← Back to Skills Marketplace
mikeliang2000

企微 Agent Ops Center

by mikeliang2000 · GitHub ↗ · v2.4.0 · MIT-0
cross-platform ⚠ suspicious
38
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install wecom-agent-ops-center
Description
在企微群实时监控AI Agent健康状态,自动告警、进程守护、敏感信息拦截和多租户隔离,支持WorkBuddy/OpenClaw。
README (SKILL.md)

企微 Agent Ops Center

在企微群里看到你的 AI Agent 在做什么。 实时心跳监控、异常秒级告警、进程自动守护、敏感信息拦截——让 AI 从「黑盒」变「透明」。

WorkBuddy / OpenClaw 用户 2 分钟接入,0 学习成本。

解决什么痛点

你的问题 我们怎么解决
Agent 挂了没人知道,第二天才发现 心跳检测 + 企微群自动告警
不知道 Agent 今天处理了什么任务 任务进度直播 + 每日健康报告
Agent 崩溃后没人重启 进程守护 + 指数退避自动重启
担心 Agent 泄露 API Key 19 种敏感信息规则实时扫描拦截
不确定 Agent 访问了哪些外部服务 网络请求白名单守卫
多个用户/团队的 Agent 数据混在一起 多租户隔离(按 WorkBuddy/OpenClaw 用户 ID 自动分区)
分不清 Agent 来自哪个平台 自动识别平台来源(WorkBuddy / OpenClaw),支持分类统计

v2.4 核心特性

  • 🧠 Agent 自动发现:安装 Skill 后自动扫描 ~/.workbuddy/skills/ 下的所有 Agent,批量注册到 Ops Center,零手动配置
  • 👥 多租户隔离:按 WorkBuddy/OpenClaw 用户 ID 自动分区,不同用户/团队的 Agent 数据完全隔离
  • 🏷️ 平台自动识别:自动检测当前平台(WorkBuddy / OpenClaw),Agent 打上 platform 标签,支持分类统计
  • 🛡️ 进程守护:Agent 崩溃后自动重启(指数退避,最多 10 次)
  • 🔒 Secrets 扫描:19 种敏感信息规则(API Key / 私钥 / 密码),支持 block / redact / warn 三种模式
  • 🌐 网络守卫:Monkey-patch http/https 模块,域名 / IP / CIDR 白名单,阻断非法外联

快速开始

1. 安装

# ClawHub 安装(推荐)
clawhub install wecom-agent-ops-center

# 或本地启动
git clone \x3Crepo> && cd wecom-agent-ops-center
npm install

2. 获取企微凭证

企微管理后台 → 应用管理 → 智能机器人
→ 创建应用(或选择已有)
→ 复制 Bot ID 和 Bot Secret

3. 配置

cp config.yaml.example config.yaml
# 编辑 config.yaml,填入 bot_id、bot_secret

最小配置(WorkBuddy 用户):

wecom:
  bot_id: "你的BotID"
  bot_secret: "你的BotSecret"
# agent-scanner 会自动扫描 ~/.workbuddy/skills/ 并批量注册!

最小配置(OpenClaw 用户):

wecom:
  bot_id: "你的BotID"
  bot_secret: "你的BotSecret"
# agent-scanner 会自动扫描 ~/.openclaw/ 并批量注册!

💡 不需要手动注册 Agent。安装后 scanner 自动扫描本地所有 Agent,批量上报到 Ops Center。平台类型(WorkBuddy/OpenClaw)和租户 ID(按用户隔离)全部自动识别。

4. 启动

# 本地开发
node connector.js

# 服务器部署(推荐,开机自启)
sudo systemctl enable --now wecom-agent-ops-center

5. 验证

  • 在企微里 @机器人 你好 → Agent 应回复
  • 浏览器打开 http://localhost:9527 → 查看 Agent 健康状态
  • Agent 离线 3 次心跳后 → 企微群收到告警卡片

命令行

node connector.js              # 启动(监控 + 消息转发)
node connector.js pair         # 生成 P2P 配对码
node connector.js join \x3CCODE>  # 加入 P2P 配对
node connector.js status       # 查看连接状态
node connector.js peers       # 查看已配对节点

环境变量(免配置文件)

export WECOM_BOT_ID=xxx
export WECOM_BOT_SECRET=xxx
export AGENT_ENDPOINT=http://...
export MONITOR_NOTIFY_CHATID=群聊ID
node connector.js

常见问题

Q:和企微自带的 AI 有什么区别? A:企微自带的是「AI 聊天」。我们是「Agent 运维工具」——不管你的 AI 是谁家的,我们帮你监控它是否正常运行。

Q:我的 Agent 不是 HTTP 端点怎么办? A:只需提供一个 health check URL(返回 200 即可)。消息转发可以走其他方式,监控走 HTTP。

Q:消息内容会经过你们的服务器吗? A:消息转换(企微格式 ↔ 标准格式)走云端 API(www.hermesai.ltd),不存储明文。监控数据全部存在本地。也可以设置 converter.enabled: false 完全本地运行。

