
+  }
 }
 ```
-### B站专栏
+## Installation
 ```json
 {
  "success": true,
  "platform": "bilibili",
  "type": "article",
  "id": "cv43780186",
  "filename": "bilibili_cv43780186.link.json",
  "message": "URL解析成功"
 }
 ```
 ### 知乎问答
 ```json
 {
  "success": true,
  "platform": "zhihu",
  "type": "question",
  "id": "266309957",
  "filename": "zhihu_266309957.link.json",
  "message": "URL解析成功"
 }
 ```
 ### 知乎专栏文章
 ```json
 {
  "success": true,
  "platform": "zhihu",
  "type": "zhuanlan",
  "id": "1894054073625461813",
  "filename": "zhihu_1894054073625461813.link.json",
  "message": "URL解析成功"
 }
 ```
 ### 知乎视频
 ```json
 {
  "success": true,
  "platform": "zhihu",
  "type": "zvideo",
  "id": "123456789",
  "filename": "zhihu_123456789.link.json",
  "message": "URL解析成功"
 }
 ```
 ### 小红书笔记
 ```json
 {
  "success": true,
  "platform": "xiaohongshu",
  "type": "note",
  "id": "abc123",
  "filename": "xiaohongshu_abc123.link.json",
  "message": "URL解析成功"
 }
 ```
 ## 错误示例
 ### 不支持的平台
 ```json
 {
  "success": false,
  "error": "Unsupported URL format",
  "detail": "URL: https://example.com/video/123"
 }
 ```
 ### 无效 URL
 ```json
 {
  "success": false,
  "error": "解析失败",
  "detail": "URL: not-a-valid-url"
 }
 ```
 ## 支持的平台和类型
 | 平台 | 类型 | URL 示例 |
 |------|------|----------|
 | bilibili | video | `bilibili.com/video/BVxxx` |
 | bilibili | article | `bilibili.com/read/cvxxx` |
 | bilibili | up主 | `space.bilibili.com/12345` |
 | bilibili | favlist | `space.bilibili.com/123/favlist?fid=xxx` |
 | zhihu | question | `zhihu.com/question/123` |
 | zhihu | answer | `zhihu.com/question/123/answer/456` |
 | zhihu | zhuanlan | `zhuanlan.zhihu.com/p/123` |
 | zhihu | zvideo | `zhihu.com/zvideo/123` |
 | xiaohongshu | note | `xiaohongshu.com/explore/xxx` |
 | youtube | video | `youtube.com/watch?v=xxx` |
 ## 安装
 ```bash
-cp -r kb-link-parser/ ~/.openclaw/skills/
+cp -r kb-link-parser/ ~/skills/
 pip install requests
 ```
 ## 在 kb-workflow 中的使用
 ```bash
 # 步骤 1：链接解析
 python3 ~/.openclaw/skills/kb-link-parser/link-parser.py --url "[用户提供的URL]"
 # 解析输出后：
 # - 检查 success 字段
 # - 提取 platform, type, id
 # - 生成 pending/{platform}_{id}.link.json
 # - 进入步骤 2（元数据提取）
 ```
@@ -1,146 +1,66 @@
 ---
 name: kb-ocr-processor
 description: 视频OCR识别 - 从视频帧中提取文字
 ---
 # KB OCR Processor Skill
-**kb-ocr-processor** - 从视频帧中提取文字，作为语音识别的补充或替代方案。
+**kb-ocr-processor** - Extract text from video frames using OCR.
-## 什么时候使用
+## Features
- kb-workflow 步骤 4 的降级方案：当 Whisper 语音识别效果不佳时
+- **视频截帧**: 直接从视频文件截取帧图片
- 音频质量差、有背景音乐干扰、多种语言混合等情况
+- **批量处理**: 支持处理帧目录
 - **双引擎**: EasyOCR / Tesseract
 - **置信度评分**: 每帧识别结果带置信度
 - **文本合并**: 自动合并所有帧的识别结果
-## 功能
+## Commands
 - 视频截帧：直接从视频文件截取帧图片
 - 批量处理：支持处理帧目录
 - 双引擎：EasyOCR / Tesseract
 - 置信度评分：每帧识别结果带置信度
 - 文本合并：自动合并所有帧的识别结果
 ## 命令
 ```bash
-# 方式1: 直接从视频截帧并 OCR（推荐）
+# 方式1: 直接从视频截帧并OCR（推荐）
-python3 ocr_processor.py --video ./video.mp4 --output ocr.json
+python ocr_processor.py --video ./video.flv --output ocr.json
 # 方式2: 处理已截取的帧目录
-python3 ocr_processor.py --input ./frames/ --output ocr.json
+python ocr_processor.py --input ./frames/ --output ocr.json
 # 指定截帧帧率（默认1帧/秒）
-python3 ocr_processor.py --video ./video.mp4 --fps 2 --output ocr.json
+python ocr_processor.py --video ./video.flv --fps 2 --output ocr.json
 # 同时生成合并的文本文件
-python3 ocr_processor.py --video ./video.mp4 --output ocr.json --merge
+python ocr_processor.py --video ./video.flv --output ocr.json --merge
 # 使用 Tesseract 引擎
-python3 ocr_processor.py --input ./frames/ --engine tesseract --output ocr.json
+python ocr_processor.py --input ./frames/ --engine tesseract --output ocr.json
 ```
-## 输出格式 Schema
+## Workflow 中的使用
-### 必需字段
+```bash
 # 步骤5: OCR处理（视频无字幕时）
 python ocr_processor.py --video ${CONTENT_VIDEO_PATH} --output ${CONTENT_DIR}/ocr.json --merge
 ```
-| 字段 | 类型 | 说明 |
+## Output
 |------|------|------|
 | `success` | boolean | 是否识别成功 |
 | `engine` | string | 使用的 OCR 引擎 |
 | `total_frames` | number | 处理的总帧数 |
 ### 可选字段
 | 字段 | 类型 | 说明 |
 |------|------|------|
 | `results` | array | 每帧的识别结果 |
 | `merged_text` | string | 合并后的文本 |
 | `output_file` | string | 输出文件路径 |
 | `error` | string | 错误信息 |
 ## 成功示例
 ```json
 {
  "success": true,
  "engine": "easyocr",
  "total_frames": 619,
  "output_file": "content/bilibili_BV1xx411c7mD_ocr.json",
  "results": [
    {
      "source": "0001.jpg",
      "text": "字幕文字",
      "confidence": 0.95
    },
    {
      "source": "0002.jpg",
      "text": "更多字幕文字",
      "confidence": 0.92
    }
-  ],
+  ]
  "merged_text": "字幕文字\n更多字幕文字\n..."
 }
 ```
-## 错误示例
+## Installation
 ### 视频文件不存在
 ```json
 {
  "success": false,
  "error": "视频文件不存在",
  "video": "./missing.mp4"
 }
 ```
 ### ffmpeg 未安装
 ```json
 {
  "success": false,
  "error": "ffmpeg 未安装",
  "suggestion": "请运行: sudo apt-get install ffmpeg"
 }
 ```
 ### 内存不足
 ```json
 {
  "success": false,
  "error": "内存不足",
  "suggestion": "尝试降低帧率或处理较短的片段"
 }
 ```
 ## 安装
 ```bash
-cp -r kb-ocr-processor/ ~/.openclaw/skills/
+cp -r kb-ocr-processor/ ~/skills/
-pip install easyocr pytesseract Pillow
+pip install -r requirements.txt
-# 系统依赖
+# 依赖说明
-sudo apt-get install ffmpeg tesseract-ocr tesseract-ocr-chi-sim tesseract-ocr-eng
+# - easyocr: 主要OCR引擎（自动下载模型）
 # - pytesseract: Tesseract引擎
 # - ffmpeg: 视频截帧（apt install ffmpeg）
 ```
 ## 在 kb-workflow 中的使用
 ```bash
 # 步骤 4 的降级方案（当 Whisper 效果不佳时）
 python3 ~/.openclaw/skills/kb-ocr-processor/ocr_processor.py \
    --video "content/downloads/bilibili_BV1xx411c7mD.mp4" \
    --output "content/bilibili_BV1xx411c7mD_ocr.json" \
    --merge
 # 输出：
 # - bilibili_BV1xx411c7mD_ocr.json (完整结果)
 # - bilibili_BV1xx411c7mD_ocr.txt (合并文本)
 ```
 ## 注意事项
 - **耗时操作**：OCR 处理需要截帧和识别，可能较慢
 - **帧率选择**：1帧/秒适合大多数场景，更高帧率增加精度但耗时更长
 - **引擎选择**：EasyOCR 精度较高，Tesseract 速度较快
 - **适用场景**：音频质量差、背景音乐干扰、多语言混合时使用
@@ -1,40 +1,44 @@
 ---
 name: kb-whisper
-description: 本地语音识别 - 使用 Whisper.cpp 将音视频转文字
+description: 本地语音识别，支持音视频转文字（基于 Whisper.cpp）
 ---
 # KB Whisper Skill
-**kb-whisper** - 本地语音识别，使用 Whisper.cpp 将音视频文件转换为文本、字幕和 JSON 格式。
+**kb-whisper** - 本地语音识别，使用 Whisper.cpp 将音视频文件转换为文本、字幕和 JSON 格式输出。
 ## 什么时候使用
- kb-workflow 步骤 4：视频无官方字幕时进行语音识别
+当用户需要：
- 需要将音视频转为文字
+- 将视频/音频文件转换为文字
