Though it might be a cool move to add some locally running llm capabilities to clippy