Summary

Open Claude Code is an open-source, clean-room implementation of Anthropic's Claude Code CLI, reverse-engineered from decompiled intelligence. It provides developers and power users with a transparent, auditable, and potentially more customizable alternative to the proprietary tool, enabling programmatic code analysis and generation.

How It Works

This project reconstructs the Claude Code CLI architecture, including its async generator agent loop, 25 tools, 4 MCP transports, and 6 permission modes, based on AI-powered analysis of the published npm package. It emphasizes a ground-up rebuild rather than using leaked source code, ensuring legal compliance while mirroring the original functionality and design principles.

Quick Start & Requirements

• Primary install/run: Use npx @ruvnet/open-claude-code "your prompt here" for instant execution without installation. Global installation is available via npm install -g @ruvnet/open-claude-code.
• Prerequisites: A valid ANTHROPIC_API_KEY environment variable is mandatory.
• Links: Releases, ADR-001, pi.ruv.io.

Highlighted Details

• Automated Nightly Releases: Features a CI/CD pipeline that detects new Claude Code releases, runs over 900 tests, performs AI-powered analysis, and publishes verified builds daily.
• Extensive Testing: Boasts 1,581 tests, aiming for 100% functional parity with the original.
• Multi-Provider Support: Integrates with Anthropic, OpenAI, Google AI, AWS Bedrock, and Google Vertex, allowing users to leverage different AI models and backends.
• MCP Integration: Supports connecting to MCP servers via stdio, SSE, Streamable HTTP, and WebSocket transports for extended functionality.

Maintenance & Community

The project utilizes the ruvnet/rudevolution submodule for its decompilation analysis. Key architectural decisions are documented in ADRs. No explicit community channels (like Discord or Slack) are listed in the README.

Licensing & Compatibility

The project is licensed under the MIT License, generally permissive for commercial use and integration. However, the reverse-engineered nature might require careful legal review for specific compliance scenarios.

Limitations & Caveats

Functionality is dependent on external AI provider APIs and requires valid API keys. As a reverse-engineered implementation, subtle behavioral differences from the original Claude Code CLI may exist. The project's reliance on decompiled intelligence, while legally framed, represents a novel approach to open-sourcing proprietary tools.