Q:不用你们监控,我只要消息转发行不行? A:可以。设置 monitor.enabled: false 降级为纯消息转发模式。

Q:bot_id 和 bot_secret 去哪找? A:企微管理后台 → 应用管理 → 智能机器人 → 点击应用名称 → 复制「机器人ID」和「Secret」。

Q:notify_chatid 怎么填? A:企微群里右键群名 → 复制群聊 ID(格式:wrkSFfCgAA...)。填到 config.yaml 的 monitor.notify_chatid

Q:端口 9527 被占用了怎么办? A:修改 config.yaml 的 connector.port 为其他端口(如 9528),重启即可。

Q:多个用户共用一个 Ops Center,Agent 数据会混在一起吗? A:不会。系统按 WorkBuddy/OpenClaw 的用户 ID 自动做多租户隔离,每个用户只能看到自己的 Agent。平台来源(WorkBuddy / OpenClaw)也会自动打标签,方便分类统计。

技术栈

  • 语言: Node.js 18+
  • WebSocket: ws 库(企微官方智能机器人协议)
  • 监控: 自建 AgentMonitor + NotifyEngine(无外部依赖)
  • 持久化: 本地 JSON 文件(StateStore)
  • 消息转换: 云端 API(www.hermesai.ltd)或本地降级

链接

  • 状态面板:http://localhost:9527
  • 健康检查:http://localhost:9527/health
  • 云端转换 API:https://www.hermesai.ltd/health
  • 问题反馈:GitHub Issues

🚀 ClawHub 首发:安装后 2 分钟,你的企微群里就能看到 AI Agent 的健康状态。

Usage Guidance
Install only if you are comfortable with WeCom message content and local agent inventory potentially going to www.hermesai.ltd. Before use, verify whether cloud conversion can truly be disabled in your runtime, keep the dashboard bound to localhost or behind authentication, avoid enabling P2P inbound exposure unless needed, review what the scanner will upload, and treat lifecycle/process controls as administrative functionality.
Capability Tags
cryptorequires-walletrequires-oauth-tokenrequires-sensitive-credentials
Capability Assessment
Purpose & Capability
The monitoring, dashboard, message forwarding, process restart, secrets scanning, and network guard features fit the stated ops-center purpose, but the capability set is high impact: full message frames and agent replies are posted to a cloud converter, local agent directories and process details can be inventoried and registered remotely, and lifecycle APIs can start, stop, restart, or kill managed processes.
Instruction Scope
The artifacts disclose some remote behavior, but the disclosure is inconsistent: docs say monitoring data is local and that converter.enabled:false makes it local-only, while the JavaScript converter always attempts cloud conversion and does not check that setting. Scanner registration fields and consent boundaries are also not clearly scoped before transmission.
Install Mechanism
package.json has no postinstall hook or hidden installer, so I did not find automatic execution at npm install time. The docs nevertheless claim installation triggers automatic scanning, and the standalone scanner defaults to registering inventory with https://www.hermesai.ltd/ops when run.
Credentials
Reading ~/.workbuddy and ~/.openclaw metadata is proportionate for auto-discovery, but also reading cached user IDs, MEMORY.md, and process command lines is broader than a user may expect. The dashboard binds to 127.0.0.1 by default, lowering exposure, but its POST APIs lack authentication and the optional P2P inbound server binds to 0.0.0.0.
Persistence & Privilege
Local JSON persistence of agent state, resources, tasks, and audit records is purpose-aligned but may retain sensitive inputs and outputs. The lifecycle manager can spawn and terminate configured processes, the scanner daemon can run recurring scans and heartbeats, and WebSocket/P2P clients automatically reconnect, so users need explicit control over long-running behavior.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install wecom-agent-ops-center
  3. After installation, invoke the skill by name or use /wecom-agent-ops-center
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v2.4.0
v2.4.0 Summary: Major upgrade introducing multi-tenant isolation, platform auto-detection, and zero-config agent registration. - Agent自动发现:自动扫描本地Agent,批量注册到Ops Center,无需手动配置 - 多租户隔离:按用户ID自动分区,不同用户/团队的数据完全隔离 - 平台自动识别:自动检测WorkBuddy或OpenClaw来源,Agent自动打标签 - 进程守护:Agent崩溃后自动重启,最多重试10次,支持指数退避 - 提升敏感信息安全:19种敏感信息规则,支持拦截、脱敏提示多种模式 - 网络守卫增强:域名/IP/CIDR白名单,异地非法外联实时阻断
Metadata
Slug wecom-agent-ops-center
Version 2.4.0
License MIT-0
All-time Installs 0
Active Installs 0
Total Versions 1
Frequently Asked Questions

What is 企微 Agent Ops Center?

在企微群实时监控AI Agent健康状态,自动告警、进程守护、敏感信息拦截和多租户隔离,支持WorkBuddy/OpenClaw。 It is an AI Agent Skill for Claude Code / OpenClaw, with 38 downloads so far.

How do I install 企微 Agent Ops Center?

Run "/install wecom-agent-ops-center" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is 企微 Agent Ops Center free?

Yes, 企微 Agent Ops Center is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does 企微 Agent Ops Center support?

企微 Agent Ops Center is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created 企微 Agent Ops Center?

It is built and maintained by mikeliang2000 (@mikeliang2000); the current version is v2.4.0.

💬 Comments