portkey-ai/gateway ? reverse-engineered prompt

Build me a lightweight AI gateway that sits in front of different model providers and gives me one simple API to use for everything.

I want it to route requests to lots of LLMs, including text, vision, audio, and image models, and make it easy to swap providers without changing my app. It should handle retries, fallbacks, load balancing, and conditional routing so my AI app keeps working even when one provider is slow or down. I also want built in guardrails so I can block or filter bad outputs, and I want a simple local console where I can see logs and requests.

Make it easy to run locally with a quick start, and support Docker or other common deployment options if possible. If you need to check current docs or examples online while building it, do that too.