finds.dev

zhayujie/bot-on-anything

A large model-based chatbot builder that can quickly integrate AI models (including ChatGPT, Claude, Gemini) into various software applications (such as Telegram, Gmail, Slack, and websites).

View on GitHubHomepage ↗

★ 4,199PythonMITupdated Jan 2025

// the take

A plugin-style chatbot connector that wires LLM backends (OpenAI, LinkAI) to messaging channels (Telegram, Slack, WeChat, Gmail, etc.) via a single config file. The core abstraction is a bridge/channel/model split that lets you swap either side without touching the other. Aimed at developers who want a self-hosted chatbot on a specific platform without writing the glue code themselves.

The channel/model separation is genuinely clean — adding a new platform means implementing one channel class, not touching the core. Config-driven switching between models and channels is useful for experimentation. The WeChat async workaround (asynchronous fetch + customer service API push to dodge the 5-second timeout) is a real solution to a real problem. Dockerfile and start/shutdown scripts mean you can deploy this without inventing your own process management.

Claude and Gemini are listed as supported in the README but checked off as incomplete in the model list — the two most relevant alternatives to OpenAI are effectively missing. Last push was January 2025 and the repo looks stalled; OpenAI SDK versions pinned in the README are already outdated. The QQ integration depends on go-cqhttp, which was abandoned by its maintainer due to legal pressure from Tencent, so that channel is dead in practice. No conversation persistence beyond in-memory history — restart the process and everyone loses context.

We read the README, the directory tree and a key source file in June 2026. Nobody pays to be listed. How we judge →

// at a glance

Stars
4,199
Forks
908
Language
Python
License
MIT
Last commit
3 Jan 2025

// nearby

If bot-on-anything isn't quite it.

All repos →