Doc Collaboration Watcher

实时监控指定协作文档变更，自动通过飞书、微信、iMessage和WebChat通知相关代理并跟踪确认与响应状态。

Security Scan

The skill's behavior mostly matches its stated purpose, but it automatically reads the user's OpenClaw config (which may contain channel credentials) without declaring that requirement and the repository includes at least one potential secret-like token in PUBLISH.md — worth review before enabling.

Purpose & Capability

Name/description (document change watcher + multi-channel notification) align with included files and declared permissions (file read/write, message:send). The single Python script and no external downloads are reasonable for the stated functionality.

Instruction Scope

SKILL.md / README and bin/doc-watcher.py instruct the skill to automatically read ~/.openclaw/config/openclaw.json and to use enabled channels from the user's OpenClaw config. Reading that config is functionally relevant, but SKILL.md/README present 'zero-config' behavior without clearly warning users that local channel configurations (which often hold tokens/credentials) will be read and used. The repo also contains a PUBLISH.md example that embeds a CLAWHUB token string (clh_RVZ_...), which could be a real token or a sensitive placeholder — storing tokens in repo files is a risk. Additionally the runtime script prints local file:// paths in notifications (exposes local paths) and the main loop uses while True: pass (a busy spin), which is a bug/operational concern.

Install Mechanism

No network downloads or arbitrary installers; install is manual (git clone + pip install watchdog). No extracted archives or remote binaries. This is low risk from an install mechanism perspective.

Credentials

The skill declares no required env vars, which is reasonable, but it silently reads the user's OpenClaw config to discover channels. OpenClaw channel configs commonly contain credentials/tokens for messaging integrations; the skill doesn't declare that it will access these secrets, nor does SKILL.md ask the user to review or approve that access. The presence of a CLAWHUB token-like string in PUBLISH.md increases the concern about sensitive data handling in the repo.

Persistence & Privilege

Skill is not always-enabled, does not request platform-wide persistence, and does not modify other skills' configuration. It writes its own log/history under the workspace logs which is expected for this kind of tool.

What to consider before installing

Plain-language checklist before installing/enabling: - Review your OpenClaw config (~/.openclaw/config/openclaw.json). It will be read automatically and may contain channel credentials (tokens, API keys). Only enable this skill if you are comfortable it can access and use your configured channels. - Inspect the repo files (especially PUBLISH.md and other docs) for any accidentally committed tokens/keys. The PUBLISH.md contains a CLAWHUB_TOKEN-like string — treat it as sensitive until confirmed otherwise; do not assume it's harmless. - The current Python script prints notification info and contains a TODO for real message sending; verify the implementation that actually sends messages before enabling in production. Confirm which channels will receive notifications and whether messages may expose local paths (the notifications include file:// links to local files). - Run the watcher in a sandbox or a test workspace first (not your main workspace). Create test .md files and confirm behavior (and that it doesn't leak data externally). - Fix or be aware of the busy-loop in main (while True: pass) which will consume CPU; modify to sleep or use a proper event loop before long-running use. - If you need stricter control, require explicit config (disable 'auto_channels') so the skill only uses channels you specify, or whitelist channels it may use. If you want, I can: - Point to the exact lines in files that read the OpenClaw config and show what fields to look for in your openclaw.json, - Suggest a small patch to replace the busy-loop with a safe sleep, and - Help redact or rotate any tokens found in the repository.

协作文档监控技能 (Doc-Collaboration-Watcher)

版本: v1.0
创建: 2026-04-07
作者: 伊娃
状态: 开发中

---

📋 技能概述

实时监控协作文档变更，自动通知所有相关子代理和沟通通道，确保多代理协作信息同步。

适用场景：

• 多代理协作项目（如 ESP32 固件 + 前端 + 后端）
• 跨团队接口对齐
• 分布式文档维护
• 需要实时同步的协作场景

---

🎯 核心功能

1. 文档变更监控

• 监控指定目录下的 Markdown 文件
• 检测文件创建、修改、删除
• 计算文件哈希，避免误报

2. 实时通知

• 飞书通道：发送变更消息
• 微信通道：发送变更消息
• iMessage 通道：发送变更消息
• WebChat 会话：发送变更消息

3. 变更历史

• 记录所有文档变更
• 保存变更时间、文件、大小
• 支持查询历史变更

4. 响应追踪

• 追踪各代理确认状态
• 超时未确认时提醒
• 生成响应统计报告

---

📦 安装

# 1. 克隆技能到 workspace
cd ~/.openclaw/workspace/skills
git clone <repo-url> doc-collaboration-watcher

# 2. 安装依赖
pip3 install watchdog

# 3. 配置监控目录
cp config.example.json config.json
# 编辑 config.json 设置监控目录、通知通道等

# 4. 启动监控
python3 bin/doc-watcher.py

---

⚙️ 配置

config.json

{
  "workspace": "/Users/bot-eva/.openclaw/workspace",
  "docs_dir": "docs",
  "log_dir": "logs",
  "watched_files": [
    "esp32-collaboration.md",
    "esp32-collaboration-discussion.md"
  ],
  "notification": {
    "channels": ["feishu", "wechat", "imessage", "webchat"],
    "response_timeout_minutes": 5,
    "evaluation_timeout_minutes": 30
  },
  "agents": [
    {"name": "伊娃 - 固件", "role": "firmware"},
    {"name": "伊娃 - 前端", "role": "frontend"},
    {"name": "伊娃 - 后端", "role": "backend"}
  ]
}