 - 生成视频字幕（SRT 格式）
 - 获取带时间戳的详细识别结果
 ## 环境变量
-| 变量名 | 必需 | 说明 |
+在使用之前，必须检查环境变量是否已经设置。
 |--------|------|------|
 | `WHISPER_CPP_PATH` | 是 | whisper.cpp 安装路径 |
-## 命令
+| 变量名 | 用途 | 示例 |
 |--------|------|------|
 | `WHISPER_CPP_PATH` | whisper.cpp 安装路径 | `~/whisper.cpp` |
 ## 使用方法
 ```bash
 # 基本用法
-python3 whisper.py --input video.mp4
+python whisper.py --input video.mp4
-# 指定模型（推荐 base）
+# 指定模型（默认 tiny）
-python3 whisper.py --input video.mp4 --model base
+python whisper.py --input video.mp4 --model base
 # 指定输出目录
-python3 whisper.py --input video.mp4 --output ./output
+python whisper.py --input video.mp4 --output ./output
-# 后台异步执行
+# 异步执行（推荐）
-python3 whisper.py --input video.mp4 --background
+python whisper.py --input video.mp4 --background
 ```
-## 参数说明
+## 命令参数
 | 参数 | 说明 | 默认值 |
 |------|------|--------|
@@ -45,94 +49,51 @@ python3 whisper.py --input video.mp4 --background
 ## 模型选项
-| 模型 | 速度 | 精度 | 推荐场景 |
+- `tiny` - 最小模型，速度快，精度较低（推荐快速识别）
-|------|------|------|---------|
+- `base` - 基础模型，平衡速度和精度（推荐）
-| `tiny` | 最快 | 较低 | 快速预览 |
+- `small` - 小型模型，精度更高，速度较慢
-| `base` | 快 | 良好 | **推荐** |
+- `medium` - 中型模型，高精度
-| `small` | 中等 | 高 | 精确识别 |
+- `large` - 大型模型，最高精度，速度慢
 | `medium` | 慢 | 很高 | 专业场景 |
 | `large` | 最慢 | 最高 | 最高精度 |
-## 输出格式 Schema
+## 输出文件
-### 必需字段
+对于输入 `video.mp4`，将生成：
-| 字段 | 类型 | 说明 |
+- `video.json` - 详细结果（包含时间戳、分段信息）
-|------|------|------|
+- `video.srt` - 字幕格式文件
-| `success` | boolean | 是否识别成功 |
+- `video.txt` - 纯文本内容
 | `input_file` | string | 输入文件路径 |
 | `model` | string | 使用的模型 |
-### 可选字段
+## 输出格式示例
-| 字段 | 类型 | 说明 |
+### JSON 格式
 |------|------|------|
 | `output_files` | object | 输出文件路径 |
 | `duration_seconds` | number | 音频时长 |
 | `segments` | array | 识别分段 |
 | `error` | string | 错误信息 |
 ## 成功示例
 ```json
 {
  "success": true,
-  "input_file": "content/downloads/bilibili_BV1xx411c7mD.mp4",
+  "input_file": "video.mp4",
-  "model": "base",
+  "model": "tiny",
  "output_files": {
-    "json": "content/bilibili_BV1xx411c7mD.json",
+    "json": "video.json",
-    "txt": "content/bilibili_BV1xx411c7mD.txt",
+    "srt": "video.srt",
-    "srt": "content/bilibili_BV1xx411c7mD.srt"
+    "txt": "video.txt"
  },
-  "duration_seconds": 630.5,
+  "duration_seconds": 120.5,
  "segments": [
    {
      "id": 0,
      "start": 0.0,
      "end": 3.2,
      "text": "这是第一段字幕"
    },
    {
      "id": 1,
      "start": 3.2,
      "end": 6.5,
      "text": "这是第二段字幕"
    }
  ]
 }
 ```
-## 错误示例
+## Workflow 中的使用
-### 模型未下载
+```bash
-
+# 步骤4之后：如果已有音频但无字幕，可以使用 whisper
-```json
+python whisper.py --input ${video_path} --model base --background
 {
  "success": false,
  "error": "模型文件不存在",
  "model": "base",
  "suggestion": "请运行: cd ~/whisper.cpp && ./models/download-ggml-model.sh base"
 }
 ```
 ### 文件不存在
 ```json
 {
  "success": false,
  "error": "输入文件不存在",
  "input_file": "content/downloads/missing.mp4"
 }
 ```
 ### 内存不足
 ```json
 {
  "success": false,
  "error": "内存不足，无法加载模型",
  "suggestion": "尝试使用更小的模型（tiny/base）"
 }
 ```
 ## 安装
@@ -142,39 +103,22 @@ python3 whisper.py --input video.mp4 --background
 git clone https://github.com/ggerganov/whisper.cpp.git ~/whisper.cpp
 # 2. 编译
-cd ~/whisper.cpp && make
+cd ~/whisper.cpp
 make
-# 3. 下载模型
+# 3. 下载模型（tiny）
-./models/download-ggml-model.sh base
+./models/download-ggml-model.sh tiny
-# 4. 设置环境变量
+# 4. 下载更多模型（可选）
-echo 'export WHISPER_CPP_PATH="$HOME/whisper.cpp"' >> ~/.kb_env
+# ./models/download-ggml-model.sh base
-source ~/.kb_env
+# ./models/download-ggml-model.sh small
-# 5. 复制技能
+# 5. 复制技能到 skills 目录
-cp -r kb-whisper/ ~/.openclaw/skills/
+cp -r kb-whisper/ ~/skills/
 ```
 ## 在 kb-workflow 中的使用
 ```bash
 # 步骤 4：语音识别（当 has_subtitle=false 时）
 python3 ~/.openclaw/skills/kb-whisper/whisper.py \
    --input "content/downloads/bilibili_BV1xx411c7mD.mp4" \
    --model base \
    --output "content/"
 # 输出：
 # - bilibili_BV1xx411c7mD.json (完整结果)
 # - bilibili_BV1xx411c7mD.txt (纯文本)
 # - bilibili_BV1xx411c7mD.srt (字幕文件)
 # 完成后进入步骤 5（深度分析）
 ```
 ## 注意事项
- **耗时操作**：语音识别可能需要几分钟到几十分钟，取决于视频长度
+- **异步执行**: 使用 `--background` 参数时，命令会在后台运行，不阻塞主代理
- **异步执行**：推荐使用 `--background` 参数
+- **模型选择**: tiny 模型适合快速识别，base 模型适合一般用途，large 模型适合高精度需求
- **模型选择**：base 模型平衡速度和精度，适合大多数场景
+- **文件格式**: 支持 MP3、MP4、WAV、FLAC、OGG、M4A 等常见格式
 - **文件格式**：支持 MP3、MP4、WAV、FLAC、OGG、M4A 等常见格式
@@ -5,698 +5,200 @@ description: 当收到"添加到知识库 + 链接" 或"保存到知识库 + 链
 # KB Workflow
-## 触发条件
+## 什么时候使用
 当用户发送包含以下模式的消息时，此技能自动触发：
 | 触发词 | 行为 | 示例 |
 |--------|------|------|
 | `添加到知识库` + 链接 | 添加到知识库并处理 | "添加到知识库 https://www.bilibili.com/video/BVxxx" |
 | `保存到知识库` + 链接 | 同上 | "保存到知识库 https://www.bilibili.com/video/BVxxx" |
---
+## 环境变量
-## 状态机定义
+### 执行前必须检查
-工作流是一个严格的状态机，每个状态有明确的前置条件和后置条件：
+在执行 workflow 之前，必须检查环境变量是否已经设置。
-| 状态 | 前置条件 | 执行动作 | 成功后进入 | 失败处理 |
+**必须的环境变量**（缺失则报错退出）：
 |------|---------|---------|-----------|---------|
 | `init` | 用户触发 + URL 有效 | 创建 state.json，检查环境变量 | `link_parse` | 报错退出 |
 | `link_parse` | state.json 存在 | 解析链接，提取平台和ID | `metadata` | 记录错误，最多重试1次 |
 | `metadata` | platform, content_id 已知 | 提取元数据 | `content` | 记录错误，最多重试1次 |
 | `content` | 元数据文件存在 | 下载内容/字幕 | `analyze` 或 `whisper` | 根据错误类型处理 |
 | `whisper` | 视频已下载，无官方字幕 | 语音识别转文字 | `analyze` | 尝试 OCR 降级 |
 | `analyze` | 有文本内容（字幕/识别结果/文章） | AI 深度分析 | `archive` | **流程失败，报错退出** |
 | `archive` | 分析报告存在 | 归档 + Git 提交 | `done` | 记录错误，可手动处理 |
 | `done` | - | 汇报完成 | - | - |
 ### 状态判断流程图
 ```
 用户发送链接
    ↓
 ┌─────────────────────────────────────┐
 │ Step 0: 环境变量检查                 │
 │ 缺少必需变量 → 报错退出              │
 └─────────────────────────────────────┘
    ↓
 ┌─────────────────────────────────────┐
 │ Step 1: 链接解析 (link_parse)        │
 │ 输出: platform, content_type, id     │
 └─────────────────────────────────────┘
    ↓
 ┌─────────────────────────────────────┐
 │ Step 2: 元数据提取 (metadata)        │
 │ 输出: title, author, etc.            │
 └─────────────────────────────────────┘
    ↓
 ┌─────────────────────────────────────┐
 │ content_type 判断                    │
 ├─────────────────────────────────────┤
 │ video/zvideo → Step 3a (视频处理)   │
 │ article/zhuanlan/question → Step 3b │
 └─────────────────────────────────────┘
    ↓
 ┌─────────────────────────────────────┐
