router-for-me / CLIProxyAPI

A high-availability proxy that transforms multiple AI CLI coding tools into unified, standard API endpoints for seamless programmatic access.

活跃维护 MIT Go Tracked
56.4k 8.6k 2 小时前
CIDocker
claude-code cluade gemini openai codex antigravity devin muse

星标趋势

数据积累中,暂无足够数据生成趋势图

AI 分析

项目摘要

CLIProxyAPI is a Go-based proxy server that wraps multiple AI CLI providers (Claude Code, ChatGPT Codex, Kimi Code, Grok Build, Muse Code, Devin) and exposes them through unified OpenAI/Gemini/Claude/Codex-compatible API interfaces, enabling programmatic access to various AI models that are typically only available via command-line tools.

为什么值得关注

With over 55K stars and 258 contributors, this project has rapidly gained traction by solving a real developer pain point: making CLI-only AI coding tools accessible through standard API interfaces. Its breadth of provider support (Kimi, OpenAI, Claude, Google Antigravity, etc.) and extremely active development (100 releases in 6 months) demonstrate strong community demand for unified AI model access.

优势

  • Unified API interface for multiple AI CLI providers under a single endpoint
  • Extremely active development with 100 releases in 6 months and strong community engagement (55K+ stars, 258 contributors)
  • Comprehensive multi-language documentation (English, Chinese, Japanese) with clear provider comparison tables and Docker support

局限性

  • Reliability and availability depend entirely on upstream CLI tools and their authentication mechanisms
  • Free model access through CLI wrappers may exist in a gray area regarding provider terms of service
  • Not a true model-serving solution; adds latency and potential failure points as a proxy layer

使用场景

  • Integrating AI CLI coding assistants into custom development workflows via standard API calls
  • Building applications that need to switch between multiple AI model providers without vendor lock-in
  • Enabling team-wide access to AI models through a single, consistent API interface
目标用户: Developers and teams who want to use AI coding assistants (Claude Code, Codex, Kimi Code, etc.) programmatically through standard OpenAI/Gemini-compatible APIs rather than interactive CLI sessions
学习曲线: 中
分析模型:LongCat-2.0 | 分析时间:11 天前