knowledge-rag

Knowledge RAG is a fully local, zero-cloud document retrieval system that lets Claude Code search your PDFs, markdown, code, and notebooks natively. It indexes up to 1800+ files in under 3 minutes, using hybrid search (BM25 + semantic vectors + cross-encoder reranking) via 12 MCP tools. Everything runs locally via ONNX, no Docker, Ollama, or API keys required.

knowledge-rag sits in the RAG / Knowledge Base category, so it is usually better judged by workflow fit, deployment style, and team needs than by isolated features alone. Based on the available metadata, it leans most clearly toward Hybrid Search with BM25 + semantic vectors + cross-encoder reranking. A representative use case is Personal document retrieval and knowledge base for developers. The listed license is MIT. GitHub stars are currently 300. Relevant tags include bm25, chromadb.