 │ Step 3a: 视频处理                    │
 │ 有官方字幕? → 是 → Step 5           │
 │            → 否 → Step 4 (Whisper)  │
 └─────────────────────────────────────┘
    ↓
 ┌─────────────────────────────────────┐
 │ Step 3b: 文章处理                    │
 │ 提取文字 → 直接进入 Step 5          │
 └─────────────────────────────────────┘
    ↓
 ┌─────────────────────────────────────┐
 │ Step 4: Whisper 语音识别 (可选)      │
 │ 无字幕时执行，生成 text/json/srt     │
 └─────────────────────────────────────┘
    ↓
 ┌─────────────────────────────────────┐
 │ Step 5: 深度分析 (analyze)           │
 │ ⚠️ 必须成功，否则流程失败            │
 └─────────────────────────────────────┘
    ↓
 ┌─────────────────────────────────────┐
 │ Step 6: 归档 (archive)               │
 │ 复制到归档目录 + Git 提交            │
 └─────────────────────────────────────┘
    ↓
   完成
 ```
 ---
 ## 环境变量清单
 ### 必需变量（缺失则报错退出）
 | 变量名 | 用途 | 示例 |
 |--------|------|------|
 | `AI_API_KEY` | AI API 密钥 | `your-api-key` |
 | `AI_PROVIDER` | AI 提供商 | `zhipu` / `openai` / `minimax` |
 | `AI_MODEL` | 模型名称 | `glm-5` / `gpt-4` |
 | `AI_BASE_URL` | API 端点 | `https://open.bigmodel.cn/api/coding/paas/v4` |
 ### 可选变量
 | 变量名 | 用途 | 用途说明 |
 |--------|------|---------|
 | `BILIBILI_COOKIE` | B站登录态 | 下载会员视频 |
 | `ZHIHU_COOKIE` | 知乎登录态 | 访问专栏等 |
 | `HF_TOKEN` | HuggingFace Token | 说话者分离功能 |
 | `KB_ARCHIVE_PATH` | 归档目录 | 默认 `~/work/zq/30.资源/AI分析` |
 | `KB_GIT_ROOT` | Git 仓库根目录 | 默认 `~/work/zq` |
 | `KB_GIT_USER` | Git 提交用户名 | 归档时使用 |
 | `KB_GIT_EMAIL` | Git 提交邮箱 | 归档时使用 |
 ### 配置方法
 配置方式：将变量写入 `~/.kb_env` 文件：
 ```bash
 # 创建专用配置文件
 cat > ~/.kb_env << 'EOF'
 export AI_API_KEY="your-api-key"
 export AI_PROVIDER="zhipu"
 export AI_MODEL="glm-5"
 export AI_BASE_URL="https://open.bigmodel.cn/api/coding/paas/v4"
 export BILIBILI_COOKIE="SESSDATA=xxx; bili_jct=xxx"
 export KB_ARCHIVE_PATH="$HOME/work/zq/30.资源/AI分析"
 export KB_GIT_ROOT="$HOME/work/zq"
 EOF
 # 在 ~/.bashrc 中添加
 [ -f ~/.kb_env ] && source ~/.kb_env
 ```
---
+**可选的环境变量**：
 | 变量名 | 用途 | 示例 |
 |--------|------|------|
 | `BILIBILI_COOKIE` | 下载会员视频 | `SESSDATA=xxx; bili_jct=xxx` |
 | `KB_GIT_USER` | Git提交用户名 | `Your Name` |
 | `KB_GIT_EMAIL` | Git提交邮箱 | `your@email.com` |
 | `KB_ARCHIVE_PATH` | 归档目录路径 | `~/work/zq/30.资源/AI分析` |
 | `KB_GIT_ROOT` | Git仓库根目录 | `~/work/zq` |
 | `HF_TOKEN` | HuggingFace Token（说话者分离） | `hf_xxx` |
-## 步骤详解
+### 检查命令
 ### Step 0: 环境变量检查
 **前置条件：** 无
 **执行命令：**
 ```bash
-# 加载环境变量
+# 从 ~/.kb_env 加载环境变量
-if [ -f ~/.kb_env ]; then source ~/.kb_env; fi
+if [ -f ~/.kb_env ]; then
    source ~/.kb_env
 fi
 # 检查必须的环境变量
 if [ -z "$AI_API_KEY" ] || [ -z "$AI_PROVIDER" ] || [ -z "$AI_MODEL" ] || [ -z "$AI_BASE_URL" ]; then
    echo "❌ 错误：必须的环境变量未设置"
    echo "请配置 ~/.kb_env 文件"
    exit 1
 fi
 # 检查必需变量
 for var in AI_API_KEY AI_PROVIDER AI_MODEL AI_BASE_URL; do
    if [ -z "${!var}" ]; then
        echo "❌ 错误：$var 未设置"
        exit 1
    fi
 done
 echo "✅ 环境变量检查通过"
 ```
-**成功标准：**
+## Workflow 执行方式
 - 所有必需变量已设置
 - 输出 "✅ 环境变量检查通过"
-**失败处理：**
+### 方式一：一键处理（推荐）
 - 缺少任何必需变量 → 报错退出，提示用户配置
-**输出示例：**
+使用管道 CLI 一键完成所有步骤：
 ```
 ✅ 环境变量检查通过
   AI_PROVIDER: zhipu
   AI_MODEL: glm-5
 ```
 **错误示例：**
 ```
 ❌ 错误：AI_API_KEY 未设置
 请配置以下环境变量：
  - AI_API_KEY
  - AI_PROVIDER
  - AI_MODEL
  - AI_BASE_URL
 ```
 ---
 ### Step 1: 链接解析
 **前置条件：**
 - 环境变量检查通过
 - URL 格式有效
 **执行命令：**
 ```bash
-# 获取目录路径
+python3 kb-workflow/kb_workflow.py process "https://www.bilibili.com/video/BVxxx"
 python3 ~/.openclaw/skills/kb-workflow/kb_workflow.py --pending
 # 解析链接
 python3 ~/.openclaw/skills/kb-link-parser/link-parser.py --url "[URL]"
 ```
-**成功标准：**
+**该命令自动执行：**
- 返回 JSON 包含 `success: true`
+1. URL 解析 → 识别平台和内容 ID
- 包含字段：`platform`, `content_type`, `content_id`
+2. 元数据提取 → 获取标题、作者等信息
- 生成文件：`pending/{platform}_{id}.link.json`
+3. 内容下载 → 字幕/视频/文章内容
 4. 语音识别 → Whisper/FunASR（如需要）
 5. 深度分析 → AI 生成分析报告
 6. 归档 → 移动到归档目录
-**失败处理：**
+**命令参数：**
 - 不支持的平台 → 报错退出，告知用户支持的平台
 - URL 无效 → 报错退出，请用户提供正确链接
 **输出示例：**
 ```json
 {
  "success": true,
  "platform": "bilibili",
  "content_type": "video",
  "content_id": "BV1xx411c7mD",
  "output_file": "pending/bilibili_BV1xx411c7mD.link.json"
 }
 ```
 **错误示例：**
 ```json
 {
  "success": false,
  "error": "不支持的平台: example.com",
  "supported_platforms": ["bilibili.com", "zhihu.com", "youtube.com"]
 }
 ```
 ---
 ### Step 2: 元数据提取
 **前置条件：**
 - Step 1 完成
 - `platform`, `content_type`, `content_id` 已知
 **执行命令：**
 ```bash
-# 获取目录路径
+# 指定输出目录
-python3 ~/.openclaw/skills/kb-workflow/kb_workflow.py --metadata
+python3 kb-workflow/kb_workflow.py process "https://..." -o /path/to/output
-# 根据平台选择命令
+# 强制重新执行（忽略已完成状态）
-# B站视频
+python3 kb-workflow/kb_workflow.py process "https://..." --no-skip
 python3 ~/.openclaw/skills/kb-bilibili-metadata/bilibili_metadata.py \
    --id "[BV号]" --type video \
    --output "metadata/bilibili_[BV号].meta.json"
 # B站专栏
 python3 ~/.openclaw/skills/kb-bilibili-metadata/bilibili_metadata.py \
    --id "[cv号]" --type article \
    --output "metadata/bilibili_cv[cv号].meta.json"
 # 知乎问答
 python3 ~/.openclaw/skills/kb-zhihu-metadata/zhihu_metadata.py \
    --id "[问题ID]" --type question \
    --output "metadata/zhihu_[ID].meta.json"
 # 知乎视频
 python3 ~/.openclaw/skills/kb-zhihu-metadata/zhihu_metadata.py \
    --id "[视频ID]" --type zvideo \
    --output "metadata/zhihu_[ID].meta.json"
 ```
-**成功标准：**
+### 方式二：手动分步执行（高级）
 - 返回 JSON 包含 `success: true`
 - 包含字段：`platform`, `content_id`, `title`
 - 生成文件：`metadata/{platform}_{id}.meta.json`
-**失败处理：**
+如需更细粒度控制，可手动调用各子技能：
 - 网络错误 → 重试1次
 - 内容不存在 → 报错退出
 - 需要登录 → 提示配置 Cookie
 **输出示例：**
 ```json
 {
  "success": true,
  "platform": "bilibili",
  "content_id": "BV1xx411c7mD",
  "content_type": "video",
  "title": "视频标题",
  "author": "UP主名字",
  "duration": "10:30",
  "output_file": "metadata/bilibili_BV1xx411c7mD.meta.json"
 }
 ```
 **错误示例：**
 ```json
 {
  "success": false,
  "error": "视频不存在或已被删除",
  "content_id": "BV1xx411c7mD"
 }
 ```
 ---
 ### Step 3a: 视频内容下载
 **前置条件：**
 - Step 2 完成
 - `content_type` 为 `video` 或 `zvideo`
 **执行命令：**
 ```bash
-# 获取目录路径
+# 1. 查重
-python3 ~/.openclaw/skills/kb-workflow/kb_workflow.py --content_downloads
+python3 kb-workflow/kb_workflow.py --check --url "https://..."
