agentleFS
ironyjkironyjk/howtotalk/crucial-conversations/SKILL.md

crucial-conversations

Skill2 starsChanged 7 months agoMITView on GitHub

Crucial Conversations (Patterson, Grenny, McMillan, Switzler) - Talk when stakes are high, opinions vary, and emotions run strong

We split this skill into the rules it gives an agent

9 rules. About 0 are common, 9 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈9

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Every conversation has a "pool" of shared meaning — all the information, ideas, theories, feelings, and opinions that participants share …

    line 34
  2. Distinctive:

    Goal of every crucial conversation: Keep filling the pool, even when it's hard.

    line 39
  3. Distinctive:

    Before opening your mouth, get clear on what you REALLY want:

    line 58
  4. Distinctive:

    When safety breaks down, step OUT of the content and restore safety:

    line 85
  5. Distinctive:

    If missing → use CRIB:

    line 89
  6. Distinctive:

    If missing → Apologize sincerely if you've been disrespectful

    line 96
  7. Distinctive:

    Key: Start with FACTS, not stories.

    line 135
  8. Distinctive:

    When the OTHER person has a story (especially angry/silent):

    line 139
  9. Distinctive:

    When preparing for or analyzing a crucial conversation:

    line 170

≈ “Common” is an estimate: at least 1 in 5 of the 308,232 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Crucial Conversations Skill

"At the core of every successful conversation lies the free flow of relevant information." — Patterson et al.

Overview

A Crucial Conversation (결정적 대화) occurs when three conditions collide: (1) stakes are high (이해관계가 크다), (2) opinions differ (의견이 다르다), (3) emotions are strong (감정이 격하다). Most people handle these moments with either silence (침묵) or violence (공격) — both destroy dialogue. This framework teaches how to stay in dialogue when it matters most.

Sub-commands

CommandDescription
/crucialFull crucial conversation preparation and script
/crucial:stateApply the STATE method to share a tough message
/crucial:safetyDiagnose and restore safety in a broken conversation
/crucial:storiesAnalyze "Master My Stories" — separate facts from interpretations

Core Concept: The Pool of Shared Meaning (공유 의미의 풀)

Every conversation has a "pool" of shared meaning — all the information, ideas, theories, feelings, and opinions that participants share openly. The larger the pool:

  • Better decisions (more information available)
  • Greater buy-in (people feel heard)
  • Stronger commitment (unity of purpose)

Goal of every crucial conversation: Keep filling the pool, even when it's hard.

Two main threats to the pool:

  1. Silence (침묵) — Withholding information from the pool
    • Masking: Understating or selectively showing true opinions
    • Avoiding: Steering away from sensitive subjects entirely
    • Withdrawing: Pulling out of the conversation altogether
  2. Violence (폭력적 대화) — Forcing meaning into the pool
    • Controlling: Coercing others to your way of thinking (cutting off, overstating, speaking absolutes)
    • Labeling: Putting a label on people or ideas to dismiss them
    • Attacking: Moving from winning the argument to making the other person suffer

Process

Step 1: Start with Heart (마음에서 시작하라)

Before opening your mouth, get clear on what you REALLY want:

  1. What do I really want for myself?
  2. What do I really want for the other person?
  3. What do I really want for the relationship?
  4. How would I behave if I really wanted these results?

The Sucker's Choice (바보의 선택): The false dilemma that you must choose between honesty and kindness. Refuse it. Ask: "How can I be 100% honest AND 100% respectful?"

Step 2: Learn to Look (관찰하는 법을 배워라)

Watch for two critical signals during the conversation:

Signal 1: When a conversation becomes crucial

  • Physical signs: stomach tightening, eyes drying, voice rising
  • Emotional signs: feeling scared, hurt, angry
  • Behavioral signs: raising voice, going quiet, pointing fingers

Signal 2: When safety is at risk

  • The other person goes to silence (masking, avoiding, withdrawing)
  • The other person goes to violence (controlling, labeling, attacking)
  • YOU go to silence or violence

Dual-processing: Simultaneously watch the CONTENT and the CONDITIONS (safety level).

Step 3: Make It Safe (안전하게 만들어라)

When safety breaks down, step OUT of the content and restore safety:

Two conditions for safety:

  1. Mutual Purpose (상호 목적) — Do they believe I care about their goals?
    • If missing → use CRIB:
      • Commit to seek mutual purpose: "I want to find something that works for both of us"
      • Recognize the purpose behind the strategy: What do they REALLY want?
      • Invent a mutual purpose: Find a higher/broader goal you share
      • Brainstorm new strategies: With mutual purpose clear, generate options together
  2. Mutual Respect (상호 존중) — Do they believe I respect them?
    • If missing → Apologize sincerely if you've been disrespectful
    • Find something genuinely respectable about the person
    • Contrast to fix misunderstandings

Contrasting (대비 기법):

  • "I DON'T want you to think [the negative thing they fear]. I DO want [your actual positive purpose]."
  • Example: "I don't want you to think I'm questioning your competence. I do want to find a way we can meet the deadline together."

