Use when a user identifies an AI agent fault, overreach, misunderstanding, unauthorized action, or asks for a retrospective, durable fault rule, trigger keyword, or future automatic invocation; also use when existing fault records need to be consulted.
Idioma del texto original: inglés