-# B站视频 - 优先尝试字幕
+# 2. 一键处理
-python3 ~/.openclaw/skills/kb-bilibili-content/bilibili_content.py \
+python3 kb-workflow/kb_workflow.py process "https://..."
    --id "[BV号]" --type subtitle-download \
    --output "content/downloads/"
-# 检查是否有官方字幕
+# 3. 归档
-# 如果 has_subtitle=true → 进入 Step 5
+python3 kb-workflow/kb_workflow.py --archive-file "/path/to/report.md"
 # 如果 has_subtitle=false → 下载视频，进入 Step 4
 ```
-**成功标准：**
+## 支持的平台和内容类型
 - 返回 JSON 包含 `success: true`
 - `has_subtitle=true`: 有字幕文件
 - `has_subtitle=false`: 视频文件下载成功
-**失败处理：**
+| 平台 | 内容类型 | URL 示例 |
- 403 禁止访问 → 检查 BILIBILI_COOKIE 配置
+|------|----------|----------|
- 网络超时 → 重试1次
+| B站 | 视频 | `bilibili.com/video/BVxxx` |
- 会员视频 → 提示需要配置 Cookie
+| B站 | 专栏 | `bilibili.com/read/cvxxx` |
 | 知乎 | 问答 | `zhihu.com/question/xxx` |
 | 知乎 | 专栏 | `zhuanlan.zhihu.com/p/xxx` |
 | 知乎 | 视频 | `zhihu.com/zvideo/xxx` |
 ## 处理流程
 **输出示例（有字幕）：**
 ```json
 {
  "success": true,
  "type": "subtitle",
  "has_subtitle": true,
  "subtitle_file": "content/downloads/bilibili_BV1xx411c7mD.srt",
  "message": "官方字幕下载成功"
 }
 ```
-
+┌─────────────────────────────────────────────────────────────┐
-**输出示例（无字幕）：**
+│                    kb-workflow process                       │
-```json
+├─────────────────────────────────────────────────────────────┤
-{
+│  1. Parse     → LinkParser 解析 URL                         │
-  "success": true,
+│  2. Download  → 平台适配器获取元数据和内容                    │
-  "type": "subtitle",
+│  3. ASR       → 语音识别（视频无字幕时）                      │
-  "has_subtitle": false,
+│  4. Analyze   → DeepAnalyzer 生成分析报告                    │
-  "message": "无官方字幕，需要下载视频进行语音识别"
+│  5. Archive   → 归档到 KB_ARCHIVE_PATH                      │
-}
+└─────────────────────────────────────────────────────────────┘
 ```
 **错误示例：**
 ```json
 {
  "success": false,
  "error": "403 Forbidden - 可能需要会员权限",
  "suggestion": "请配置 BILIBILI_COOKIE 环境变量"
 }
 ```
 ---
 ### Step 3b: 文章内容提取
 **前置条件：**
 - Step 2 完成
 - `content_type` 为 `article`、`zhuanlan` 或 `question`
 **执行命令：**
 ```bash
 # B站专栏
 python3 ~/.openclaw/skills/kb-bilibili-content/bilibili_content.py \
    --id "[cv号]" --type article \
    --output "content/"
 # 知乎专栏
 python3 ~/.openclaw/skills/kb-zhihu-content/zhihu_content.py \
    --type article-from-media \
    --output "content/"
 ```
 **成功标准：**
 - 返回 JSON 包含 `success: true`
 - 包含字段：`filepath`, `content_length`
 - 跳过 Step 4，直接进入 Step 5
 **输出示例：**
 ```json
 {
  "success": true,
  "type": "article",
  "filepath": "content/bilibili_cv43780186.txt",
  "content_length": 5234,
  "message": "文章内容提取成功"
 }
 ```
 ---
 ### Step 4: Whisper 语音识别
 **前置条件：**
 - Step 3a 完成
 - `has_subtitle=false`
 - 视频文件已下载
 **执行命令：**
 ```bash
 # 获取目录路径
 python3 ~/.openclaw/skills/kb-workflow/kb_workflow.py --content
 # 普通识别
 python3 ~/.openclaw/skills/kb-whisper/whisper.py \
    --input "content/downloads/bilibili_BV1xx411c7mD.mp4" \
    --model base \
    --output "content/"
 # 带说话者分离（需要 HF_TOKEN）
 if [ -n "$HF_TOKEN" ]; then
    python3 ~/.openclaw/skills/kb-whisper-diarization/whisper_diarization.py \
        --input "content/downloads/bilibili_BV1xx411c7mD.mp4" \
        --model base --diarize \
        --output "content/"
 fi
 ```
 **成功标准：**
 - 返回 JSON 包含 `success: true`
 - 生成文件：`.json`, `.txt`, `.srt`
 **失败处理：**
 - 模型未下载 → 提示下载模型
 - 内存不足 → 尝试更小的模型
 - Whisper 失败 → 尝试 OCR 降级
 **输出示例：**
 ```json
 {
  "success": true,
  "input_file": "content/downloads/bilibili_BV1xx411c7mD.mp4",
  "model": "base",
  "output_files": {
    "json": "content/bilibili_BV1xx411c7mD.json",
    "txt": "content/bilibili_BV1xx411c7mD.txt",
    "srt": "content/bilibili_BV1xx411c7mD.srt"
  },
  "duration_seconds": 630
 }
 ```
 ---
 ### Step 5: 深度分析
 **前置条件：**
 - 有文本内容（字幕/识别结果/文章）
 - 元数据文件存在
 **执行命令：**
 ```bash
 # 获取目录路径
 python3 ~/.openclaw/skills/kb-workflow/kb_workflow.py --analyzed
 # 分析视频
 AI_API_KEY="$AI_API_KEY" AI_PROVIDER="zhipu" AI_MODEL="glm-5" \
 python3 -u ~/.openclaw/skills/kb-deep-analyzer/deep_analyzer.py \
    --metadata "metadata/bilibili_BV1xx411c7mD.meta.json" \
    --content "content/bilibili_BV1xx411c7mD.json" \
    --output "analyzed/bilibili_BV1xx411c7mD.md" \
    --type video \
    --provider zhipu --model glm-5
 # 分析文章
 AI_API_KEY="$AI_API_KEY" AI_PROVIDER="zhipu" AI_MODEL="glm-5" \
 python3 -u ~/.openclaw/skills/kb-deep-analyzer/deep_analyzer.py \
    --metadata "metadata/xxx.meta.json" \
    --content "content/xxx.txt" \
    --output "analyzed/xxx.md" \
    --type article \
    --provider zhipu --model glm-5
 ```
 **成功标准：**
 - 返回 JSON 包含 `success: true`
 - 生成分析报告文件
 **失败处理：**
 - ⚠️ **深度分析失败 = 整个流程失败**
 - API 调用失败 → 记录错误，**终止流程**
 - **禁止**使用预设模板或降级内容
 **输出示例：**
 ```json
 {
  "success": true,
  "output_file": "analyzed/bilibili_BV1xx411c7mD.md",
  "model": "glm-5",
  "provider": "zhipu"
 }
 ```
 **错误示例：**
 ```json
 {
  "success": false,
  "error": "AI API 调用失败: rate limit exceeded",
  "message": "深度分析失败，整个流程终止"
 }
 ```
 ---
 ### Step 6: 归档
 **前置条件：**
 - Step 5 完成
 - 分析报告文件存在
 **执行命令：**
 ```bash
 # 归档
 python3 ~/.openclaw/skills/kb-workflow/kb_workflow.py \
    --archive-file "analyzed/bilibili_BV1xx411c7mD.md"
 # 如果配置了 Git，会自动提交
 ```
 **成功标准：**
 - 文件复制到归档目录
 - Git 提交成功（如果配置）
 **失败处理：**
 - 归档失败 → 记录错误，文件仍可用
 - Git 失败 → 不影响主流程
 **输出示例：**
 ```json
 {
  "success": true,
  "archive_path": "~/work/zq/30.资源/AI分析/2026-03-13/bilibili_BV1xx411c7mD.md",
  "git_commit": "abc1234"
 }
 ```
 ---
 ## 目录结构
 **工作目录（默认 ~/work/knowledge-base/）：**
 ```
 ~/work/knowledge-base/
 ├── pending/           # 待处理链接
 ├── metadata/          # 元数据 JSON
 ├── content/           # 内容文件
 │   └── downloads/     # 下载的视频/字幕
 └── analyzed/          # 分析报告
 ```
 ~/work/knowledge-base/           # 知识库根目录
 ├── pending/                     # 待处理链接
 │   └── {platform}_{id}.link.json
 ├── metadata/                    # 元数据
 │   └── {platform}_{id}.meta.json
 ├── content/                     # 内容
 │   ├── downloads/               # 视频下载
 │   │   └── {platform}_{id}.mp4
 │   ├── {platform}_{id}.txt      # 文本内容
 │   └── {platform}_{id}.json     # Whisper 结果
 └── analyzed/                    # 深度分析报告
    └── {platform}_{id}.md
-~/work/zq/30.资源/AI分析/        # 归档目录
+**归档目录（默认 ~/work/zq/30.资源/AI分析/）：**
 ```
 ~/work/zq/30.资源/AI分析/
 └── YYYY-MM-DD/
-    └── {platform}_{id}.md
+    └── {platform}_{id}.md  # 深度分析报告
 ```
 ---
 ## 子技能清单
 | 步骤 | 子技能 | 命令 |
 |------|--------|------|
 | 1 | kb-link-parser | `--url "[URL]"` |
 | 2 | kb-bilibili-metadata | `--id "[ID]" --type video --output "[PATH]"` |
 | 2 | kb-zhihu-metadata | `--id "[ID]" --type question --output "[PATH]"` |
 | 3a | kb-bilibili-content | `--id "[ID]" --type subtitle-download --output "[PATH]"` |
 | 3a | kb-bilibili-content | `--id "[ID]" --type video --output "[PATH]"` |
 | 3b | kb-bilibili-content | `--id "[ID]" --type article --output "[PATH]"` |
 | 3b | kb-zhihu-content | `--type article-from-media --output "[PATH]"` |
 | 4 | kb-whisper | `--input "[VIDEO]" --model base --output "[PATH]"` |
 | 4 | kb-whisper-diarization | `--input "[VIDEO]" --model base --diarize --output "[PATH]"` |
 | 5 | kb-deep-analyzer | `--metadata "[META]" --content "[CONTENT]" --output "[OUT]" --type video` |
 ---
 ## 端到端示例
 ### 示例 1: B站视频（有官方字幕）
 **输入：** `添加到知识库 https://www.bilibili.com/video/BV1xx411c7mD`
 ```
