Knowledge RAG

•
306
•

Enables Claude Code to search and retrieve context from personal documents using a local, hybrid RAG system.

Knowledge RAG is a 100% local, privacy-first Retrieval-Augmented Generation (RAG) system designed for seamless integration with Claude Code via the Model Context Protocol (MCP). It allows users to augment Claude's capabilities by providing relevant context from their personal documents, including PDFs, Markdown, text, and code files. The system features a sophisticated hybrid search mechanism, combining semantic understanding with precise BM25 keyword matching using Reciprocal Rank Fusion (RRF), enhanced by smart keyword routing. All data processing occurs locally, ensuring privacy, and parallel embedding generation with ChromaDB provides fast and accurate context retrieval for AI models.

主要功能

01Word-boundary aware Keyword Routing for domain-specific queries
02Hybrid Search (Semantic + BM25 with RRF fusion)
03Native MCP Integration for Claude Code tools
043 GitHub stars
05Local Embeddings via Ollama (nomic-embed-text) with parallel processing
06Multi-Format Parser (PDF, Markdown, TXT, Python, JSON)

使用案例

01Augmenting Claude Code's knowledge base with proprietary or local documentation.
02Answering specific technical questions or conceptual queries by searching a custom knowledge repository.
03Retrieving precise information such as CVEs, tool names, or code snippets from personal files.