codexray
X-ray vision for your codebase — semantic knowledge graph & MCP server with 16 tools that saves AI coding agents 30%+ tokens. TF-IDF search, call graphs, impact analysis, dead code detection. Works with Claude Code, Cursor & Windsurf.
claude mcp add --transport stdio neuralrays-codexray npx -y codexray
How to use
CodeXRay CodexRay MCP server exposes a local, zero-config tool to access the CodeXRay semantic knowledge graph and 16 MCP tools for code intelligence. Once launched via npx codexray, you gain interactive access to the CodeXRay suite and can perform project-wide symbol queries, semantic searches, call graph explorations, and more without sending code to the cloud. Use the 16 MCP tools to quickly overview a project, drill into specific symbols, inspect dependencies, and analyze code hotspots. The server is designed to run locally on your machine, keeps data private, and indexes your codebase to provide fast, TF-IDF and graph-based search results for developers and AI agents alike. The provided CLI and MCP tools let you perform both exact name-based lookups and meaning-based searches, enabling you to locate and understand code paths, relationships, and potential issues efficiently.
How to install
Prerequisites:
- Node.js 18–24 installed on your machine
- npm or pnpm or yarn installed
- A supported shell (bash/zsh on macOS/Linux, PowerShell/cmd on Windows)
- Install Node.js if you don’t have it:
-
macOS/Linux via NodeSource (example): curl -fsSL https://deb.nodesource.com/setup_18.x | sudo -E bash - sudo apt-get install -y nodejs
-
Windows: download and install from https://nodejs.org/
- Install CodeXRay globally via npm (recommended for MCP usage):
npm install -g codexray
- Run the MCP server using npx (no local install required per Quick Start):
npx codexray
- If you prefer a fully local install (without npx each time), you can install npm package globally and run it:
npm install -g codexray
codexray start
- Optional: verify installation
codexray --version
Prerequisites recap:
- Node.js 18–24
- npm/pnpm/yarn for package management
- Basic shell access to run commands
Additional notes
Tips and common issues:
- The MCP server runs locally; no API keys are required. Ensure your firewall allows local network traffic if you’re integrating with external tools.
- When using npx, the first run will fetch the codexray package; this may take a moment on slow connections.
- If you encounter permission errors on macOS/Linux, consider using sudo for the global installation commands or adjust npm permissions.
- The 16 MCP tools include overview, context, search, semantic search, and various exploration and analysis tools. Use codexray serve to start the interactive server interface if you prefer a GUI/interactive flow.
- Data privacy: by default, CodeXRay indexes your project locally; ensure your project directory is excluded from any cloud-sync services if necessary.
- If indexing large codebases, allocate sufficient CPU/RAM to avoid timeouts during initial indexing.
Related MCP Servers
code-graph-rag
The ultimate RAG for your monorepo. Query, understand, and edit multi-language codebases with the power of AI and knowledge graphs
axon
Graph-powered code intelligence engine — indexes codebases into a knowledge graph, exposed via MCP tools for AI agents and a CLI for developers.
roam-code
Architectural intelligence layer for AI coding agents. Structural graph, architecture governance, multi-agent orchestration, vulnerability mapping. 139 commands, 101 MCP tools, 26 languages, 100% local.
mpm-vibe-coding
MPM is an MCP context-engineering layer for Vibe Coding, focused on three delivery bottlenecks: context loss, uncontrolled changes, and non-verifiable outcomes.
CodeMCP
Code intelligence for AI assistants - MCP server, CLI, and HTTP API with symbol navigation, impact analysis, and architecture mapping
fossil
The code quality toolkit for the agentic AI era. Find dead code, clones, and scaffolding across 15 languages. MCP server + CLI.