通知与机器人集成¶

InnoClaw 通过机器人适配器支持与消息平台的集成。这使用户可以直接从聊天应用程序与其工作空间进行交互。

支持的渠道¶

平台	状态码	Webhook 端点
飞书	已支持	`/api/bot/feishu`
企业微信	已支持	`/api/bot/wechat`

架构¶

        sequenceDiagram
    participant User
    participant Platform as Feishu / WeChat
    participant Webhook as InnoClaw Webhook
    participant Processor as Bot Processor
    participant AI as AI Provider

    User->>Platform: Send message
    Platform->>Webhook: POST webhook event
    Webhook->>Processor: Parse & validate
    Processor->>AI: Process with RAG
    AI->>Processor: AI response
    Processor->>Platform: Reply message
    Platform->>User: Show response

飞书机器人¶

设置¶

在飞书开放平台创建飞书应用
配置以下环境变量：

FEISHU_BOT_ENABLED=true
FEISHU_APP_ID=cli_your_app_id
FEISHU_APP_SECRET=your_app_secret
FEISHU_VERIFICATION_TOKEN=your_verification_token
FEISHU_ENCRYPT_KEY=your_encrypt_key
FEISHU_PUSH_SECRET=your_push_secret

在飞书应用配置中设置 Webhook URL：
```
https://your-domain.com/api/bot/feishu
```

Connection Modes¶

Feishu supports two connection modes:

模式	描述
HTTP Webhook	Feishu sends HTTP POST events to your webhook URL. Requires public URL.
WebSocket	Long-lived connection using `@larksuiteoapi/node-sdk` WSClient. Started automatically in `instrumentation.ts`. No public URL needed.

功能特性¶

接收并回复文本消息
Download and process file attachments (text files < 100 KB)
Audio message transcription
Interactive card responses with real-time agent progress tracking
Agent mode integration (full tool access)
支持加密事件负载
自动 Webhook 验证

Bot Commands¶

Users can interact with the bot using slash commands:

Command	Parameters	描述
`/workspace`	`<path>` (optional)	Bind a workspace directory, or show the current binding
`/mode`	`<agent\|plan\|ask>`	Switch the agent mode for this chat
`/status`	—	Show chat state (workspace, mode, history count, chat ID)
`/clear`	—	Clear conversation history (preserves workspace and mode)
`/help`	—	Show available commands and modes

Agent Modes in Feishu¶

模式	Tools Available	描述
agent	All tools (bash, readFile, writeFile, grep, kubectl, etc.)	Full autonomous execution
plan	readFile, listDirectory, grep	Read-only analysis and planning
ask	readFile, listDirectory, grep	Simple question answering

Interactive Cards¶

The Feishu bot uses interactive cards for rich progress display:

Progress Card — Shows real-time agent execution progress with tool call status
Final Card — Displays completed execution with tool call summary and response text
Error Card — Shows agent execution error with details
Command Response Card — Used for slash command responses

Push API¶

The Push API enables sending messages from the web application to Feishu chats:

POST /api/bot/feishu/push
Authorization: Bearer <FEISHU_PUSH_SECRET>

Request Body:

{
  "chatId": "oc_xxxxxxxxxxxx",
  "title": "Agent Message",
  "content": "Message content here",
  "type": "card"
}

This enables bidirectional communication between the web UI and Feishu. The chatId can be obtained via the /status command.

事件处理¶

飞书机器人处理以下事件类型：

事件	描述
`url_verification`	初始 Webhook URL 验证
`im.message.receive_v1`	来自用户的传入消息

企业微信机器人¶

设置¶

在企业微信管理后台创建应用
配置以下环境变量：

WECHAT_BOT_ENABLED=true
WECHAT_CORP_ID=your_corp_id
WECHAT_CORP_SECRET=your_corp_secret
WECHAT_TOKEN=your_token
WECHAT_ENCODING_AES_KEY=your_aes_key
WECHAT_AGENT_ID=your_agent_id

在企业微信应用配置中设置 Webhook URL：
```
https://your-domain.com/api/bot/wechat
```

功能特性¶

接收并回复文本消息
支持明文和加密两种消息模式
Webhook 签名验证
自动访问令牌管理

验证模式¶

企业微信支持两种 Webhook 验证模式：

模式	使用的参数
明文	`msg_signature`、`timestamp`、`nonce` 配合 SHA-1 签名
加密	AES-256-CBC 加密消息体

通用配置¶

工作空间绑定¶

机器人集成使用已配置工作空间的 RAG 流水线处理消息。机器人处理器将传入消息路由到相应的工作空间和 AI 提供商。

For Feishu, use the /workspace <path> command to bind a workspace. New chats are automatically bound to the first directory in WORKSPACE_ROOTS.

安全性¶

所有 Webhook 请求都通过平台特定的签名/令牌验证进行校验
两个平台均支持加密消息模式
API 密钥和密钥仅存储在服务器端，永远不会暴露给客户端

通知负载格式¶

机器人响应遵循各平台的原生消息格式：

Feishu (Interactive Card):

{
  "msg_type": "interactive",
  "card": {
    "header": { "title": { "tag": "plain_text", "content": "Agent Response" } },
    "elements": [{ "tag": "markdown", "content": "AI response content" }]
  }
}

Feishu (Text):

{
  "msg_type": "text",
  "content": {
    "text": "AI response based on your workspace files"
  }
}

企业微信：

<xml>
  <MsgType>text</MsgType>
  <Content>AI response based on your workspace files</Content>
</xml>