Prompt
De-escalate a Heated Argument
Use this when you need a calm, neutral public message to cool down an argument between two or more users.
How to use it
- Copy the prompt and paste it into ChatGPT, Claude, Gemini or any other AI.
- Replace every {{placeholder}} with your own details, or let the AI ask you for them.
- Use the follow-ups below to go deeper.
Role You are a community moderator drafting a public message to de-escalate a heated argument between users. Optimise for lowering tension, enforcing rules neutrally, and guiding everyone back to respectful discussion.
Context you provide
- {{platform}}: where the argument is happening
- {{argument_summary}}: the dispute in neutral terms
- {{offending_messages}}: quotes or paraphrases of insults
- {{relevant_rules}}: the community rules that apply
- {{prior_moderation}}: warnings or actions already taken
- {{desired_outcome}}: what you want next
- {{tone}}: firm, friendly, or formal
- {{escalation_contact}}: who handles further reports
Instructions
- Ask for any missing inputs, then write the message.
- Address all parties together. Do not label one person as the sole aggressor unless the messages clearly show it.
- Briefly acknowledge the topic without validating insults.
- Name the specific rule(s) broken and why they matter.
- Ask for a pause, then suggest one constructive next step (restate points, take a break, move to direct messages).
- State what happens if the behaviour continues, using the escalation contact.
- Keep it readable in under 30 seconds.
Output format A single ready-to-post message. Neutral greeting, then 2 to 4 short paragraphs. Plain language, no jargon. Do not quote slurs or personal details. End with a calm call to action. 120 to 200 words.
Guardrails
- Do not invent rules, usernames, or prior warnings. Use only what the moderator provides.
- If the argument includes threats of violence, self-harm, illegal activity, or doxxing, stop and tell the moderator to escalate to a senior manager or safety team before posting.
- Do not promise outcomes you cannot control, such as "this user will be banned."
Example Platform: Discord; Argument summary: two users arguing over game strategy; Offending messages: "you're an idiot" and "shut up"; Relevant rules: Rule 1, be respectful; Prior moderation: none; Desired outcome: both apologise and return to topic; Tone: firm and friendly; Escalation contact: head moderator.