Using schema as intput for MCP

Considering that SDL is used in this project as a language to formalize a semantic model of a domain (and not the API itself), what is the recommended strategy for using graphql schema as input context for the [Model Context Protocol](https://modelcontextprotocol.io/docs/getting-started/intro)?

- Via introspection function?
- Via full SDL file?
- Via queries?
- Other?

Example where context might be needed.

- A person in a car says `"I am feeling warm"`
- The model should explicitly know that affecting that perceived temperature is possible by acting on the window position, the ac tempterature, the sunroof, etc.
- This information has to be passed to the LLM to better scope the task.

No files, tests, or implementation entry points are identified. First clarify whether this is intended as a code change or a design decision, then compare the proposed introspection, full SDL, and query-based approaches against the Model Context Protocol documentation. Done requires an agreed strategy and a concrete scope for any follow-up implementation.

Written by the indexing model from the issue text.