px0-ai/px0 ? reverse-engineered prompt

Build me a Go based backend for a prompt management platform that lets me version prompts, update them without redeploying the main app, and control who can change them. I want it to support prompt templates, running prompts against different model providers, and keeping track of versions so I can roll back if something goes wrong. It should also have a way to manage reusable agent skills and structured tool schemas, plus an MCP server so AI assistants can call the API directly.

Please make it easy to run locally with Docker Compose, include a health check endpoint, and set up the database, cache, and observability pieces so I can see what is happening in Grafana and Prometheus. If you need to, look up the current docs online and follow the project README closely. I mostly want something I can start up, test, and extend without having to change application code every time a prompt changes.