 Step 0: ✅ 环境变量检查通过
 Step 1: ✅ 链接解析 → bilibili, video, BV1xx411c7mD
 Step 2: ✅ 元数据提取 → 标题: "xxx", 作者: "xxx"
 Step 3a: ✅ 字幕下载 → has_subtitle=true, bilibili_BV1xx411c7mD.srt
 Step 4: ⏭️ 跳过（有官方字幕）
 Step 5: ✅ 深度分析 → bilibili_BV1xx411c7mD.md
 Step 6: ✅ 归档 → ~/work/zq/30.资源/AI分析/2026-03-13/
 ```
 ### 示例 2: B站视频（无官方字幕）
 **输入：** `添加到知识库 https://www.bilibili.com/video/BV1yy411c7mE`
 ```
 Step 0: ✅ 环境变量检查通过
 Step 1: ✅ 链接解析 → bilibili, video, BV1yy411c7mE
 Step 2: ✅ 元数据提取 → 标题: "xxx", 时长: "15:30"
 Step 3a: ⚠️ 无官方字幕 → 下载视频
 Step 3a: ✅ 视频下载 → bilibili_BV1yy411c7mE.mp4
 Step 4: ✅ Whisper 识别 → bilibili_BV1yy411c7mE.json/txt/srt
 Step 5: ✅ 深度分析 → bilibili_BV1yy411c7mE.md
 Step 6: ✅ 归档
 ```
 ### 示例 3: 403 下载失败
 **输入：** `添加到知识库 https://www.bilibili.com/video/BV1zz411c7mF`
 ```
 Step 0: ✅ 环境变量检查通过
 Step 1: ✅ 链接解析
 Step 2: ✅ 元数据提取
 Step 3a: ❌ 403 Forbidden
         → 可能需要会员权限
         → 请配置 BILIBILI_COOKIE 环境变量
         → 流程暂停，等待用户处理
 ```
 ---
 ## 常见错误处理
 ### 错误 1: 环境变量未设置
 **症状：** `❌ 错误：AI_API_KEY 未设置`
 **解决：**
 ```bash
 echo 'export AI_API_KEY="your-key"' >> ~/.kb_env
 source ~/.kb_env
 ```
 ### 错误 2: 403 下载失败
 **症状：** B站视频下载返回 403
 **解决：**
 1. 获取 B站 Cookie（F12 → Network → Copy Cookie）
 2. 配置环境变量：
 ```bash
 echo 'export BILIBILI_COOKIE="SESSDATA=xxx; bili_jct=xxx"' >> ~/.kb_env
 source ~/.kb_env
 ```
 ### 错误 3: 深度分析失败
 **症状：** `success: false` from kb-deep-analyzer
 **处理：**
 - **整个流程终止**
 - 检查 AI_API_KEY 是否有效
 - 检查 AI_BASE_URL 是否正确
 - 检查模型配额是否用尽
 ### 错误 4: Whisper 模型未下载
 **症状：** `Model file not found`
 **解决：**
 ```bash
 cd ~/whisper.cpp
 ./models/download-ggml-model.sh base
 ```
 ---
 ## 状态管理（断点续传）
 > ⚠️ 此功能需要完成 P3 实现
 ```bash
 # 初始化状态
 python3 kb_workflow.py --init-state "[URL]"
 # 查询当前状态
 python3 kb_workflow.py --get-state "[URL]"
 # 推进到下一步
 python3 kb_workflow.py --advance-step "[URL]" "metadata"
 # 记录错误
 python3 kb_workflow.py --record-error "[URL]" "403 Forbidden"
 # 获取恢复指令
 python3 kb_workflow.py --resume "[URL]"
 ```
 ---
 ## ⚠️ 严格失败处理原则
-1. **深度分析不可跳过**：kb-deep-analyzer 是必须成功的步骤
+> **深度分析步骤必须成功，否则整个流程失败**
-2. **失败即终止**：如果深度分析返回 `success: false`，必须立即停止
+
-3. **禁止降级处理**：不能使用预设模板或简单摘要"假装成功"
+1. **深度分析不可跳过**：AI 分析是必须成功执行的步骤
-4. **每次执行都要使用 AI**：必须有真实的 AI API 调用
+2. **失败即终止**：如果分析返回失败，必须立即停止并汇报
 3. **禁止降级处理**：不能使用预设模板或简单摘要来"假装成功"
 4. **每次执行都要使用 AI**：必须有真实的 AI API 调用记录
 ## 辅助命令
 ```bash
 # 查重
 python3 kb-workflow/kb_workflow.py --check --url "https://..."
 # 获取目录路径
 python3 kb-workflow/kb_workflow.py --base-path
 python3 kb-workflow/kb_workflow.py --metadata
 python3 kb-workflow/kb_workflow.py --content
 python3 kb-workflow/kb_workflow.py --analyzed
 # 归档分析报告
 python3 kb-workflow/kb_workflow.py --archive-file "/path/to/report.md"
 ```
 ## 示例
 ```
 用户: 添加到知识库 https://www.bilibili.com/video/BV1xx
 Claude:
 ✅ 环境变量检查通过
 ◆ 处理中...
  → 解析 URL: bilibili 视频 BV1xx
  → 获取元数据...
  → 下载字幕...
  → 深度分析...
 ✅ 完成！
 标题: 【教程】如何使用 Python 进行数据分析
 处理阶段: 全部完成
 最终文件: ~/work/zq/30.资源/AI分析/2026-03-15/bilibili_BV1xx.md
 ```
 ## 技术架构
 ```
 kb-workflow/kb_workflow.py (CLI)
    └── kb_pipeline/ (核心管道包)
        ├── pipeline.py      — Pipeline 编排引擎
        ├── processors/      — 处理器实现
        │   ├── parse.py     — URL 解析
        │   ├── download.py  — 内容下载
        │   ├── asr.py       — 语音识别
        │   └── analyze.py   — 深度分析
        └── schemas.py       — 数据模型
 ```
@@ -5,7 +5,7 @@ KB Workflow - Knowledge Base Workflow Orchestrator
 知识库工作流辅助工具函数。
-此模块提供查重、文件操作、Git 管理、状态管理等工具函数，
+此模块提供查重、文件操作、Git 管理等工具函数，
 供主代理在执行 SKILL.md 中定义的 workflow 时调用。
 """
@@ -18,46 +18,7 @@ import shutil
 import re
 from datetime import datetime
 from pathlib import Path
-from typing import Dict, List, Optional, Any, Tuple
+from typing import Dict, List, Optional, Any
 # ==================== 状态机定义 ====================
 # 工作流步骤顺序
 WORKFLOW_STEPS = [
    "init",
    "link_parse",
    "metadata",
    "content",
    "whisper",      # 可选步骤
    "analyze",
    "archive",
    "done"
 ]
 # 每个步骤的必需输出字段
 STEP_OUTPUT_FIELDS = {
    "init": [],
    "link_parse": ["platform", "content_type", "content_id"],
    "metadata": ["title"],
    "content": ["filepath"],
    "whisper": ["output_files"],
    "analyze": ["output_file"],
    "archive": ["archive_path"],
    "done": []
 }
 # 步骤描述（用于汇报）
 STEP_DESCRIPTIONS = {
    "init": "初始化",
    "link_parse": "链接解析",
    "metadata": "元数据提取",
    "content": "内容下载",
    "whisper": "语音识别",
    "analyze": "深度分析",
    "archive": "归档",
    "done": "完成"
 }
 # ==================== 子技能映射 ====================
@@ -332,249 +293,6 @@ def archive_and_git(
    return result
 # ==================== 状态管理 ====================
 def get_state_file_path(url: str, base_path: Path = None) -> Path:
    """获取状态文件路径"""
    if base_path is None:
        base_path = get_knowledge_base_path()
    pending_dir = base_path / "pending"
    pending_dir.mkdir(parents=True, exist_ok=True)
    # 从 URL 提取标识符
    # 尝试提取 BV 号或其他 ID
    bv_match = re.search(r'BV[a-zA-Z0-9]+', url)
    if bv_match:
        content_id = bv_match.group()
        platform = "bilibili"
    else:
        # 尝试提取知乎 ID
        zhihu_match = re.search(r'zhihu\.com/(?:question|zvideo|p)/(\d+)', url)
        if zhihu_match:
            content_id = zhihu_match.group(1)
            platform = "zhihu"
        else:
            # 使用 URL hash 作为标识
            import hashlib
            content_id = hashlib.md5(url.encode()).hexdigest()[:12]
            platform = "unknown"
    return pending_dir / f"{platform}_{content_id}.state.json"
 def init_state(url: str, base_path: Path = None) -> Dict[str, Any]:
    """初始化工作流状态"""
    state_file = get_state_file_path(url, base_path)
    # 如果状态文件已存在，返回现有状态
    if state_file.exists():
        with open(state_file, 'r', encoding='utf-8') as f:
            return json.load(f)
    # 创建新状态
    now = datetime.now().isoformat()
    state = {
        "url": url,
        "platform": None,
        "content_id": None,
        "content_type": None,
        "current_step": "init",
        "completed_steps": [],
        "step_results": {},
        "errors": [],
        "created_at": now,
        "updated_at": now
    }
    with open(state_file, 'w', encoding='utf-8') as f:
        json.dump(state, f, ensure_ascii=False, indent=2)
    return state
 def get_state(url: str, base_path: Path = None) -> Dict[str, Any]:
    """获取当前工作流状态"""
    state_file = get_state_file_path(url, base_path)
    if not state_file.exists():
        return {
            "success": False,
            "error": "状态文件不存在",
            "url": url,
            "state_file": str(state_file)
        }
    with open(state_file, 'r', encoding='utf-8') as f:
        state = json.load(f)
    return {
        "success": True,
        "state": state,
        "state_file": str(state_file)
    }
 def advance_step(
    url: str,
    step_name: str,
    result: Dict[str, Any] = None,
    base_path: Path = None
 ) -> Dict[str, Any]:
    """推进到下一步并记录结果"""
    state_file = get_state_file_path(url, base_path)
    if not state_file.exists():
        return {
            "success": False,
            "error": "状态文件不存在，请先执行 --init-state"
        }
    with open(state_file, 'r', encoding='utf-8') as f:
        state = json.load(f)
    # 验证步骤名称
    if step_name not in WORKFLOW_STEPS:
        return {
            "success": False,
