What it does AstrBot is an open-source Python framework that turns LLMs into chatbots across mainstream messaging platforms. It handles the plumbing—message routing, persona management, context compression, multimodal I/O, and plugin loading—so you can focus on bot behavior rather than reverse-engineering QQ or WeChat protocols.

The interesting bit The project treats IM platforms as interchangeable backends. It supports 15+ platforms officially (including QQ, Telegram, Slack, Discord, Feishu, DingTalk, and LINE) and offers a plugin marketplace with 1000+ community extensions. The “Agent Sandbox” is a notable addition: an isolated execution environment for code and shell calls, which is rarer in chatbot frameworks than it should be.

Key highlights

• Broad platform coverage: Official adapters for QQ, WeChat Work/Official Accounts, Telegram, Slack, Discord, Feishu, DingTalk, LINE, KOOK, Misskey, Mattermost, and more.
• LLM-agnostic: Works with OpenAI, Anthropic, Gemini, DeepSeek, Ollama, LM Studio, plus LLMOps platforms like Dify, Coze, and Alibaba Bailian.
• Voice pipeline: Built-in STT (Whisper, SenseVoice) and TTS (Edge, Azure, FishAudio, GPT-Sovits, and several Chinese cloud providers).
• Deployment flexibility: One-liner uv install, Docker, desktop app, AUR, Replit, RainYun cloud, and panel integrations (BT-Panel, 1Panel, CasaOS).
• Plugin ecosystem: Claims 1000+ plugins via marketplace; dynamic badge suggests active growth.

Caveats

• macOS friction: First uv launch triggers security checks, adding 10–20s startup delay.
• Upgrade path split: uv deployments cannot self-update through the WebUI; you must return to the CLI.
• WhatsApp: Listed as “Coming Soon” despite being in the official platform table.

Verdict Worth a look if you need a single backend for multi-platform chatbots, especially in Chinese IM ecosystems (QQ, WeChat, Feishu) where protocol support is often scarce. Skip if you want a minimal, single-platform bot—this is a chassis, not a sports car.