Code Graph MCP by Sdsrss is a free MCP server for AI coding assistants. It provides a structured understanding of codebases, enabling AI agents to interact with code more effectively. Operating locally, this Rust-based tool ensures privacy with zero external runtime dependencies. Developers can connect various AI clients, enhancing their coding workflows with advanced analysis capabilities. Code Graph MCP acts as a high-performance code knowledge graph server, designed to improve the accuracy and speed of AI-driven development tasks.

What does Code Graph MCP connect to?

Code Graph MCP serves as an MCP server, connecting to a variety of AI agents and coding assistants. It supports clients such as Claude Code, Cursor, GitHub Copilot, Windsurf, Cline, OpenAI Codex CLI, and Google Gemini. This tool operates using the MCP protocol, which functions over stdio via JSON-RPC 2.0. Its primary role is to provide these AI clients with a structured understanding of codebases, acting as a high-performance code knowledge graph server.

Core features and tools

The Code Graph MCP server exposes specific callable tools, including GetCodeGraph, which delivers the complete code graph as JSON. Beyond this, it offers capabilities for semantic code search, call graph traversal, HTTP route tracing, dead code detection, and impact analysis. Within the Claude Code Plugin, these capabilities are directly accessible through slash commands like /understand, /trace, and /impact, allowing AI agents to perform detailed code inspections.

Who is it for?

Code Graph MCP is primarily for developers who use AI coding assistants. It helps these users gain deeper insights into their codebases and perform advanced analysis with AI support. The tool enhances the capabilities of AI agents by providing structured code knowledge. This is not the right fit if you do not utilize AI coding assistants in your development workflow.

How to get started

No credentials are required, as Code Graph MCP operates locally and privately. Setup involves configuring the AI client, for example, by editing the mcp.json file for GitHub Copilot in Visual Studio to connect to the local MCP server.

Use cases

• Enhanced Code Understanding: AI agents query structured code graphs for deep understanding of code relationships, enabling faster, more accurate suggestions.
• Advanced Code Analysis: AI agents perform semantic code searches, traverse call graphs, detect dead code, and conduct impact analysis for code changes.
• HTTP Route Tracing: The tool maps HTTP route paths to backend handler functions across frameworks like Express, Flask/FastAPI, Go net/http, and Rust axum.

Code Graph MCP by Sdsrss offers a private, local solution for enhancing AI coding assistant capabilities. Its Rust-based architecture provides a high-performance code knowledge graph server, supporting various AI clients with advanced analysis features. Developers can gain significant advantages in code understanding and analysis without external dependencies.