AI That Lives a Life of Its Own
Tale is a multi-agent AI conversation system with autonomous daily planning. The AI character schedules its own day — waking up, working, resting, playing — like a real person. Multi-platform support and a full-featured WebUI dashboard included.
- Multi-Agent Architecture — ChatLLM (conversation) + PlanLLM (scheduling) + ToolLLM (tool calling)
- Autonomous Daily Planning — AI generates 24h schedules simulating real-life routines
- Tool Calling — Browser, search, weather, calculator; extensible via plugins
- WebUI Dashboard — Real-time chat, config editor, log viewer, adapter management
- Multi-Platform — QQ / WeChat PC (incl. Moments) / WebSocket, hot-pluggable adapters
- Plugin System — 6 extension points, auto-discovery via manifest.json
中文版 README 请见 docs/README.zh.md
git clone https://github.com/Qixuan112/Tale-AI.git
cd Tale-AI
pip install -r requirements.txt
python main.py # Start core + WebUI → http://127.0.0.1:32456On first run, config files are auto-created in data/config/. Edit services.yaml with your API keys and start chatting.
Configuration Guide: See docs/config-guide.md for detailed instructions — YAML vs
.env, WebUI setup, and common pitfalls.
Security: WebUI binds to localhost by default. Always add authentication behind a reverse proxy before exposing it.
User Input → ChatLLM (intent) → PlanLLM (plan) / ToolLLM (tools)
↕
EventBus ← Plugin System ← Adapters (QQ/WeChat/WS)
↕
WebUI (Flask) → Dashboard · Chat · Config · Logs
| Component | License | Description |
|---|---|---|
wxauto (core/adapter/src/wechat_pc/wxauto/) |
Apache 2.0 | WeChat PC automation library |
| UIAutomation (bundled in wxauto) | Apache 2.0 | Windows UI automation wrapper |
GNU AGPL v3 — Free to use, modify, and distribute. Network service providers must disclose modified source code.