Akon Labs · 2026-04-24 · major
GitNexus — Code Intelligence Graph with MCP for AI Editors
Code intelligence platform that maps codebases into dependency graphs and exposes 16 MCP tools to AI editors (Claude Code, Cursor, Codex, Windsurf). Trending on GitHub with 29.6k stars and +667 added today.
A code knowledge graph that gives AI editors structural awareness of your entire codebase via 16 MCP tools.
What is it?
GitNexus indexes your codebase into a graph database, mapping every function, class, import, and call chain across 14+ languages. It then exposes this graph as an MCP server, so AI editors like Claude Code, Cursor, and Codex can answer structural questions — what depends on this function, what would break if I rename this type — without re-scanning files on every query.
How does it work?
A CLI or browser-based web UI builds the initial index. The resulting graph is served via a local MCP server process. AI editors connect and gain access to 16 tools: impact analysis (all callers/dependents of a symbol), hybrid search (fuzzy + semantic over the graph), 360-degree symbol views, multi-file rename with dependency tracking, raw Cypher queries, and pre-commit change analysis. Indexes update incrementally on file changes.
Why does it matter?
AI coding agents fail on large codebases because they lack structural memory — they either ignore unknown files or hallucinate dependencies. GitNexus gives agents a precomputed relational map they can query cheaply, reducing hallucinated refactors and broken imports. Works with Claude Code, Cursor, Codex, and Windsurf without changing the agent's core architecture.
Who is it for?
Engineering teams using AI editors on large or unfamiliar codebases
Try it
npx gitnexus index && npx gitnexus mcp start