Codegraph

About

Codegraph provides continuously updated code intelligence by building a function-level dependency graph across 11 programming languages using Tree-sitter. It ensures the graph is always current through a unique three-tier incremental change detection system, achieving sub-second rebuilds without relying on costly cloud API calls. Designed specifically for AI agents, Codegraph includes a 17-tool Model Context Protocol (MCP) server, enabling deep code context retrieval and powerful semantic search capabilities. Users can optionally enhance its features with their preferred LLM providers, maintaining control over their code and avoiding vendor lock-in.

Key Features

• Git diff impact analysis to identify blast radius of code changes
• Integrated 17-tool Model Context Protocol (MCP) server for AI agents
• Sub-second incremental code graph rebuilds with three-tier change detection
• 9 GitHub stars
• Local and LLM-enhanced semantic search for code functions
• Function-level dependency tracing across 11 programming languages

Use Cases

• Analyzing the impact of code changes before merging pull requests
• Understanding complex code dependencies and call graphs across a multi-language codebase
• Providing always-fresh and accurate code context to AI agents