diegosouzapw/OmniRoute ? reverse-engineered prompt

I want to build an AI gateway to manage all the different LLM services I use. It should work like a universal proxy, giving me a single, OpenAI-compatible endpoint for everything. The most important feature is smart routing to save money and improve reliability. The system should automatically try to use free or low-cost models first, and if a provider is down, it needs to automatically fall back to a different one.

It needs a clean web dashboard where I can configure providers, set up routing policies or "combos", and see which services are up or down on a health status page. I also want to see analytics on my usage and costs. The gateway should support various API types like chat, image generation, and embeddings. Lastly, let's also package this as a downloadable desktop app, maybe built with Electron. Use TypeScript for everything.