Knowledge Rag is a local, privacy-first document search tool designed for developers and technical professionals. It enables lightning-fast, hybrid search (BM25, semantic vectors, cross-encoder reranking) across a wide variety of file types (PDFs, markdown, code, notebooks), with all processing and storage running locally on your machine—no cloud or data outflow. Supporting 13 Model Context Protocol (MCP) tools and running efficiently via ONNX (including optional GPU acceleration), it ensures robust, scalable, and fully offline retrieval-augmented generation for technical documentation, codebases, and personal knowledge management.
Visit Knowledge Rag's official website for product details and getting started.
Join the community discussions, report issues, and contribute to the Knowledge Rag project on GitHub.