cgviz

Explore your codebase as an interactive graph — one command, zero config.

cgviz turns a CodeGraph database (a tree-sitter–parsed knowledge graph stored at .codegraph/codegraph.db) into a fast, browser-based visualization of every symbol, call, import, and containment relationship in your project. It’s built for understanding architecture, tracing call flows, and impact analysis at a glance.

• WebGL rendering (Sigma.js + Graphology) — handles thousands of nodes smoothly
• Focus mode — click any symbol to gather its caller/callee cluster, centered and labeled
• 2D and immersive 3D views (Three.js), toggleable
• Multiple layouts — force-directed, radial, hierarchical, and tree
• Full-text search and filters by node kind, edge kind, and file path
• View modes — all edges, call graph, inheritance, and structure
• Live updates — re-renders when the .codegraph database changes
• Dark and light themes with a refined color palette

• Python ≥ 3.10
• A project containing a CodeGraph database (.codegraph/codegraph.db)

The PyPI package ships with the frontend pre-built, so installing it requires no Node.js.

pip3 install cgviz

From the root of a project that has a .codegraph/ directory:

cgviz

…or point it at any project (or directly at a .db file):

cgviz /path/to/your/project

This starts a local server and opens your browser with the interactive graph. cgviz walks up from the given path to auto-detect .codegraph/codegraph.db.

cgviz [OPTIONS] [PATH]

All options can also be set via environment variables with the CGVIZ_ prefix (e.g. CGVIZ_PORT=8080).

import cgviz

# Blocking launch (Ctrl-C to stop)
cgviz.launch("/path/to/project", port=7777, open_browser=True)

cgviz opens the CodeGraph SQLite database read-only, serves it through a FastAPI backend, and renders it with a React + Sigma.js / Three.js frontend. It never modifies your code or your database.

The frontend (TypeScript / React) lives in frontend/ and is built into src/cgviz/static/. Building it requires Node.js ≥ 18.

# Python package, editable + dev tools
pip3 install -e ".[dev]"

# Frontend dependencies
cd frontend && npm install && cd ..

# Build the frontend into the package
python3 scripts/build_frontend.py

# Run backend + frontend dev servers together
python3 scripts/dev.py /path/to/project

# Run tests
pytest

MIT