abhigyanpatwari/gitnexus

GitNexus is a code intelligence tool that parses any GitHub repository or ZIP file and builds a knowledge graph representing everything inside the codebase, every function, class, file, dependency, and the relationships between them. Think of a knowledge graph as a map of your code: instead of just reading files line by line, it shows you which functions call which other functions, which modules depend on which others, and how the entire codebase is architecturally connected. The main use case is giving AI coding assistants a deeper understanding of your project before they suggest changes. When tools like Claude Code, Cursor, or GitHub Copilot only read individual files, they miss the big picture, they might suggest a change that breaks a dependency somewhere else or misses a critical call chain. GitNexus solves this by providing those AI agents with graph-aware context through an MCP (Model Context Protocol) server, a standard way for AI agents to query external data sources. There are two ways to use it. The CLI mode indexes a repository locally, runs an MCP server, and integrates with your editor so AI agents can query the graph at any time. This is intended for daily development work. The web UI mode requires no installation, you visit gitnexus.vercel.app, drop in a repository, and explore it interactively in the browser with a built-in graph visualization and a chat interface for asking questions about the code. You would reach for GitNexus when you are working on a large or unfamiliar codebase and want your AI coding assistant to understand the full architecture before it starts suggesting edits, or when you want to explore an open-source project and understand how the pieces fit together. The tech stack is TypeScript, using Tree-sitter for parsing code into a syntax tree and LadybugDB for storing the knowledge graph. It runs in Node.js via npm and in the browser via WebAssembly.