---
name: "social_crisis_response"
description: "Run a social media brand crisis like a comms professional: first-hour triage, holding statements, and when to apologize vs. stay quiet vs. correct the record. Triggers on reputation threats — viral complaints, outages, scandals, misinformation, pile-ons — affecting a brand or public figure's social presence."
---

# Social Crisis Response

## Purpose

Turn a chaotic social-media firestorm into a sequenced, professional response. This skill triages the crisis in the first hour, drafts a defensible holding statement, chooses the right response posture using research-backed crisis communication theory, and keeps everything on the record — while keeping a human in the approval loop for anything public-facing.

It is the crisis escalation point for routine social work: a single complaint is customer care, not a crisis (see triage in `references/triage.md`).

## Workflow

**0. Scope.** Confirm this is actually a crisis: fast-spreading negative mentions, press pickup, or a trending pile-on — not one angry customer. See `references/triage.md` for the severity ladder (Low → Medium → High).

**1. Freeze and assess (first 15 minutes).** See `references/first-hour-timeline.md`.
- Pause all scheduled/promo posts. Tone-deaf scheduled content during a crisis reads as contempt.
- Establish what is *verified*, what is *unknown*, and what is already public. Never communicate a fact you haven't verified.
- Name an incident lead. One voice, one fact sheet.

**2. Classify (minutes 15–30).** Place the crisis on the responsibility ladder:
- **Victim** — you're a victim too (tampering, rumor, natural disaster).
- **Accidental** — you caused harm but unintentionally (technical error, logistics failure).
- **Preventable** — human error or misconduct you should have stopped.
See `references/apology-decision.md` — prior crisis history bumps you one level more severe.

**3. Acknowledge (within the first hour).** Publish one timestamped holding statement: what happened, what you know, what you're doing next, and when the next update lands. Templates in `references/holding-statements.md`. Keep it factual — no cause narrative, no blame assignment, no legal conclusions yet.

**4. Respond (hour 1–24).** Match the posture to the classification:
- **Real failure → rebuild:** full, specific apology + concrete corrective action. Never a non-apology ("we're sorry you feel that way").
- **Accident → diminish + rebuild:** acknowledge, explain the unintentional cause briefly, commit to a fix.
- **False claims → correct with evidence:** calmly post verifiable facts; never apologize for what didn't happen; don't engage bad-faith actors emotionally.
- **Unverified pile-on → stay quiet on details, stay visible:** repeat the acknowledgment cadence, correct only what you can prove.

**5. Follow through (days 2–30).** Progress updates with dated evidence. Centralize everything in one pinned thread/newsroom post so there's a single citable record. Correct stale claims; don't let old answers stand.

## Output Contract

| Deliverable | What it contains |
|---|---|
| `incident-facts.md` | Verified / unknown / public facts table, incident lead, timestamp |
| `holding-statement.md` | Timestamped statement + next-update time, drafted for human approval |
| `response-plan.md` | Classification (victim/accidental/preventable), posture (deny/diminish/rebuild/correct/stay-quiet), channel plan, monitoring cadence |
| `follow-up-log.md` | Dated updates, evidence, and corrections until resolution |

## Operating Rules

1. **Human posts, never the agent.** Draft everything; a human (and legal/leadership for serious ones) approves and posts. Never publish a crisis statement unilaterally.
2. **No invented facts.** Every claim in a statement must be verified. Unverifiable = cut it or leave it in "unknown."
3. **Never apologize for what didn't happen.** A false apology creates liability and erodes trust when the truth comes out.
4. **Don't delete.** Leave negative comments up unless they violate platform policy — deletion fuels screenshots and anger.
5. **Match the apology to responsibility.** Under-accommodating a preventable crisis (excuses instead of apology) backfires harder than over-apologizing for an accident.
6. **One record.** All public communication points back to the single pinned/timestamped thread. No contradictory side replies.
7. **Speed ≠ recklessness.** The goal is a verified acknowledgment inside an hour, not a perfect explanation. The central question gets answered — fast vagueness that dodges the issue is as bad as silence.
8. **Debrief.** When resolved, log what happened, what worked, and what the plan missed — before the next crisis.