pitbaden/omniroute ? reverse-engineered prompt

Build me OmniRoute, a web app for routing AI requests through one simple endpoint. I want to connect different LLM providers, set up smart fallback rules so requests keep working if one model fails, and make it easy to switch between free or low cost options without changing my tools.

Please include a clean dashboard where I can add providers, see health and usage, manage routing combos or policies, set rate limits and caching, and view analytics and logs so I can tell what is being used and what it is costing me. It should feel like a practical control center for an AI gateway, not just a settings page.

Also make sure it works with OpenAI style API requests so coding tools and other clients can point to it easily. If anything is unclear, look up current docs online if you need to.