Code Graph RAG
A self-hosted AI code analysis tool searches and edits mixed-language repositories through knowledge graphs, with MIT licensing and Claude Code integration.
Code Graph RAG helps developers understand and change unfamiliar repositories, including monorepos that mix languages. It maps code relationships into a knowledge graph so questions and edits can use context across files.
- Codebase questions and search: Ask where a feature lives, retrieve functions by name or intent, and explore imports, dependencies, and calls.
- Reviewed edits: Target functions or exact code blocks with diff previews, or use ast-grep to search and replace code by structural pattern.
- Data flow and dead code: Follow values through assignments and calls to input or output operations, and identify functions unreachable from entry points.
- Runtime context: Add observed calls from test runs or production profiles to the graph to capture behavior that static analysis misses.
- Claude Code integration: An MCP server lets Claude Code and other MCP clients query, search, and edit the repository.
The self-hosted core is free and open source under the MIT license. It uses Memgraph for code relationships and Qdrant for semantic search. Full language support includes Python, TypeScript, JavaScript, Rust, Go, Java, C, C++, C#, PHP, Lua, and Dart; Ruby has structural support. Optional compiler frontends add more precise relationships where the relevant toolchain is available.
Your graph can run in your own environment, while cloud AI connections send requests to external providers. Enterprise services offer on-premise and air-gapped deployments, plus managed cloud hosting for the graph database and AI agent connection.
