Bridge any AI agent backend to WeChat using the weixin-agent-sdk framework with simple Agent interface, login, and message loop.

weixin agent sdk Skill by [ara.so](https://ara.so) — Daily 2026 Skills collection. weixin agent sdk is a TypeScript framework that bridges any AI backend to WeChat (微信) via the Clawbot channel. It uses long polling to receive messages — no public server required — and exposes a minimal Agent interface so you can plug in OpenAI, Claude, or any custom logic in minutes. Installation Node.js = 22 required. Quick Start 1. Login (scan QR code once) 2. Implement the Agent interface Core API Agent Interface login() Triggers QR code scan and persists session to ~/.openclaw/ . Only needs to run once. start(agent) Starts the message loop. Blocks until process exits. Automatically reconnects on session expiry. Common Patterns Multi turn Conversation with History OpenAI Agent (Full Example) Send Image Response Send File Response ACP (Agent Client Protocol) Integration If you have an ACP compatible agent (Claude Code, Codex, kimi cli, etc.), use the weixin acp package — no code needed. weixin acp launches your agent as a subprocess and communicates via JSON RPC over stdio. Environment Variables (OpenAI Example) Variable Required Description OPENAI API KEY Yes OpenAI API key OPENAI BASE URL No Custom API base URL (OpenAI compatible services) OPENAI MODEL No Model name, default gpt 5.4 SYSTEM PROMPT No System prompt for the assistant Built in Slash Commands Send these in WeChat chat to control the bot: Command Description /echo <message Echoes back directly (bypasses Agent), shows channel latency /toggle debug Toggles debug mode — appends full latency stats to each reply Supported Message Types Incoming (WeChat → Agent) Type media.type Notes Text — Plain text in request.text Image image Downloaded & decrypted, filePath = local file Voice audio SILK auto converted to WAV (requires silk wasm ) Video video Downloaded & decrypted File file Downloaded & decrypted, original filename preserved Quoted message — Quoted text appended to request.text , quoted media as media Voice to text — WeChat transcription delivered as request.text Outgoing (Agent → WeChat) Type Usage Text Return { text: "..." } Image Return { media: { type: "image", url: "..." } } Video Return { media: { type: "video", url: "..." } } File Return { media: { type: "file", url: "...", fileName: "..." } } Text + Media Return both text and media together Remote image Set url to an HTTPS link — SDK auto downloads and uploads to WeChat CDN Monorepo / pnpm Setup Troubleshooting Session expired ( errcode 14 ) The SDK automatically enters a 1 hour cooldown and then reconnects. No manual intervention needed. Audio not converting from SILK to WAV Install the optional dependency: npm install silk wasm Bot not receiving messages after restart State is persisted in ~/.openclaw/get updates buf . The bot resumes from the last position automatically. Remote image URL not sending Ensure the URL is HTTPS and publicly accessible. The SDK downloads it before uploading to WeChat CDN. login() QR code not appearing Ensure your terminal supports rendering QR codes, or check ~/.openclaw/ for the raw QR data.