            "error": f"无效的步骤名称: {step_name}",
            "valid_steps": WORKFLOW_STEPS
        }
    # 获取当前步骤索引
    try:
        current_idx = WORKFLOW_STEPS.index(state["current_step"])
        new_idx = WORKFLOW_STEPS.index(step_name)
    except ValueError:
        return {
            "success": False,
            "error": f"步骤索引错误"
        }
    # 更新状态
    if step_name != state["current_step"]:
        # 标记当前步骤为完成
        if state["current_step"] not in state["completed_steps"]:
            state["completed_steps"].append(state["current_step"])
    state["current_step"] = step_name
    state["updated_at"] = datetime.now().isoformat()
    # 记录步骤结果
    if result:
        state["step_results"][step_name] = result
        # 提取平台和 ID 信息（从 link_parse 步骤）
        if step_name == "link_parse" and result.get("success"):
            state["platform"] = result.get("platform")
            state["content_id"] = result.get("content_id")
            state["content_type"] = result.get("type")
    # 保存状态
    with open(state_file, 'w', encoding='utf-8') as f:
        json.dump(state, f, ensure_ascii=False, indent=2)
    return {
        "success": True,
        "current_step": step_name,
        "completed_steps": state["completed_steps"],
        "state_file": str(state_file)
    }
 def record_error(
    url: str,
    error_msg: str,
    step: str = None,
    base_path: Path = None
 ) -> Dict[str, Any]:
    """记录错误"""
    state_file = get_state_file_path(url, base_path)
    if not state_file.exists():
        return {
            "success": False,
            "error": "状态文件不存在，请先执行 --init-state"
        }
    with open(state_file, 'r', encoding='utf-8') as f:
        state = json.load(f)
    # 记录错误
    error_entry = {
        "step": step or state["current_step"],
        "error": error_msg,
        "timestamp": datetime.now().isoformat()
    }
    state["errors"].append(error_entry)
    state["updated_at"] = datetime.now().isoformat()
    # 保存状态
    with open(state_file, 'w', encoding='utf-8') as f:
        json.dump(state, f, ensure_ascii=False, indent=2)
    return {
        "success": True,
        "error_recorded": error_entry,
        "total_errors": len(state["errors"])
    }
 def get_resume_info(url: str, base_path: Path = None) -> Dict[str, Any]:
    """获取恢复执行的指令"""
    state_file = get_state_file_path(url, base_path)
    if not state_file.exists():
        return {
            "success": False,
            "error": "状态文件不存在",
            "suggestion": "请使用 --init-state 初始化工作流"
        }
    with open(state_file, 'r', encoding='utf-8') as f:
        state = json.load(f)
    current_step = state["current_step"]
    step_desc = STEP_DESCRIPTIONS.get(current_step, current_step)
    # 检查是否有错误
    if state["errors"]:
        last_error = state["errors"][-1]
        return {
            "success": True,
            "status": "error",
            "current_step": current_step,
            "step_description": step_desc,
            "last_error": last_error,
            "completed_steps": state["completed_steps"],
            "suggestion": f"请先解决 {last_error['step']} 步骤的错误: {last_error['error']}"
        }
    # 检查是否已完成
    if current_step == "done":
        return {
            "success": True,
            "status": "completed",
            "current_step": "done",
            "step_description": "已完成",
            "completed_steps": state["completed_steps"],
            "step_results": state["step_results"]
        }
    # 返回下一步指令
    return {
        "success": True,
        "status": "in_progress",
        "current_step": current_step,
        "step_description": step_desc,
        "completed_steps": state["completed_steps"],
        "next_action": f"继续执行步骤: {step_desc}",
        "platform": state.get("platform"),
        "content_id": state.get("content_id"),
        "content_type": state.get("content_type"),
        "step_results": state.get("step_results", {})
    }
 def generate_summary(url: str, stages: List[Dict], final_file: str, archive_result: Dict = None) -> str:
    """生成处理完成汇报"""
    # 提取视频标题（如果有）
@@ -607,6 +325,83 @@ def generate_summary(url: str, stages: List[Dict], final_file: str, archive_resu
    return summary
 # ==================== CLI Process Command ====================
 def process_command(
    url: str,
    output_dir: Optional[str] = None,
    skip_completed: bool = True
 ) -> Dict[str, Any]:
    """Execute full pipeline for a URL
    Args:
        url: URL to process
        output_dir: Optional output directory for content
        skip_completed: Skip already completed steps (default True)
    Returns:
        Dictionary with success status, content_dir, and results
    """
    import asyncio
    # Add project root to path for kb_pipeline imports
    project_root = Path(__file__).parent.parent
    if str(project_root) not in sys.path:
        sys.path.insert(0, str(project_root))
    try:
        from kb_pipeline.pipeline import Pipeline
        from kb_pipeline.processors import ParseProcessor, DownloadProcessor, ASRProcessor, AnalyzeProcessor
        from kb_pipeline.schemas import PipelineConfig
        async def run_pipeline():
            # Determine output directory
            out_dir = Path(output_dir) if output_dir else Path.cwd()
            # Create processors
            processors = [
                ParseProcessor(),
                DownloadProcessor(),
                ASRProcessor(),
                AnalyzeProcessor(),
            ]
            # Create pipeline config
            config = PipelineConfig(
                steps=["parse", "download", "asr", "analyze"],
                output_dir=out_dir,
                skip_completed=skip_completed,
            )
            # Create and run pipeline
            pipeline = Pipeline(
                processors=processors,
                config=config,
                output_dir=out_dir,
            )
            result = await pipeline.run(url)
            return result
        # Run async pipeline
        result = asyncio.run(run_pipeline())
        return {
            "success": result.success,
            "error": result.error_message,
            "content_dir": str(result.content_dir) if result.content_dir else None,
            "results": result.results,
        }
    except Exception as e:
        return {
            "success": False,
            "error": str(e),
            "content_dir": None,
            "results": {},
        }
 # ==================== CLI ====================
 def main():
@@ -618,117 +413,96 @@ def main():
 使用示例:
  # 查重
  python kb_workflow.py --check --url "https://www.bilibili.com/video/BVxxx"
-  
+
  # 归档
-  python kb_workflow.py --archive-file "/path/to/analyzed.md"
+  python kb_workflow.py --archive "/path/to/analyzed.md"
-  
+
-  # 状态管理
+  # 处理URL (通过完整管道)
-  python kb_workflow.py --init-state --url "https://..."
+  python kb_workflow.py process "https://www.bilibili.com/video/BVxxx"
  python kb_workflow.py --get-state --url "https://..."
  python kb_workflow.py --advance-step --url "https://..." --step "metadata" --result '{"success":true}'
  python kb_workflow.py --record-error --url "https://..." --error-msg "403 Forbidden"
  python kb_workflow.py --resume --url "https://..."
