GitNexus is an open-source code intelligence tool that converts a software repository into a searchable knowledge graph, making it easier for both developers and AI coding assistants to understand complex codebases. It parses source code using Tree-sitter, indexes symbols and relationships, and builds a graph of files, classes, functions, imports, and call chains.
Some of its main capabilities include:
-
Knowledge graph generation: Visualizes relationships between modules, functions, classes, and dependencies.
-
AI-friendly code search: Provides Graph RAG and hybrid search so AI agents can answer questions using the actual code structure rather than only text similarity.
-
MCP integration: Exposes the indexed repository through the Model Context Protocol (MCP), allowing tools like Cursor, Claude Code, and other compatible agents to query the codebase efficiently.
-
Privacy-focused: The web UI can run entirely in the browser, so code does not need to be uploaded to a remote server.
-
Multi-repository support: A single MCP server can manage multiple indexed repositories through a global registry.
Typical workflow:
-
Run
gitnexus analyze (or npx gitnexus analyze) in your repository.
-
GitNexus indexes the codebase and stores its graph.
-
Connect an MCP-compatible AI assistant.
-
Ask questions such as:
-
"Where is this function called?"
-
"What is the impact of changing this API?"
-
"Show the execution flow for this feature."
If you want to explore it further, you can visit:
If your goal is to use GitNexus with Cursor, Claude Code, Codex, or another AI coding assistant, I can also walk you through the setup step by step.