AstrBot is a self-hosted agent chatbot platform that connects large language models to instant-messaging apps. You run one AstrBot instance, attach it to platforms such as QQ, Telegram, Slack, Discord, Feishu, DingTalk, WeChat Work, and LINE, and configure the model providers, personas, knowledge base, and plugins from a web UI. It targets individual assistants, customer-service bots, automation assistants, and team knowledge bases inside existing chat workflows.
Conversations support multimodal input, agent tool use, MCP (Model Context Protocol) servers, skills, and automatic context compression. AstrBot can also hand conversations to external agent platforms such as Dify, Alibaba Cloud Bailian, and Coze. An agent sandbox isolates code and shell execution with session-level resource reuse, and a built-in web ChatUI includes the sandbox and web search.
Model support spans OpenAI-compatible services, Anthropic, Google Gemini, DeepSeek, Moonshot, Zhipu, and self-hosted Ollama or LM Studio, plus separate speech-to-text (Whisper, SenseVoice) and text-to-speech (OpenAI, Gemini, Edge TTS, Azure, FishAudio, GPT-SoVITS, and others) services. It requires Python 3.12 or later and can be deployed with uv, Docker, a desktop app, a launcher, or panel installers such as 1Panel and CasaOS.
Features
- Messaging platforms: QQ, OneBot v11, Telegram, WeCom, WeChat Official Accounts, Feishu, DingTalk, Slack, Discord, LINE, Satori, KOOK, Misskey, and Mattermost, with community adapters for Matrix, Rocket.Chat, and VoceChat
- Model services: OpenAI-compatible APIs, Anthropic, Gemini, Moonshot, Zhipu, DeepSeek, Ollama, LM Studio, ModelScope, and several API gateways
- Agent capabilities: multimodal chat, tool use, MCP servers, skills, knowledge base, persona settings, and automatic context compression
- LLMOps integrations: Dify, Alibaba Cloud Bailian applications, and Coze
- Plugin marketplace: plugins installed from the web UI with one click
- Agent sandbox: isolated execution of code and shell calls with session-level resource reuse
- Web UI and ChatUI: browser-based administration plus a chat interface with sandbox and web search
- Speech services: speech-to-text through Whisper, SenseVoice, and MiMo Omni; text-to-speech through OpenAI, Gemini, Edge TTS, Azure, Minimax, Volcano Engine, and others
- Deployment options: uv tool install, Docker Compose, desktop app, launcher, BT-Panel, 1Panel, and CasaOS
- Internationalisation: localised UI and documentation
