modelcontextprotocol/python-sdk ? reverse-engineered prompt

Build me a simple Python server that can provide tools and context to an LLM like Claude. Think of it as a specialized API designed specifically for an AI assistant to use.

The server should let me define a few different things. First, "resources," which are like read-only data sources the AI can query to get information. Second, "tools," which are actual Python functions the AI can decide to run to perform actions, like a calculator or an API call. And finally, I'd like to be able to create reusable "prompts" that act as templates for interactions.

Please build this using a modern Python setup with uv for package management. The framework should be easy to use, maybe with decorators like you see in FastAPI. Please include a basic example with a simple calculator tool and a resource that gives a personalized greeting, so I can see how it all works and test it out locally.