Integrates local codebases and documentation into a knowledge base server for Model Context Protocol (MCP) clients.
Sponsored
Knowledge Base is a local knowledge base server designed to augment Model Context Protocol (MCP) clients like Codex and Gemini CLI with project-specific context. It efficiently indexes your codebase and documentation, extracting relevant information and generating embeddings locally without external services. By combining full-text and semantic vector search, it provides context-aware results, enabling AI agents to better understand and interact with your proprietary code and documentation.
Key Features
01Hybrid search fusing full-text and semantic vector search via Reciprocal Rank Fusion
02Context-aware parsing of source code and Markdown with trust levels
03Autonomous in-process embedding generation using built-in ML models
040 GitHub stars
05Incremental repository indexing with delta-sync for efficiency
Use Cases
01Enhancing AI coding assistants with project-specific context
02Providing a local, offline knowledge base for proprietary code
03Enabling AI agents to perform context-aware searches within a codebase
