YaoApp/yao

每日信息看板 · 2026-03-03
开源项目
Category
github_search
Source
3
Score
2026-03-03T01:57:35Z
Published

AI 总结

Yao 发布开源自治智能体引擎,通过事件/邮件/定时触发与多智能体编排、内置 GraphRAG 和原生 MCP 支持,让智能体可主动自运行并在单一可执行文件内完成数据/API/UI运行,降低部署与工具集成成本。
#GitHub #repo #开源项目 #Yao #Autonomous Agents #MCP #GraphRAG #Agent #RAG

内容摘录

Yao — Build Autonomous Agents. Just Define the Role.

Yao is an open-source engine for autonomous agents — event-driven, proactive, and self-scheduling.

!Mission Control

**Quick Links:**

**🏠 Homepage:** https://yaoapps.com

**🚀 Quick Start:** https://yaoapps.com/docs/documentation/en-us/getting-started

**📚 Documentation:** https://yaoapps.com/docs

**✨ Why Yao?** https://yaoapps.com/docs/why-yao

**🤖 Yao Agents:** https://github.com/YaoAgents/awesome ( Preview )

---
What Makes Yao Different?

| Traditional AI Assistants | Yao Autonomous Agents |
| ----------------------------- | ------------------------------------- |
| Entry point: Chatbox | Entry point: Email, Events, Schedules |
| Passive: You ask, they answer | Proactive: They work autonomously |
| Role: Tool | Role: Team member |
The entry point is not a chatbox — it's email, events, and scheduled tasks.

---
Features
Autonomous Agent Framework

Build agents that work like real team members:
**Three Trigger Modes** — Clock (scheduled), Human (email/message), Event (webhook/database)
**Six-Phase Execution** — Inspiration → Goals → Tasks → Run → Deliver → Learn
**Multi-Agent Orchestration** — Agents delegate, collaborate, and compose dynamically
**Continuous Learning** — Agents accumulate experience in private knowledge bases
Native MCP Support

Integrate tools without writing adapters:
**Process Transport** — Map Yao processes directly to MCP tools
**External Servers** — Connect via SSE or STDIO
**Schema Mapping** — Declarative input/output schemas
Built-in GraphRAG
**Vector Search** — Embeddings with OpenAI/FastEmbed
**Knowledge Graph** — Entity-relationship retrieval
**Hybrid Search** — Combine vector similarity with graph traversal
Full-Stack Runtime

Everything in a single executable:
**All-in-One** — Data, API, Agent, UI in one engine
**TypeScript Support** — Built-in V8 engine
**Single Binary** — No Node.js, Python, or containers required
**Edge-Ready** — Runs on ARM64/x64 devices