Repo GraphRAG
Leverage LightRAG and Tree-sitter to build a repository knowledge graph from code and documentation for Q&A and implementation planning.
소개
Repo GraphRAG is an MCP (Model Context Protocol) server designed to transform your codebase and documentation into an interactive knowledge graph. By leveraging LightRAG and Tree-sitter, it intelligently analyzes code and text documents within a repository to construct a comprehensive knowledge base. This powerful tool then facilitates intelligent Q&A, provides detailed implementation plans, and suggests concrete change steps, making it invaluable for understanding, modifying, and evolving complex software projects.
주요 기능
- Builds and incrementally updates a repository knowledge graph from code and text documents.
- Generates detailed implementation plans and concrete change steps for modification requests.
- Answers questions about the codebase based on the knowledge graph, optionally combined with vector search.
- Supports knowledge graph extraction for 12 programming languages including Python, C++, Java, and JavaScript.
- Features intelligent entity merging based on semantic similarity to consolidate code and documentation references.
- 0 GitHub stars
사용 사례
- Facilitating code comprehension and onboarding for new developers by providing a navigable knowledge graph.
- Generating structured implementation plans for new features or modifications within a project.
- Understanding complex codebases and documentation through interactive Q&A.