I typically provide small screenshots to llms so this seems fine for that usecase, providing an entire screens context seems cause confusion with a lot of llms.