colbymchenry/codegraph

Pre-indexed code knowledge graph, auto syncs on code changes, for Claude Code, Codex, Gemini, Cursor, OpenCode, AntiGravity, Kiro, CoPilot, and Hermes Agent — fewer tokens, fewer tool calls, 100% local

What it solves

AI agents typically discover code structure by repeatedly reading files, using grep, and rebuilding call paths manually. This process is slow, token-intensive, and expensive. CodeGraph solves this by providing a pre-built semantic knowledge graph of every symbol, call edge, and dependency in a codebase, allowing agents to retrieve surgical context in a single call rather than crawling files one by one.

How it works

CodeGraph uses a Rust-powered kernel to perform full structural extraction and cross-file resolution across a wide variety of programming languages. It builds a local index (stored in .codegraph/) and provides an MCP (Model Context Protocol) server that AI agents can query. The system includes an auto-sync feature that watches for file changes and updates the graph in real-time, ensuring the index never becomes stale.

Who it’s for

Developers using AI coding agents such as Claude Code, Cursor, GitHub Copilot, Gemini, and others who want to reduce the number of tool calls, lower token costs, and speed up the response time of their agents when working with large or complex codebases.

Highlights

  • Broad Language Support: Supports a vast array of languages including TypeScript, Python, Go, Rust, Java, C#, PHP, Ruby, C/C++, Swift, Kotlin, and more.
  • 100% Local: The graph is built and stored locally on the user's machine.
  • Performance Gains: Benchmarks show significant reductions in tool calls (up to 88% fewer) and faster response times (up to 3.6x faster) compared to file-reading agents.
  • Surgical Context: Enables agents to identify call paths and the blast radius of changes without needing to read every file in a sequence.
  • Zero-Config Sync: Automatically updates the index as files are modified, added, or deleted.

Related

  • Project
    aovestdipaperino/tokensave

    tokensave is a Rust‑based local service that builds a semantic knowledge graph of a codebase. AI coding agents query this graph via MCP tools to get symbols, relationships and snippets in one call, dramatically cutting token usage and eliminating costly file‑scanning operations. It supports 50+ languages, 12+ agent integrations, optional per‑branch indexing, crash‑resilient sync, and fully offline operation.

  • Project
    Context-Engine-AI/Context-Engine

    A semantic code search and symbol intelligence system that provides AI coding assistants with deep codebase context, memory, and navigation tools via MCP.

  • Project
    justrach/codedb

    A high-performance code intelligence server for AI agents that provides structural indexing and fast symbol search to help agents understand large codebases.

  • Project
    Cranot/roam-code

    A local codebase intelligence layer that provides AI coding agents with pre-resolved graph facts and structural verification to reduce token usage and ensure code quality.

  • Project
    CodeBendKit/codeseek

    A code intelligence CLI tool that combines AST-based call graph analysis and hybrid semantic search to help developers and AI agents navigate large codebases.