Using Claude Code in WeChat, Just Open-Sourced This Skill on GitHub.

I had Claude Code read the source code of the ClawBot package.

It contains the entire protocol suite for a WeChat Bot:how HTTP long polling receives messages, how AES encryption is handled, the flow for QR code authentication, how CDN uploads/downloads media files, it’s all there.

No dependency on OpenClaw, just write your own following this protocol.

Any Agent can connect to WeChat.

I directly dictated to Claude Code and open-sourced a Skill in 10 minutes.

Then I started chatting with Claude Code on my computer from within WeChat.

Open Source Address: https://github.com/Wechat-ggGitHub/wechat-claude-code

OpenClaw’s ability to connect to WeChat relies on the protocol mentioned above.

What if I completely replace OpenClaw’s message handling logic and swap it with Claude Code?

A message comes from the mobile WeChat app, the local process receives it, directly throws it to Claude Code for processing, and sends the result back to WeChat.

<span leaf="" style="color: rgba(0, 0, 0, 0.9);font-size: 17px;font-family: mp-quote, 'PingFang SC', system-ui, -apple-system, BlinkMacSystem