Summary

Claude Code Proxy provides a transparent proxy solution for capturing, visualizing, and optionally routing Claude Code (claude.ai/code) requests and conversations. It offers developers real-time insights into API interactions, aiding debugging and enabling flexible multi-LLM agent configurations.

How It Works

The project employs a Go-based transparent proxy that intercepts traffic between Claude Code and the Anthropic API. It logs all requests and responses to an SQLite database and presents this data through a React Remix-powered web dashboard for live visualization and analysis. An optional subagent routing feature allows directing specific Claude Code agents to alternative LLM providers like OpenAI's models.

Quick Start & Requirements

• Prerequisites: Go 1.20+ and Node.js 18+ for local development, or Docker for containerized deployment.
• Installation: Clone the repository, copy config.yaml.example to config.yaml, and run make install followed by make dev for local setup. For Docker, build the image (docker build -t claude-code-proxy .) and run the container, mounting volumes for persistent data and configuration.
• Usage: Set export ANTHROPIC_BASE_URL=http://localhost:3001 and launch Claude Code. The web dashboard is accessible at http://localhost:5173.
• Docs: GitHub Repository

Highlighted Details

• Transparent Proxy: Seamlessly intercepts and monitors Claude Code API requests without user disruption.
• Agent Routing: Dynamically routes specific Claude Code agents to different LLM providers (e.g., GPT-4o, O3) via configuration.
• Request Monitoring: Comprehensive SQLite logging of all API interactions, including request/response bodies, with searchable history.
• Live Dashboard: Real-time visualization of requests, responses, and conversation threads, offering interactive exploration.

Maintenance & Community

The project is hosted on GitHub. Specific details regarding active contributors, community channels (e.g., Discord/Slack), or a public roadmap are not detailed in the README.

Licensing & Compatibility

The project is released under the permissive MIT License. This license generally allows for commercial use and integration into closed-source projects without significant restrictions.

Limitations & Caveats

The README does not specify alpha/beta status or list known bugs. The agent routing functionality is optional and requires explicit configuration in config.yaml or via environment variables. Functionality is dependent on the internal structure and API endpoints of Claude Code and Anthropic.