AI agents don't exist in a vacuum. Behind every agent is a stack of open-source infrastructure — frameworks, runtimes, observability tools, and developer utilities that make the agent economy possible.
Today we're expanding Fushu beyond agents, MCP servers, and tools to include the open-source repositories that power everything.
Why OSS Repos Matter
The agent ecosystem is built on open source. CrewAI orchestrates multi-agent teams. Ollama runs LLMs locally. browser-use lets agents interact with any website. Langfuse provides observability. These aren't just libraries — they're the foundation of the agent economy.
Top 10 OSS Repos on Fushu
- AutoGPT (170,000 stars) — An experimental open-source autonomous AI agent
- Langflow (140,000 stars) — Low-code app builder for RAG and multi-agent AI
- Ollama (115,000 stars) — Run LLMs locally with a single command
- browser-use (78,000 stars) — Make AI agents interact with any website
- RAGFlow (70,000 stars) — Open-source RAG engine for deep document understanding
- OpenHands (69,400 stars) — Open-source AI software engineer (formerly OpenDevin)
- Open WebUI (65,000 stars) — Self-hosted web UI for LLMs (formerly Ollama WebUI)
- n8n (62,000 stars) — Fair-code workflow automation platform
- Dify (60,000 stars) — Open-source LLM app development platform
- Open Interpreter (58,000 stars) — Natural language interface for your computer
What This Means
By indexing OSS repos alongside agents and tools, Fushu becomes the complete map of the AI agent ecosystem. Developers can discover not just agents, but the infrastructure to build them.