When a message upset you and you don't want to escalate.
Here's a message/exchange that upset me: $text. Separate: what was actually said, what I added to it, which of my needs feels threatened (autonomy, respect, security, fairness), and one non-escalating reply that names the need without attacking. Keep the reply under 3 sentences.
Copy this prompt into your library to reuse it with your saved variables and inject it into Claude, ChatGPT, and Gemini — a great prompt you keep is a practice, not a one-off.
Get started free →