One binary. Zero dependencies. 29 languages. 32 MCP tools. Local code intelligence for AI coding agents.

Documentation · npm · Issues

Any MCP-compatible AI agent, IDE, or CLI tool connects out of the box. cortex install auto-detects your setup and writes the config.

An agent asking "what calls processOrder" gets a 200-token graph result instead of burning 20,000 tokens reading files. That's 100x fewer tokens on a single structural question.

Cortex indexes your repository into a SQLite call graph and exposes it over the Model Context Protocol. 127 files indexed in 535ms. Incremental re-index in 13ms. Handles 100K to 1M+ line repositories.

npx @1337xcode/cortex install

curl -fsSL https://raw.githubusercontent.com/1337Xcode/cortex/main/install.sh | sh

git clone https://github.com/1337Xcode/cortex.git && cd cortex
cargo build --release
cp target/release/cortex ~/.local/bin/

cortex index    # parse your repo, build the graph
cortex install  # detect Claude Code / Cursor, write MCP config
cortex serve    # start the MCP server

Your agent queries the graph instead of reading raw files.

cortex impact UserService.getUser       # blast radius: what breaks if I change this?
cortex explain DatabasePool.acquire     # offline function explanation, zero LLM calls
cortex security report                  # taint flows, OWASP patterns, dependency vulns
cortex diff main feature-branch         # call graph diff between branches
cortex viz --export graph.html          # interactive 3D graph in a standalone HTML file
cortex ci --fail-on-taint               # CI quality gate with exit codes
cortex hotspots --months 6              # high-churn risky code ranked by risk score
cortex coverage --lcov coverage.lcov    # untested functions ranked by caller count
cortex modules                          # Leiden community detection module boundaries
cortex federate add ../auth-service     # query across repos with unified graph

• 29 languages parsed via tree-sitter
• 32 MCP tools exposed over stdio, or 5 in smart mode (the ask meta-tool routes internally)
• Sub-second incremental re-indexing via native OS file watcher (inotify, FSEvents, ReadDirectoryChangesW)
• Taint flow analysis, OWASP Top 10 detection, SBOM generation all running locally, no cloud
• Cross-session memory that marks observations stale when linked code changes
• Multi-repo federation for querying across repositories with a unified graph
• Hybrid search combining FTS5 BM25 with optional local ONNX vector search
• CI quality gates with configurable thresholds and exit codes
• 3D interactive graph visualization exportable to standalone HTML
• Portable JSON bundle (cortex.json) for team sharing via git. cortex bundle export serializes the full graph as a portable JSON file. Commit it to git so your team skips the index step entirely.
• Works offline. No cloud, no API keys, no Docker, no language runtimes.

graph TD
    A[Repository Files] --> B[File Watcher<br/>notify crate]
    B --> C[Indexer Pipeline<br/>Rayon + tree-sitter]
    C --> D[SQLite Graph Store<br/>WAL + FTS5 + sqlite-vec]
    D --> E[MCP Server<br/>Tokio + JSON-RPC 2.0]
    E --> F[AI Coding Agent]

    C --> G[Security Pass<br/>Taint + OWASP + SBOM]
    G --> D

    D --> H[Bundle Exporter<br/>cortex.json]
    H --> I[Git Repository<br/>Team Sharing]

    D --> J[Memory Layer<br/>Staleness-aware observations]

• Getting started
• IDE setup
• CLI reference
• MCP tools
• Configuration
• Architecture

Full docs site: 1337xcode.github.io/cortex

MIT