justlovemaki/AIClient2API

What it solves

AIClient2API is an API proxy service that converts free large models restricted to client-only use (such as Gemini CLI, Grok, Codex, and Kiro) into standard, OpenAI-compatible interfaces. This allows users to bypass client limitations and use these advanced models within any application that supports the OpenAI API specification, such as Cherry-Studio, NextChat, or Cline.

How it works

Built on Node.js, the project uses a modular architecture with strategy and adapter patterns to simulate client requests and encapsulate them into a local API. It features a built-in TLS Sidecar (Go uTLS) to bypass Cloudflare 403 blocks and supports intelligent conversion between OpenAI, Claude, and Gemini protocols. To ensure high availability, it includes an account pool manager for multi-account polling, automatic failover, and health checks.

Who it’s for

It is designed for developers and AI enthusiasts who want to access advanced models (like Claude Opus or Gemini Pro) for free or at a lower cost by leveraging client-only interfaces, and for those who need a unified API gateway to manage multiple AI providers in one place.

Highlights

• Unified Interface: Provides a single OpenAI-compatible endpoint for models from Gemini, Claude, Grok, Codex, and others.
• Client Bypass: Uses OAuth authorization and TLS fingerprint simulation to break through rate limits and official restrictions.
• Account Pool Management: Supports multi-account polling and automatic degradation to maintain 99.9% service availability.
• Web UI Console: Includes a management interface for real-time configuration, health monitoring, and API testing.
• Agent-Ready: Natively supports interaction with AI agents like OpenClaw and Claude Code via specialized skill guides.