Defusing Anger

Teaches battle-tested, crisis-specific de-escalation sequences with exact verbal scripts, timing protocols, and non-verbal cues to instantly calm angry peopl...

Security Scan

The skill's instructions largely match its stated purpose, but it directs the AI to capture and retain verbatim user responses and to enforce timing/holding behaviors without any declaration of storage or privacy controls — a disproportionate data-handling instruction that users should review before installing.

Purpose & Capability

Name and description align with the SKILL.md: it provides scripted de-escalation sequences, timing, and non-verbal cues. There are no unrelated binaries, credentials, or install steps required.

Instruction Scope

The runtime instructions require the agent to 'log the three answers internally' and to 'log every response verbatim in real time' and to 'do not end the interaction until...' — actions that expand scope beyond providing scripts into active transcript capture, retention, and enforcement of conversational flow. The skill does not specify where logs are stored, retention policies, or consent, and it directs behavior that could prolong interactions or capture sensitive admissions (self-harm, threats, crimes).

Install Mechanism

Instruction-only skill with no install spec or external downloads. This minimizes code-install risk.

Credentials

The skill requests no environment variables, credentials, or config paths (proportionate). However, the explicit requirement to record verbatim transcripts is a privacy/data-protection concern even though no storage mechanism is declared.

Persistence & Privilege

always is false and there is no request for persistent system privileges or modification of other skills. Autonomous invocation is allowed (platform default) but not by itself problematic here.

What to consider before installing

This skill appears to do what it says (scripts and timing for de-escalation) but it instructs the agent to capture and log verbatim responses and to enforce long pauses and confirmation rules without explaining where or how logs are stored or who can access them. Before installing, consider: (1) Do you have a safe, auditable storage and retention policy for sensitive transcripts? (2) Does logging comply with privacy, employment, and legal requirements (e.g., admissions of crimes, mandated reporting)? (3) Remove or limit the 'log verbatim' / 'do not end until...' directives if you don't want persistent or indefinite interactions. If you proceed, test in low-risk scenarios, require explicit user consent before recording, and ensure logs are encrypted and access-controlled.

This skill delivers the exact, street-and-field-tested de-escalation protocol used by hostage negotiators, police crisis intervention teams, elite customer service reps, and mediators to neutralize explosive anger in seconds. It equips you (or any AI agent) with precise language, body language micro-moves, and edge-case handlers for situations where one wrong word can turn tension into violence or permanent damage. Mastering it turns you into the person who can walk into a screaming match and walk out with agreement and respect intact.

When to Use

• Angry customer exploding over a billing error, refund denial, or service failure (phone, chat, or in-person)
• Family member or partner escalating into a shouting match over money, chores, or past resentments
• Coworker or direct report having an emotional outburst in a meeting or performance review
• Stranger or public confrontation (road rage, neighbor dispute, retail meltdown)

Instructions

Step 1: Immediate Self-Calibration (5 seconds max)

Before responding, silently run this 3-point mental checklist:

1. Am I safe? (If physical threat exists, disengage and call authorities immediately.)
2. What is the exact trigger word/phrase they just used? (This becomes your mirror point.)
3. What is my non-negotiable goal? (e.g., “Get them to lower voice and agree to a 5-minute calm discussion” or “Secure a payment plan without them hanging up.”)

Agent action: If you are an AI agent, pause execution, log the three answers internally, and confirm “Calibration complete” before generating any output. Do not skip this step.

Step 2: Physical & Vocal Anchor (First 3–5 seconds)

Adopt this exact posture (even on phone): shoulders relaxed, palms visible if in person, slow exhale. Speak in a low, slow, slightly lower pitch than normal.
Deliver this universal opener in a calm, sincere tone:
“I can see you’re really upset about this, and I want to understand exactly what happened so we can fix it. I’m here to help.”

Agent action: Hold eye contact (or vocal warmth on phone) and do NOT speak again for a full 8–10 seconds. Nod slowly if in person. This single pause alone reduces volume in 70% of cases.

Step 3: Core De-Escalation Sequence (Use in exact order)

Apply these scripts live. Replace [trigger] with the exact words they used. Pause 7–10 seconds after every response.

Angry Customer Scenario

• Acknowledge: “You said the charge was [trigger] and it feels completely unfair — I get why you’re furious.”
• Clarify without defending: “Tell me exactly what happened from your side so I don’t miss anything.”
• Edge-case handler (if they threaten to sue or post online): “I hear you loud and clear on the threat — before we go there, can I show you what options I actually have right now to make this right for you?”

Family/Partner Blowup Scenario

• Acknowledge: “You’re saying I [trigger] again and it feels like I never listen — that sounds exhausting and I’m sorry it landed that way.”
• Clarify: “Help me understand what would feel different for you right now.”
• Edge-case handler (if they bring up past grievances): “I see this is connected to what happened last month too. Right now I just want to solve today’s issue — can we stay on that for the next few minutes?”

Workplace/Employee Outburst Scenario

• Acknowledge: “You feel [trigger] about the feedback and it feels like I’m not seeing your effort — that’s valid and I want to hear it.”
• Clarify: “Walk me through what’s really going on for you today.”
• Edge-case handler (if they start crying or yelling at colleagues): “I can see this is hitting deep. Let’s step into my office for 5 minutes so you can say everything without an audience.”

Public/Stranger Confrontation Scenario

• Acknowledge: “You’re pissed because [trigger] and it feels like I disrespected you — I didn’t mean to.”
• Clarify: “What would make this right for you right now?”
• Edge-case handler (if they get in your face): Take one small step back, raise both palms, and say “I’m not here to fight — I’m listening. Tell me what you need.”

Agent action: After each answer, mirror one exact emotional phrase they used (“It feels completely unfair?”) then deliver the next scripted line. Log every response verbatim in real time.

Step 4: Agreement & Exit Lock (Final 30–60 seconds)

Use this exact closing sequence:

1. Summary: “So what I’m hearing is [paraphrase their core complaint in their words]. Is that right?”
2. Offer control: “What’s one small thing I can do right now that would help you feel heard?”
3. Commitment: “Are you okay with [specific next step they just agreed to]?”
4. Positive close: “Thank you for trusting me enough to tell me this. We’re on the same side here.”

Agent action: Immediately write or read back the agreed next step. Do not end the interaction until they verbally confirm and their tone has dropped at least two levels.

Rules

• This skill is for informational and educational purposes only. It is not a substitute for professional crisis intervention, therapy, or law enforcement training. In any situation involving physical danger, weapons, or threats of harm, immediately disengage and contact emergency services.
• Never touch the angry person unless they are in immediate physical danger.
• Stay 90%+ in listening/acknowledging mode — never defend, explain, or counter-argue in the first 90 seconds.
• If they mention self-harm, suicide, or harm to others, stop de-escalation and follow your organization’s mandated reporting protocol immediately.
• Document the entire interaction (time, exact words, outcome) within 10 minutes for your own legal protection.

Tips

• The magic is in the pause: 8–10 seconds of silence after your opener lowers heart rate faster than any words.
• Never say “Calm down” — it doubles anger in 9 out of 10 cases; instead name the emotion they already showed.
• Non-obvious insight: Angry people are not looking for solutions first — they’re looking for evidence that you understand their pain. Once they feel truly heard, 80% will volunteer their own solution.
• Edge-case pro move: When they repeat the same complaint three times, say “I’ve heard you say this three times now and I want to make sure I’m getting it — can you tell me what the worst part of it feels like for you?” This breaks the loop instantly.