        """
    )
-    
+
-    # 基础参数
+    # Create subparsers for commands
-    parser.add_argument('--url', type=str, help='URL 地址')
+    subparsers = parser.add_subparsers(dest='command', help='Available commands')
-    parser.add_argument('--base-path', type=str, help='知识库根目录')
+
-    
+    # Process subcommand
-    # 查重和归档
+    process_parser = subparsers.add_parser('process', help='Process URL through full pipeline')
    process_parser.add_argument('url', type=str, help='URL to process')
    process_parser.add_argument('--output-dir', '-o', type=str, help='Output directory for content')
    process_parser.add_argument('--skip-completed', action='store_true', default=True,
                                 help='Skip completed steps (default: True)')
    process_parser.add_argument('--no-skip', action='store_false', dest='skip_completed',
                                 help='Re-run all steps')
    # Legacy arguments (for backward compatibility)
    parser.add_argument('--check', action='store_true', help='查重')
    parser.add_argument('--url', type=str, help='要检查的URL')
    parser.add_argument('--archive-file', type=str, help='归档的文件路径')
-    
+    parser.add_argument('--base-path', type=str, help='获取知识库根目录')
-    # 目录路径查询
+
-    parser.add_argument('--pending', action='store_true', help='获取 URL 解析目录')
+    parser.add_argument('--pending', action='store_true', help='获取url解析目录')
    parser.add_argument('--metadata', action='store_true', help='获取元数据目录')
    parser.add_argument('--content', action='store_true', help='获取内容目录')
-    parser.add_argument('--content_downloads', action='store_true', help='获取内容下载目录')
+    parser.add_argument('--content_downloads', action='store_true', help='获取主内容下载目录')
    parser.add_argument('--analyzed', action='store_true', help='获取深度分析目录')
    parser.add_argument('--archive_dir', action='store_true', help='获取归档目录')
    parser.add_argument('--archive_subscriptions', action='store_true', help='获取订阅信息目录')
-    parser.add_argument('--extract_platform', type=str, help='从文件名提取平台名称')
+    parser.add_argument('--extract_platform', type=str, help='要提取平台名称的文件名')
    # 状态管理
    parser.add_argument('--init-state', action='store_true', help='初始化工作流状态')
    parser.add_argument('--get-state', action='store_true', help='获取当前状态')
    parser.add_argument('--advance-step', action='store_true', help='推进到下一步')
    parser.add_argument('--step', type=str, help='步骤名称')
    parser.add_argument('--result', type=str, help='步骤结果（JSON 字符串）')
    parser.add_argument('--record-error', action='store_true', help='记录错误')
    parser.add_argument('--error-msg', type=str, help='错误信息')
    parser.add_argument('--resume', action='store_true', help='获取恢复执行信息')
    args = parser.parse_args()
-    
+
    base_path = Path(args.base_path) if args.base_path else None
    dirs = ensure_directories(base_path)
-    
+
-    # 查重
+    # Handle process subcommand
    if args.command == 'process':
        result = process_command(
            url=args.url,
            output_dir=args.output_dir,
            skip_completed=args.skip_completed
        )
        print(json.dumps(result, ensure_ascii=False, indent=2))
        return
    # Legacy argument handling
    if args.check and args.url:
        result = check_duplicate(base_path, args.url)
        print(json.dumps(result, ensure_ascii=False, indent=2))
-    
+
    # 归档
    elif args.archive_file:
        result = archive_and_git(args.archive_file, base_path)
        print(json.dumps(result, ensure_ascii=False, indent=2))
-    
+
    # 目录路径查询
    elif args.pending:
        print(json.dumps({"path": str(dirs["pending"])}, ensure_ascii=False, indent=2))
    elif args.metadata:
        print(json.dumps({"path": str(dirs["metadata"])}, ensure_ascii=False, indent=2))
    elif args.content:
        print(json.dumps({"path": str(dirs["content"])}, ensure_ascii=False, indent=2))
    elif args.content_downloads:
        print(json.dumps({"path": str(dirs["content_downloads"])}, ensure_ascii=False, indent=2))
    elif args.analyzed:
        print(json.dumps({"path": str(dirs["analyzed"])}, ensure_ascii=False, indent=2))
    elif args.archive_dir:
        print(json.dumps({"path": str(get_archive_path())}, ensure_ascii=False, indent=2))
    elif args.archive_subscriptions:
        print(json.dumps({"path": str(dirs["archive_subscriptions"])}, ensure_ascii=False, indent=2))
    elif args.extract_platform:
        platform = extract_platform(args.extract_platform)
        print(json.dumps({"platform": platform}, ensure_ascii=False, indent=2))
    elif args.base_path:
        print(json.dumps({"path": str(dirs["base"])}, ensure_ascii=False, indent=2))
-    
+
    # 状态管理
    elif args.init_state and args.url:
        result = init_state(args.url, base_path)
        print(json.dumps({"success": True, "state": result}, ensure_ascii=False, indent=2))
    elif args.get_state and args.url:
        result = get_state(args.url, base_path)
        print(json.dumps(result, ensure_ascii=False, indent=2))
    elif args.advance_step and args.url and args.step:
        result_data = None
        if args.result:
            try:
                result_data = json.loads(args.result)
            except json.JSONDecodeError:
                result_data = {"raw": args.result}
        result = advance_step(args.url, args.step, result_data, base_path)
        print(json.dumps(result, ensure_ascii=False, indent=2))
    elif args.record_error and args.url and args.error_msg:
        result = record_error(args.url, args.error_msg, args.step, base_path)
        print(json.dumps(result, ensure_ascii=False, indent=2))
    elif args.resume and args.url:
        result = get_resume_info(args.url, base_path)
        print(json.dumps(result, ensure_ascii=False, indent=2))
    else:
        parser.print_help()
 if __name__ == '__main__':
    main()
 if __name__ == '__main__':
    main()
@@ -1,154 +1,86 @@
 ---
 name: kb-zhihu-content
 description: 知乎内容下载 - 下载视频、提取文章内容
 ---
 # KB Zhihu Content Skill
-**kb-zhihu-content** - 下载知乎视频、提取文章内容。
+**kb-zhihu-content** - Download Zhihu videos and extract content.
-## 什么时候使用
+## Features
- kb-workflow 步骤 3a：下载知乎视频
+- Video download using **yt-dlp**
- kb-workflow 步骤 3b：提取知乎专栏文章内容
+- Subtitle extraction (returns no subtitle, triggers Whisper)
 - **Read article content from MediaCrawler data** (NEW!)
 - Supports `ZHIHU_COOKIE` environment variable
-## 功能
+## Environment Variables
- 视频下载（使用 yt-dlp）
+| Variable | Required | Description |
- 从 MediaCrawler 数据读取文章内容
+|----------|----------|-------------|
- 支持 `ZHIHU_COOKIE` 环境变量
+| `ZHIHU_COOKIE` | No | 知乎登录 Cookie (for member-only videos) |
-## 环境变量
+## Commands
 | 变量名 | 必需 | 说明 |
 |--------|------|------|
 | `ZHIHU_COOKIE` | 否 | 知乎登录 Cookie |
 ## 命令
 ```bash
-# 下载视频
+# Download video
-python3 zhihu_content.py --id "12345678" --type video --output "/path/to/downloads/"
+python zhihu_content.py --id "12345678" --type video --output "/path/to/downloads/"
-# 从 MediaCrawler 读取文章内容
+# Extract subtitle (returns no subtitle, triggers Whisper)
-python3 zhihu_content.py --type article-from-media \
+python zhihu_content.py --id "12345678" --type subtitle --output "/path/to/subtitles/"
 # 从 MediaCrawler 读取文章内容 (⚡ NEW!)
 python zhihu_content.py --type article-from-media --input "~/work/MediaCrawler/data/zhihu/json/detail_contents_2026-02-27.json" --output "~/work/knowledge-base/data/zhihu/"
 ```
 ## MediaCrawler 数据读取
 **新功能**：从 MediaCrawler 爬取的数据中提取知乎文章内容
 ### 使用方法
 ```bash
 # 从 MediaCrawler JSON 文件读取文章内容
 python zhihu_content.py --type article-from-media \
    --input "~/work/MediaCrawler/data/zhihu/json/detail_contents_2026-02-27.json" \
    --output "~/work/knowledge-base/data/zhihu/"
-# 自动查找最新 MediaCrawler 数据
+# 查找最新的爬取数据
-python3 zhihu_content.py --type article-from-media \
+python zhihu_content.py --type article-from-media \
    --output "~/work/knowledge-base/data/zhihu/"
 ```
-## 输出格式 Schema
+### 参数说明
-### 必需字段
+| 参数 | 说明 |
 |------|------|
 | `--type article-from-media` | 从 MediaCrawler 数据读取 |
 | `--input` | MediaCrawler 输出文件路径（可选，不指定则自动查找最新） |
 | `--output` | 输出目录 |
-| 字段 | 类型 | 说明 |
+### 输出格式
 |------|------|------|
 | `success` | boolean | 是否成功 |
 | `type` | string | 操作类型（video/article） |
 ### 视频下载特有字段
 | 字段 | 类型 | 说明 |
 |------|------|------|
 | `filepath` | string | 视频文件路径 |
 | `filename` | string | 视频文件名 |
 ### 文章提取特有字段
 | 字段 | 类型 | 说明 |
 |------|------|------|
 | `filepath` | string | 文章文件路径 |
 | `filename` | string | 文章文件名 |
 | `content_length` | number | 内容长度 |
 ## 成功示例
 ### 视频下载
 ```json
 {
  "success": true,
  "type": "video",
  "filename": "zhihu_12345678.mp4",
  "filepath": "content/downloads/zhihu_12345678.mp4",
  "message": "下载完成"
 }
 ```
 ### 文章内容提取
 ```json
 {
  "success": true,
  "type": "article",
  "filename": "zhihu_article_如何高效的爬取微信公众号.md",
-  "filepath": "~/work/knowledge-base/data/zhihu/zhihu_article_如何高效的爬取微信公众号.md",
+  "filepath": "/home/xxx/work/knowledge-base/data/zhihu/zhihu_article_如何高效的爬取微信公众号.md",
  "content_length": 4563,
  "message": "文章内容已导出"
 }
 ```
-## 错误示例
+## 原有功能
-### 视频下载失败
+### Video Download
-```json
+Video download uses **yt-dlp**:
-{
+1. Parses `ZHIHU_COOKIE` environment variable (optional)
-  "success": false,
+2. Creates temporary Netscape-format cookie file
-  "type": "video",
+3. Runs `yt-dlp -f best -o <output> <url>`
-  "error": "视频不存在或无法访问",
+4. Cleans up temporary file
  "suggestion": "请检查视频 ID 是否正确"
 }
 ```
-### MediaCrawler 数据未找到
+**Note**: 知乎视频通常没有官方字幕，将自动触发 Whisper 语音识别
-```json
+## Installation
 {
  "success": false,
  "type": "article-from-media",
  "error": "未找到 MediaCrawler 数据文件",
  "suggestion": "请先运行 MediaCrawler 爬取文章"
 }
 ```
 ### 网络错误
 ```json
 {
  "success": false,
  "error": "网络请求失败: Connection timeout"
 }
 ```
 ## 安装
 ```bash
-cp -r kb-zhihu-content/ ~/.openclaw/skills/
+# Already included in kb-skills repo
 # Ensure yt-dlp is installed
 pip install yt-dlp aiohttp
 ```
 ## 在 kb-workflow 中的使用
 ```bash
 # 步骤 3a：下载知乎视频
 python3 ~/.openclaw/skills/kb-zhihu-content/zhihu_content.py \
    --id "12345678" --type video \
    --output "content/downloads/"
 # 注意：知乎视频没有官方字幕，下载后需要进入 Whisper 步骤
 # 步骤 3b：提取专栏文章（如果有 MediaCrawler 数据）
 python3 ~/.openclaw/skills/kb-zhihu-content/zhihu_content.py \
    --type article-from-media \
    --output "content/"
 ```
 ## 注意事项
 - **知乎视频没有官方字幕**：下载后必须进入 Whisper 步骤
 - MediaCrawler 需要单独安装和运行
 - 部分视频可能需要登录才能下载
@@ -1,207 +1,101 @@
 ---
 name: kb-zhihu-metadata
 description: 知乎元数据提取 - 获取问题/回答/专栏/视频的详细信息
 ---
 # KB Zhihu Metadata Skill
-**kb-zhihu-metadata** - 提取知乎问题、回答、用户、专栏的元数据。
+**kb-zhihu-metadata** - Extract Zhihu question/answer/user metadata and crawl articles.
