What it does

ChatGPT-weBot turns a logged-in Windows WeChat client into a programmable bot. It intercepts messages via DLL injection, then routes them to OpenAI’s API (GPT-3.5+) or a local Stable Diffusion instance. You set wake keywords for DMs and group chats, configure a system character, and the bot replies in-thread—including @-mentions in groups. It also handles conversation memory, rollbacks, regeneration, and a “conclusion” feature that compresses history to save tokens.

The interesting bit

The project doesn’t fake a WeChat web session or use unofficial protocols. It literally injects wxinject.dll into the running WeChat.exe process and monitors messages from inside the official client. The README is oddly proud that you’ll “never get banned by using official WeChat execution”—a claim worth side-eyeing, since DLL injection is hardly what Tencent’s ToS envisions as “official.”

Key highlights

• Requires WeChat exactly version 3.9.5.81; newer versions must be downgraded or installed in parallel
• Supports dual WeChat instances with a provided .bat script and two separate installs
• Multithreaded Stable Diffusion generation with (negative) prompts, English only
• “WebChatGPT” function for web-aware queries
• Auto-recovery from errors without manual service restart
• Character system via sys_character.json for roleplay modes

Caveats

• Windows-only; needs Python 3.7+ and administrator privileges to run the injector
• The “never get banned” claim is dubious—DLL injection is detectable and violates most platforms’ terms
• Last meaningful update was October 2023; pinned to an increasingly old WeChat version
• Stable Diffusion integration is English-prompt only

Verdict

Worth a look if you need a self-hosted WeChat bot and can tolerate the DLL-injection dance on a dedicated Windows box. Skip it if you wanted something cloud-native, cross-platform, or that won’t break when WeChat auto-updates.