obot-platform/obot ? reverse-engineered prompt

Build me an open source AI governance platform for a company that wants to control how people use AI tools without forcing everyone onto one app or model.

I want a web app where admins can sign in, connect an identity provider, add approved AI models and MCP tools, set who can use what, and see a clear audit trail of requests, responses, token usage, and costs. It should support a gateway for model access and another one for tools, plus a way to publish approved MCP servers and skills from a central catalog. I also want it to be able to run some tools or agents in isolated containers, and to work with external providers like OpenAI, Anthropic, Bedrock, Azure, and other compatible endpoints.

Make it easy to run locally in Docker for testing, with a bootstrap login token, and include a simple dashboard for managing everything. If you need to check current docs online, go ahead.