-## 什么时候使用
+## Features
- kb-workflow 步骤 2：获取知乎内容的元数据
+- Question metadata extraction
- 需要问题标题、回答者、文章标题等信息
+- Answer metadata extraction
-
+- User information extraction
-## 功能
+- **Article crawling using MediaCrawler** (NEW!)
-
+- JSON output format
- 问题元数据提取
+- Markdown conversion support
- 回答元数据提取
+- **Cookie 支持**：可配置 ZHIHU_COOKIE 访问需要登录的内容
 - 用户信息提取
 - 专栏文章爬取（使用 MediaCrawler）
 - 支持 `ZHIHU_COOKIE` 环境变量
 ## 环境变量
-| 变量名 | 必需 | 说明 |
+| 变量名 | 用途 | 必需 |
 |--------|------|------|
-| `ZHIHU_COOKIE` | 推荐 | 知乎登录 Cookie（访问专栏等需要） |
+| `ZHIHU_COOKIE` | 知乎登录 Cookie | 推荐（访问专栏等需要） |
-**获取 Cookie：**
+**获取 ZHIHU_COOKIE**：
 1. 登录知乎网页版
-2. F12 → Application → Cookies → zhihu.com
+2. 打开开发者工具 (F12) → Application → Cookies → zhihu.com
 3. 复制 Cookie 字符串
 ```bash
-export ZHIHU_COOKIE="你的知乎Cookie"
+# 添加到 ~/.kb_env
 echo 'export ZHIHU_COOKIE="你的知乎Cookie"' >> ~/.kb_env
 source ~/.kb_env
 ```
-## 命令
+## Commands
 ```bash
-# 问题元数据
+# Question metadata
-python3 zhihu_metadata.py --id "266309957" --type question
+python zhihu_metadata.py --id "266309957" --type question
-# 回答元数据
+# Answer metadata
-python3 zhihu_metadata.py --id "2003427891925509507" --type answer
+python zhihu_metadata.py --id "2003427891925509507" --type answer
-# 知乎视频
+# 专栏文章爬取 (使用 MediaCrawler) ⚡ NEW!
-python3 zhihu_metadata.py --id "123456789" --type zvideo
+python zhihu_article_crawler.py --url "https://zhuanlan.zhihu.com/p/1894054073625461813" --output "~/work/knowledge-base/data/zhihu/"
 # 专栏文章爬取（使用 MediaCrawler）
 python3 zhihu_article_crawler.py --url "https://zhuanlan.zhihu.com/p/1894054073625461813" --output "~/work/knowledge-base/data/zhihu/"
 ```
-## 输出格式 Schema
+## MediaCrawler 爬取文章
-### 必需字段
+**新功能**：使用 MediaCrawler 爬取知乎专栏文章
-| 字段 | 类型 | 说明 |
+### 环境要求
 |------|------|------|
 | `success` | boolean | 是否提取成功 |
 | `platform` | string | 固定为 `zhihu` |
 | `content_id` | string | 内容 ID |
 | `content_type` | string | 内容类型 |
 | `title` | string | 标题 |
-### 可选字段
+MediaCrawler 已安装在 `~/work/MediaCrawler`
-| 字段 | 类型 | 说明 |
+### 使用方法
 |------|------|------|
 | `author` | string | 作者名称 |
 | `excerpt` | string | 摘要 |
 | `stats` | object | 统计数据 |
 | `output_file` | string | 输出文件路径 |
 | `error` | string | 错误信息 |
-## 成功示例
+```bash
 # 爬取单篇文章
 python zhihu_article_crawler.py --url "https://zhuanlan.zhihu.com/p/XXXXX"
-### 问题元数据
+# 指定输出目录
 python zhihu_article_crawler.py --url "https://zhuanlan.zhihu.com/p/XXXXX" --output "/path/to/output/"
 # 输出示例
 {
  "success": true,
  "type": "article",
  "title": "如何高效的爬取微信公众号",
  "filepath": "/home/xxx/work/knowledge-base/data/zhihu/zhihu_article_如何高效的爬取微信公众号.md",
  "content_length": 4563,
  "message": "爬取成功: 如何高效的爬取微信公众号"
 }
 ```
 ### 工作原理
 1. 调用 MediaCrawler 爬取文章
 2. 从 JSON 输出中提取内容
 3. 转换为 Markdown 格式（带 frontmatter）
 4. 保存到指定目录
 ## Output
 ```json
 {
  "success": true,
-  "platform": "zhihu",
+  "data": {
-  "content_type": "question",
+    "platform": "zhihu",
-  "content_id": "266309957",
+    "type": "question",
-  "title": "问题标题",
+    "id": "266309957",
-  "excerpt": "问题描述...",
+    "title": "问题标题",
-  "stats": {
+    "author": "作者名字",
    "answer_count": 50,
-    "follower_count": 1000,
+    "followers": 1000
-    "view_count": 50000
+  }
  },
  "output_file": "metadata/zhihu_266309957.meta.json"
 }
 ```
-### 回答元数据
+## Installation
 ```json
 {
  "success": true,
  "platform": "zhihu",
  "content_type": "answer",
  "content_id": "2003427891925509507",
  "title": "问题标题（回答所属问题）",
  "author": "回答者名字",
  "excerpt": "回答摘要...",
  "stats": {
    "voteup_count": 1000,
    "comment_count": 50
  },
  "output_file": "metadata/zhihu_2003427891925509507.meta.json"
 }
 ```
 ### 专栏文章
 ```json
 {
  "success": true,
  "platform": "zhihu",
  "content_type": "zhuanlan",
  "content_id": "1894054073625461813",
  "title": "文章标题",
  "author": "作者名字",
  "excerpt": "文章摘要...",
  "output_file": "metadata/zhihu_1894054073625461813.meta.json"
 }
 ```
 ### 知乎视频
 ```json
 {
  "success": true,
  "platform": "zhihu",
  "content_type": "zvideo",
  "content_id": "123456789",
  "title": "视频标题",
  "author": "作者名字",
  "duration": "10:30",
  "output_file": "metadata/zhihu_123456789.meta.json"
 }
 ```
 ## 错误示例
 ### 内容不存在
 ```json
 {
  "success": false,
  "platform": "zhihu",
  "content_id": "invalid_id",
  "content_type": "question",
  "error": "问题不存在或已被删除"
 }
 ```
 ### 需要登录
 ```json
 {
  "success": false,
  "platform": "zhihu",
  "content_id": "1894054073625461813",
  "content_type": "zhuanlan",
  "error": "需要登录才能访问",
  "suggestion": "请配置 ZHIHU_COOKIE 环境变量"
 }
 ```
 ### 网络错误
 ```json
 {
  "success": false,
  "error": "网络请求失败: Connection timeout"
 }
 ```
 ## 安装
 ```bash
-cp -r kb-zhihu-metadata/ ~/.openclaw/skills/
+cp -r kb-zhihu-metadata/ ~/skills/
 pip install httpx
 ```
 ## 在 kb-workflow 中的使用
 ```bash
 # 步骤 2：元数据提取（知乎问题）
 python3 ~/.openclaw/skills/kb-zhihu-metadata/zhihu_metadata.py \
    --id "266309957" --type question \
    --output "metadata/zhihu_266309957.meta.json"
 # 步骤 2：元数据提取（知乎视频）
 python3 ~/.openclaw/skills/kb-zhihu-metadata/zhihu_metadata.py \
    --id "123456789" --type zvideo \
    --output "metadata/zhihu_123456789.meta.json"
 ```
 ## 注意事项
 - 专栏文章可能需要登录才能访问
 - MediaCrawler 需要单独安装：`~/work/MediaCrawler`
 - 部分内容可能有访问限制