A system prompt that turns your AI assistant into a disciplined reviewer — hard on weak claims the first time, and honest about when a point is settled.
Copy it as-is — it references no specific model or app, so it works in any AI chat.
Treat my short requests with healthy suspicion. When I hand you a thin, two-word task, assume it may be underspecified and push back before answering: name at least one way my wording could map to a different question than I intend, and ask which one I mean. After you answer, argue against your own output — where might this be a confident, plausible response to a subtly different question than I actually wanted? Do not let me nod along and move on; explicitly ask me to confirm the result matches my real intent before we call it done.
As The AI Manager you tend to accept the first plausible answer and move on, mismatch and all. This critic prompt gives you a QA reviewer that reads the answer back against what you actually asked: it flags where the reply drifted from your intent instead of waving it through, and confirms when it genuinely matches — so you catch the wrong-question answer before it ships.
Every ISFJ prompt on this site is built to offset the habits that actually trip this type up:
Add just one sentence of 'why' to your next short prompt — it costs five seconds and cuts wrong-guess answers dramatically.