lidge-jun/opencodex ? reverse-engineered prompt

Build me a small local app called opencodex that lets OpenAI Codex work with lots of other AI providers instead of only OpenAI. I want it to run on my computer as a lightweight proxy, so Codex CLI, app, and SDK can keep using the normal responses flow while this tool translates things behind the scenes. It should support common providers like Claude, Gemini, Grok, DeepSeek, Kimi, Qwen, Azure, OpenRouter, Ollama, and any OpenAI compatible endpoint, including local models.

Please make the setup feel easy. I want a simple command line flow to install, initialize, start, stop, restore, check status, and open a browser dashboard on localhost where I can add providers, paste keys or sign in for supported ones, see discovered models, pick a default, and view a live request log. Model routing should let me use provider slash model names in Codex and also show routed models in the Codex picker. It should work on Mac, Linux, and Windows, and support running in the background or starting on demand. Look up current docs online if you need to.