Claude Code with intelligent code graphs for large codebases

Claude Code Graph enhances Claude Code with live structural analysis of your codebase. It builds and maintains graphs of your code's architecture, dependencies, and relationships - giving Claude deep contextual understanding of even the largest projects.

🚀 Quick Start

Prerequisites

Make sure you have Claude Code installed first:

npm install -g @anthropic-ai/claude-code

Installation

1. Install Claude Code Graph globally:

   npm install -g claude-code-graph

2. Navigate to your project and start coding:

   cd your-project/
   ccg start

That's it! The first run will automatically:

• Initialize graph structures
• Build code analysis graphs
• Start the live update daemon
• Launch Claude Code with graph intelligence

🎯 What It Does

Claude Code Graph provides Claude with:

• Dependency Analysis: Understand how files and modules relate to each other
• Call Graph Mapping: Track function calls and inheritance chains across languages
• AST Structure: Deep understanding of code syntax and semantics
• Architecture Overview: High-level view of your codebase organization
• Live Updates: Graphs stay current as you code

Supported Languages

• JavaScript/TypeScript: Module dependencies via madge
• Python: Call graphs via pyan3, import analysis
• C/C++: Include dependencies, symbol analysis
• Universal: AST parsing with tree-sitter for all languages

📋 Commands

🔧 Advanced Setup

Manual Workflow (if needed)

# 1. Initialize in your project
ccg init

# 2. Check dependencies 
ccg doctor

# 3. Build graphs
ccg build

# 4. Start daemon for live updates
ccg daemon start

# 5. Launch Claude Code
ccg start

Dependencies

Required:

• Node.js 18+
• Python 3.8+
• madge (for JS/TS analysis)

Optional (improves analysis):

• tree-sitter-cli: npm install -g tree-sitter-cli
• pyan3: pip install pyan3
• Python packages: pip install watchdog networkx psutil aiofiles

Check with: ccg doctor

📊 Performance

Claude Code Graph is optimized for large codebases:

• Smart Filtering: Skips node_modules, build directories, etc.
• File Limits: Processes up to 500 files for tree-sitter, 200 for C++, 50 for Python
• Incremental Updates: Only processes changed files via daemon
• Caching: Intelligent caching for faster subsequent runs

For projects with 1000+ files, initial graph building takes 2-5 minutes instead of 30+ minutes.

📁 Project Structure

your-project/
├── .graph/                 # Graph data (auto-created)
│   ├── js.json            # JavaScript/TypeScript modules  
│   ├── py.dot             # Python call graph
│   ├── cpp.json           # C++ dependencies
│   ├── ts.json            # Tree-sitter AST
│   ├── metrics.json       # Graph statistics
│   └── daemon.lock        # Daemon status
├── .gitignore             # Updated to exclude .graph/
└── your code...

🛠️ Development

# Clone the repo
git clone https://github.com/aibozo/claude-code.graph.git
cd claude-code-graph

# Install dependencies
npm install

# Build tools
npm run build

# Install globally for testing
npm install -g .

# Run tests
npm test

🤝 Contributing

1. Fork the repository
2. Create a feature branch: git checkout -b feature-name
3. Make your changes and test thoroughly
4. Submit a pull request

📝 License

MIT License - see LICENSE.md for details.

🐛 Issues & Support

• Report bugs: GitHub Issues
• For Claude Code issues: Use /bug command or Claude Code Issues

🔗 Links

• Claude Code Documentation
• Claude Code Repository
• Anthropic Website

Made with ❤️ for better AI-assisted coding