coder2api 0.1.1

Turn Gemini-CLI, Claude Code, and Codex into a local OpenAI-compatible API endpoint.

coder2api

A unified wrapper for Gemini CLI Proxy, Claude Code API, and ChatMock.

This tool provides a single entry point to run and manage these local coding agent proxies, exposing them via a unified API endpoint.

Installation

This project uses uv for dependency management.

uv sync

Because coder2api includes a Hatch build hook, running uv sync (or pip install) will automatically:

1. Install Python dependencies.
2. Run npm install and npm run build for the Gemini CLI Proxy (Node.js).

Note: You must have Node.js and npm installed on your system.

Usage

Unified Server

Start all services and the unified proxy:

uv run coder2api serve

This will start:

• Gemini CLI Proxy on port 3001
• ChatMock on port 3002
• Claude Code API on port 3003
• Unified Proxy on port 8069

You can access the APIs via the proxy:

• Codex (ChatMock): http://localhost:8069/codex/v1/...
• Claude Code: http://localhost:8069/cc/v1/...
• Gemini: http://localhost:8069/gemini/openai/... or http://localhost:8069/gemini/anthropic/...

CLI Wrappers

You can also use the CLI wrappers for individual tools:

ChatMock (Codex):

uv run coder2api codex --help
uv run coder2api codex login

Claude Code API:

uv run coder2api cc

Gemini CLI Proxy:

uv run coder2api gemini -- --help

Logs

Logs for the background services are written to the logs/ directory when running serve.

License

This project is licensed under the GNU General Public License v3.0 - see the LICENSE file for details.