abhigyanpatwari/GitNexus

What it solves

GitNexus is a context engine for enterprise codebases that prevents AI coding agents from making breaking changes due to a lack of architectural awareness. It solves the problem of AI agents missing critical dependencies, breaking call chains, or shipping "blind edits" because they lack a deep understanding of the codebase's structure.

How it works

GitNexus indexes a codebase into a knowledge graph that tracks every dependency, call chain, cluster, and execution flow. It then exposes this graph through a Model Context Protocol (MCP) server, providing AI agents with specialized tools for impact analysis, symbol tracing, and structural checks. It precomputes relational intelligence (clustering and tracing) at index time, allowing agents to retrieve complete architectural context in a single query rather than through multiple iterative search steps.

Who it’s for

It is designed for developers using AI-powered editors and agents such as Cursor, Claude Code, Windsurf, Codex, and Antigravity who need reliable, graph-backed architectural clarity for large-scale enterprise repositories.

Highlights

• 17 MCP Tools: Includes tools for blast radius analysis (impact), shortest path tracing between symbols (trace), and API route mapping (route_map).
• Bespoke Agent Skills: Automatically installs skills into .claude/skills/ or .agents/skills/ for tasks like debugging, refactoring, and taint analysis.
• Precomputed Intelligence: Uses Leiden community detection to identify functional areas and generate repo-specific skills.
• Deep Editor Integration: Supports hooks for Claude Code, Cursor, and Codex to enrich searches and prompt for re-indexing after commits.
• Hybrid Search: Combines BM25, semantic search, and Reciprocal Rank Fusion (RRF) for querying the graph.