@icesword760/openclaw-wechat

OpenClaw channel plugin for WeChat personal accounts via WeChatPadPro (iPad protocol).

Features

• Direct messages and group chats
• Keyword trigger prefix (e.g. @ai) to coexist with personal use
• AI reply prefix (e.g. 🤖) for visual distinction
• Per-peer session routing for independent conversations
• Quote reply support (extracts referenced message context)
• Image and file exchange (inbound download + outbound send)
• QR code login flow with terminal display
• DM pairing / allowlist / open access policies
• Multi-account support

Prerequisites

1. WeChatPadPro running and accessible (see WeChatPadPro docs)
2. A TOKEN_KEY generated via the WeChatPadPro admin API
3. OpenClaw installed (npm i -g openclaw)

Install

openclaw plugins install @icesword760/openclaw-wechat

Configure

openclaw config set channels.wechat.enabled true
openclaw config set channels.wechat.serverUrl http://localhost:8849
openclaw config set channels.wechat.token YOUR_TOKEN_KEY

Optional settings

# Only process messages starting with "@ai"
openclaw config set channels.wechat.triggerPrefix "@ai"

# Prefix AI replies with an emoji
openclaw config set channels.wechat.replyPrefix "🤖 "

# Independent conversation context per user
openclaw config set session.dmScope "per-peer"

# DM access policy: "pairing" (default), "allowlist", or "open"
openclaw config set channels.wechat.dmPolicy "pairing"

Full config example

{
  "channels": {
    "wechat": {
      "enabled": true,
      "serverUrl": "http://localhost:8849",
      "token": "YOUR_TOKEN_KEY",
      "triggerPrefix": "@ai",
      "replyPrefix": "🤖 ",
      "dmPolicy": "pairing",
      "debugMessages": false
    }
  },
  "session": {
    "dmScope": "per-peer"
  }
}

Usage

1. Start the OpenClaw gateway:

openclaw gateway

2. Scan the QR code displayed in your terminal with WeChat.

3. If using pairing policy, approve senders:

openclaw pairing approve wechat <CODE>

4. Send messages starting with your trigger prefix (e.g. @ai hello!).

File Exchange

The plugin supports bi-directional file exchange:

Receiving files from users:

• Images are injected into the AI's visual context (multimodal models can see them)
• Documents/files are downloaded locally and their paths are provided to the AI

Sending files to users:

• The AI uses MEDIA:/path/to/file in its reply
• Images (png/jpg/gif/webp) are sent as WeChat image messages
• Other files (PDF, Excel, zip, etc.) are uploaded to WeChat CDN and sent as file attachments

Architecture

User (WeChat) <-> WeChatPadPro (iPad Protocol) <-> This Plugin <-> OpenClaw Agent

• Inbound: HTTP polling via POST /message/HttpSyncMsg
• Outbound text: POST /message/SendTextMessage
• Outbound images: POST /message/SendImageNewMessage
• Outbound files: POST /message/UploadFileToCDN + POST /message/SendAppMessage

License

MIT