Indexes markdown collections into a searchable BM25 index and exposes them as Model Context Protocol (MCP) tools.
Sponsored
The kb-mcp tool serves as a versatile server and command-line interface for managing and querying markdown-based knowledge bases. It efficiently indexes your markdown document collections into a robust BM25 search engine, offering an optional hybrid BM25 + vector search for superior semantic understanding. Configured entirely through flexible RON files, it allows for easy definition of diverse knowledge base structures and integrates seamlessly with the Model Context Protocol (MCP), providing powerful document retrieval and management tools for AI agents and other systems.
Key Features
010 GitHub stars
02BM25 keyword search for markdown collections
03Optional hybrid BM25 + vector semantic search with RRF fusion
04CLI for direct interaction with knowledge bases
05Flexible configuration via RON files for collections and sections
06MCP server integration exposing tools like search, get_document, and kb_write
Use Cases
01Indexing and searching project documentation, notes, or internal wikis
02Providing contextual knowledge and memory retrieval for AI agents
03Managing multiple distinct markdown knowledge collections within a single system
