Yggabhigyanpatwari/GitNexus
↗ GitHubGitNexus: The Zero-Server Code Intelligence Engine - GitNexus is a client-side knowledge graph creator that runs entirely in your browser. Drop in a GitHub repo or ZIP file, and get an interactive knowledge graph wit a built in Graph RAG Agent. Perfect for code exploration
21,135
Stars
2,411
Forks
76
Watchers
220
Open Issues
Safety Rating A
No security concerns identified. The project is a well-documented, high-star open-source developer tool with a clear legitimate purpose. It explicitly emphasizes local-first privacy (no code uploaded, no network calls from CLI). No hardcoded secrets, malicious patterns, suspicious dependencies, or prompt injection attempts were found in the available content. The README even includes a proactive disclaimer warning users about fraudulent cryptocurrency tokens using the GitNexus name, which is a responsible signal from the maintainers.
ℹAI-assisted review, not a professional security audit.
AI Analysis
GitNexus is a client-side code intelligence engine that indexes any codebase into a knowledge graph and exposes it via an MCP server, CLI, and browser-based UI. It parses source files using Tree-sitter ASTs, resolves cross-file dependencies, detects functional communities via Leiden clustering, traces execution flows, and builds hybrid BM25+semantic search indexes. AI agents (Cursor, Claude Code, Codex, Windsurf) connect via MCP to get precomputed blast-radius analysis, 360-degree symbol context, process-grouped search, multi-file rename, and git-diff impact analysis. A fully client-side web UI running on WebAssembly allows zero-install graph exploration and Graph RAG chat directly in the browser. Supports 14+ programming languages.
Use Cases
- Giving AI coding assistants (Cursor, Claude Code, Codex) deep architectural awareness of a codebase via MCP to prevent breaking changes
- Visualizing and exploring code structure as an interactive knowledge graph in the browser without any server
- Performing blast-radius analysis before committing changes to understand downstream impact
- Generating repository wikis and architecture documentation from indexed knowledge graphs
- Enabling smaller LLMs to reason effectively about large codebases by precomputing relational structure
Tags
Security Findings (2)
No hardcoded secrets or API keys are visible in the README or repository metadata. The README explicitly states API keys are stored in localStorage only for the web UI, and the CLI makes no network calls.
No prompt injection attempts detected. Repository content is straightforward technical documentation with no embedded instructions attempting to manipulate AI analysis.
Project Connections
code-review-graph
→Both projects parse codebases into knowledge graphs using Tree-sitter, expose them via MCP for AI coding assistants, perform blast-radius analysis, support semantic/vector search, community detection, and execution flow tracing. They are direct functional alternatives solving the same core problem with similar architecture.
clawvault
→ClawVault provides persistent markdown-native memory and knowledge graph context for AI agents across sessions, while GitNexus provides structural code intelligence. Together they give AI agents both project memory and codebase architecture awareness.
skyclaw
→Skyclaw is an autonomous AI agent runtime that can consume MCP servers. GitNexus exposes codebase intelligence via MCP, making it a natural context provider for Skyclaw agents operating on codebases.
kavach
→Kavach monitors and restrains AI agent file system operations for safety, while GitNexus gives AI agents deep architectural awareness to make better-informed edits. The two address complementary aspects of safe AI-assisted development.
MiroFish
→MiroFish also uses GraphRAG for knowledge graph construction and multi-agent reasoning over structured graph data, though applied to simulation rather than code. Both represent different application domains of graph-augmented retrieval for AI agents.