---

🚀 使用

启动监控

# 后台运行
python3 bin/doc-watcher.py &

# 或使用 systemd（Linux）
sudo systemctl start doc-watcher

查看变更历史

cat logs/doc_change_history.json

测试通知

# 修改监控文件
echo "# 测试变更" >> docs/esp32-collaboration.md

# 观察通知输出

---

📊 通知格式

飞书/微信/iMessage

📢 文档变更通知

文件：esp32-collaboration.md
时间：2026-04-07 20:50:15
变更类型：修改
大小：15.2 KB → 16.8 KB

要求：
✅ 5 分钟内确认收到
✅ 30 分钟内评估影响

查看详情：file:///Users/bot-eva/.openclaw/workspace/docs/esp32-collaboration.md

WebChat

🔍 检测到协作文档变更

**文件**: esp32-collaboration.md
**时间**: 2026-04-07 20:50:15
**变更**: v1.0 → v1.1

**影响评估**:
- 接口变更：是/否
- 需要配合：是/否
- 紧急程度：高/中/低

[查看详情] [确认收到] [需要讨论]

---

🔧 集成 OpenClaw

作为 OpenClaw 技能

# skill.yaml
name: doc-collaboration-watcher
version: 1.0.0
description: 协作文档实时监控
author: Eva
tools:
  - file:watch
  - message:send
  - cron:schedule
permissions:
  - file:read
  - file:write
  - message:send

OpenClaw 配置

// openclaw.json
{
  "skills": {
    "doc-collaboration-watcher": {
      "enabled": true,
      "config": {
        "docs_dir": "docs",
        "channels": ["feishu", "wechat", "imessage"]
      }
    }
  }
}

---

📈 监控指标

变更统计

• 每日变更次数
• 各文件变更频率
• 平均响应时间

响应统计

• 确认率（5 分钟内）
• 评估及时率（30 分钟内）
• 超时次数

报告示例

## 文档监控周报 (2026-04-01 ~ 2026-04-07)

### 变更统计
- 总变更：23 次
- esp32-collaboration.md: 15 次
- esp32-collaboration-discussion.md: 8 次

### 响应统计
- 确认率：95% (22/23)
- 评估及时率：87% (20/23)
- 超时：1 次（伊娃 - 前端，2026-04-05 14:30）

### 改进建议
- 增加变更摘要（避免频繁查看全文）
- 添加变更对比（diff）功能

---

🤝 协作流程

标准流程

1. 伊娃 A 修改文档
   ↓
2. 监控系统检测到变更
   ↓
3. 自动通知所有通道和代理
   ↓
4. 各代理确认收到（5 分钟内）
   ↓
5. 相关代理评估影响（30 分钟内）
   ↓
6. 需要配合 → 调整工作计划
   ↓
7. 无需配合 → 继续原计划

接口变更流程

1. 在文档中修改接口定义
   ↓
2. 更新版本记录表
   ↓
3. 系统自动通知
   ↓
4. 相关方评估（30 分钟）
   ↓
5. 确认无误 → 各自实现
   ↓
6. 有分歧 → 子代理评审 → 老板决策

---

🧪 测试计划

阶段 1：基础监控

• 文件修改检测
• 文件创建检测
• 哈希计算正确
• 变更历史记录

阶段 2：通知功能

• 飞书通知
• 微信通知
• iMessage 通知
• WebChat 通知

阶段 3：响应追踪

• 确认状态追踪
• 超时提醒
• 统计报告

阶段 4：OpenClaw 集成

• 技能打包
• 配置管理
• 权限控制
• ClawHub 发布

---

📝 版本记录

版本	日期	变更	作者
v1.0	2026-04-07	初始版本	伊娃

---

🔗 相关链接

• 主文档: /Users/bot-eva/.openclaw/workspace/docs/esp32-collaboration.md
• 讨论记录: /Users/bot-eva/.openclaw/workspace/docs/esp32-collaboration-discussion.md
• 监控脚本: /Users/bot-eva/.openclaw/workspace/bin/doc-watcher.py
• ClawHub: https://clawhub.ai/skills/doc-collaboration-watcher

---

本技能由伊娃开发，用于多代理协作文档实时监控