Have the agent attempt to name/classify the current screen based on possible options.

I think the overarching thing here is how do we give the app context on the current app? We do have custom commands, but this is different.

@BrianGardnerAtl had some ideas on this, but it could be valuable to prove the LLM with app specific context to support more ambiguous prompts.

The issue names no file, test, or entry point. Start by clarifying the proposed app-specific context and possible screen options with the issue author; done should be a defined agent behavior for naming or classifying the current screen and a testable result.

Written by the indexing model from the issue text.