vitali87/code-graph-rag

The ultimate RAG for your monorepo. Query, understand, and edit multi-language codebases with the power of AI and knowledge graphs

What it solves

Code-Graph-RAG addresses the difficulty of understanding and maintaining large, multi-language codebases. It allows developers to query, edit, and optimize code using natural language by grounding AI responses in a structural knowledge graph of the codebase rather than relying on simple text search.

How it works

The system consists of two primary components:

  1. Multi-language parser: Uses Tree-sitter to extract functions, classes, methods, and modules, storing them in a Memgraph knowledge graph. It enhances this with compiler-grade frontends (like libclang for C/C++ or go/types for Go) and dynamic runtime tracing to capture calls that static analysis cannot see.
  2. RAG system: An interactive CLI that translates natural language queries into Cypher queries to retrieve precise structural information from the graph, which is then used to ground AI-powered code editing and optimization.

Who it’s for

Developers and engineering teams working with complex monorepos or multi-language projects who need to efficienty navigate, document, and refactor their code.

Highlights

  • Multi-language support: Works across Python, TypeScript, Rust, Go, Java, C/C++, C#, and others under a unified schema.
  • AST-based editing: Performs surgical patching of code with diff previews before applying changes.
  • Runtime overlay: Merges eBPF profiles or test run traces into the graph to expose dynamic dispatch.
  • Dead code detection: Identifies unused code by walking call and reference edges from entry points.
  • MCP Server: Integrates with Claude Code and other Model Context Protocol clients.

Related

  • Project
    Jakedismo/codegraph-rust

    CodeGraph turns a codebase into a semantically searchable knowledge graph, allowing AI agents to reason about architecture and dependencies instead of just searching for text.

  • Project
    colbymchenry/codegraph

    CodeGraph builds a local, up‑to‑date knowledge graph of a codebase (30+ languages) and plugs into AI coding assistants (Claude, Cursor, Copilot, etc.). By providing agents with a single, precise payload of symbols and call‑paths, it cuts tool calls, token usage and latency—benchmarks show up to 44 % cost reduction and 88 % fewer tool calls.

  • Project
    FalkorDB/code-graph

    CodeGraph is a tool that transforms codebases into interactive knowledge graphs stored in FalkorDB, enabling developers to visualize dependencies and query code structure using LLM-powered GraphRAG.

  • Project
    DeusData/codebase-memory-mcp
  • Project
    forloopcodes/contextplus

    An MCP server that turns large codebases into searchable, hierarchical feature graphs using RAG, AST parsing, and spectral clustering for high-accuracy AI coding context.