CullinanCloud/omniroute ? reverse-engineered prompt

Build me a local first AI gateway and dashboard called OmniRoute that lets me connect one endpoint to a bunch of model providers and automatically route requests to the best available option. I want it to support common coding tools like Claude Code, Cursor, Copilot, Cline, and similar apps, with smart fallback when one provider runs out or fails. Please include a clean dashboard where I can see my connected providers, available models, free tier usage, and remaining quota in a simple live view.

It should also help save tokens with compression so prompts and chat history use less context when possible, and it should work in the browser as a PWA and as a desktop app. I’d like basic setup from environment variables, good docs, and a polished UI that feels production ready. If you need current provider or integration details, look them up online first so the setup is realistic.