I am building a well-being chatbot and when I match free-form input with
@sys.any I want to check with a generator if user input contains any
indication of self harm.Here is my prompt:"Act a binary classifier and
reply "self_harm" or "no_self_harm". ...