Step 4: Master My Stories (나의 이야기를 통제하라)

Between what happens and how we feel, there's a story we tell ourselves:

See/Hear → Tell a Story → Feel → Act
(관찰)     (이야기)       (감정)  (행동)

Three "Clever Stories" (교묘한 이야기) that keep us stuck:

  1. Victim Story (피해자 이야기): "It's not my fault" — I am innocent, blameless
  2. Villain Story (악당 이야기): "It's all your fault" — The other person is evil/stupid/wrong
  3. Helpless Story (무력감 이야기): "There's nothing I can do" — I have no options

Counter-technique — Turn victims into actors, villains into humans, helpless into able:

  • "What am I pretending not to notice about MY role in this?"
  • "Why would a reasonable, rational, decent person do what they did?"
  • "What do I really want? What would I do right now to move toward that?"

Step 5: STATE My Path (STATE 기법으로 말하라)

The core method for sharing tough messages:

StepActionExample
Share your factsStart with the least controversial, most objective elements"I've noticed you've missed three of our last five 1:1 meetings"
Tell your storyExplain your interpretation — tentatively"I'm starting to wonder if these meetings aren't valuable to you"
Ask for others' pathsInvite the other person's viewpoint"I'd like to hear how you see it. What's going on?"
Talk tentativelyExpress your story as a story, not a fact"I could be wrong, but..." / "From my perspective..."
Encourage testingMake it safe for others to push back"I really want to hear if you see this differently"

Key: Start with FACTS, not stories. Facts are the least controversial starting point and hardest to disagree with.

Step 6: Explore Others' Paths (상대방의 길을 탐색하라)

When the OTHER person has a story (especially angry/silent):

AMPP Listening Tools:

  • Ask: "What's going on?" / "I'd really like to hear your perspective"
  • Mirror: "You say you're fine, but your tone tells me otherwise"
  • Paraphrase: "So you're saying... Let me make sure I understand..."
  • Prime (when all else fails): Take your best guess at what they're feeling — "Are you upset because you feel I went over your head?"

ABC Agreement:

  • Agree: Find common ground first
  • Build: Add elements they might have missed
  • Compare: When you genuinely disagree, compare your views (don't say "wrong")

Step 7: Move to Action (행동으로 옮겨라)

Dialogue without decision is just talk. Before ending:

  1. Decide HOW to decide:
    • Command (명령): One person decides (appropriate when it's their authority)
    • Consult (자문): Gather input, then one person decides
    • Vote (투표): Majority rules (when efficiency > quality)
    • Consensus (합의): Everyone agrees (highest quality, slowest)
  2. Document the decision:
    • Who does What by When
    • How will we follow up?

Output Format

When preparing for or analyzing a crucial conversation:

## Crucial Conversation Analysis

**Situation:** [description]
**Why it's crucial:** Stakes: ___ / Opinions differ: ___ / Emotions: ___

### Start with Heart
- What I really want for me: ___
- What I really want for them: ___
- What I really want for the relationship: ___

### My Stories (check for clever stories)
- Facts (camera-testable): ___
- Story I'm telling myself: ___
- Victim/Villain/Helpless elements: ___
- Reframed story: ___

### Safety Assessment
- Mutual Purpose present? ___
- Mutual Respect present? ___
- Contrasting statement if needed: "I don't want... I do want..."

### STATE Script
- **S** (Facts): ___
- **T** (My story, tentatively): ___
- **A** (Ask): ___
- **T** (Tentative language): ___
- **E** (Encourage testing): ___

### Move to Action
- Decision method: ___
- Who/What/When: ___

Anti-patterns

  1. Leading with your story, not facts — "You don't care about this project" vs. "The report was submitted two days late"
  2. Sucker's Choice — "I can either be honest OR keep the relationship" — refuse this false dilemma
  3. Playing it safe — Avoiding the conversation entirely because it's uncomfortable
  4. Telling the Villain Story — Attributing the worst possible motive to the other person
  5. Contrasting incorrectly — Only using the "I don't want" half without the positive "I do want"
  6. Forcing your path — Using STATE without genuinely asking for their perspective
  7. Content fixation — Getting so wrapped up in the topic that you miss safety breaking down
  8. Skipping Move to Action — Having a great dialogue but leaving without Who/What/When
  9. Premature problem-solving — Jumping to solutions before both sides feel fully heard
  10. Consensus fetish — Insisting on consensus when consulting or command would be more appropriate

References

  • crucial-conversations/references/state-method.md — Detailed STATE method guide with examples
  • Patterson, Grenny, McMillan, Switzler, Crucial Conversations: Tools for Talking When Stakes Are High (3rd ed., 2021)
  • Patterson et al., Crucial Accountability (2013)
  • VitalSmarts (now Crucial Learning): https://cruciallearning.com

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

Reviews can't be read right now.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.