From grimoire
Assesses and improves team psychological safety using Edmondson's 7-item survey and evidence-based leader behaviors. For team leads wanting to increase speaking up, admitting mistakes, and productive disagreement.
How this skill is triggered — by the user, by Claude, or both
Slash command
/grimoire:build-psychological-safetyThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Create the conditions where every team member can speak up, admit mistakes, ask questions, and challenge assumptions without fear of professional or social consequences.
Create the conditions where every team member can speak up, admit mistakes, ask questions, and challenge assumptions without fear of professional or social consequences.
Adopted by: Google's Project Aristotle (2016) — a two-year study of 180 teams — identified psychological safety as the #1 predictor of team effectiveness, above technical skill, individual talent, or process; Amy Edmondson's research spans 30+ years and 1,000+ organizations across healthcare, aviation, finance, and technology; the framework has been adopted by Pixar (the Braintrust model), Bridgewater (radical transparency), and was central to NASA's post-Columbia accident inquiry into why engineers felt unable to raise known safety concerns Impact: Google Project Aristotle found that teams high in psychological safety were 17% more likely to perform in the top quartile, had 19% lower attrition, and generated 27% more revenue; Edmondson's original hospital study (Harvard, 1999) showed teams with high psychological safety reported 10× more medication errors — not because they made more errors, but because they felt safe reporting them; in high-risk industries (aviation, nuclear), the same pattern holds: low psychological safety suppresses reporting of near-misses, which is the primary mechanism through which major accidents are prevented Why best: "Open door" policies are passive — they signal availability but do not remove the interpersonal risk of speaking up; team-building activities build social cohesion but do not address fear of professional consequences; psychological safety requires active, visible leader behavior that models fallibility and explicitly rewards dissenting input — it cannot be declared or mandated, only cultivated through consistent behavior
Sources: Edmondson "The Fearless Organization" (Harvard Business School Press, 2018); Google re:Work "Guide: Understand Team Effectiveness" (2016, rework.withgoogle.com); Clark "The 4 Stages of Psychological Safety" (Berrett-Koehler, 2020); Edmondson "Psychological Safety and Learning Behavior in Work Teams" (Administrative Science Quarterly, 1999)
Use Edmondson's 7-item survey, rated 1–7 (strongly disagree to strongly agree). Collect anonymously:
Items 1, 3, and 5 are reverse-scored. Average across all items. Compare against Edmondson's published team benchmarks:
Run this survey quarterly. Track the trend, not the absolute score.
Psychological safety is learned through observation of the leader. Team members calibrate their behavior by watching whether leaders face negative consequences for admitting uncertainty or mistakes.
Concrete actions:
Admit mistakes openly in team settings:
"I was wrong about that estimate — I didn't account for the integration work.
Here's what I'm doing differently next time."
Signal the limits of your knowledge:
"I don't know the answer to that. Let me find out and come back to you by Thursday."
Invite challenge to your own positions:
"I'm proposing we go with option A, but I want to be pushed on this.
What am I missing? What's the strongest case for option B?"
The leader who is never wrong, never uncertain, and never challenged trains the team that these behaviors are unacceptable. This is the single most impactful lever — and the one most within the leader's direct control.
Teams need clarity on what kinds of failures are acceptable. Without explicit norms, people default to hiding all failures.
Distinguish:
Communicate this distinction explicitly in team retrospectives and in post-incident conversations:
"I want us to report near-misses and small failures before they become large ones.
That requires people to feel safe raising them. I will never penalize someone
for surfacing a problem early. I will be concerned if I find out about a problem
late because people didn't feel they could raise it."
Celebrate instances of early problem reporting — not just successes.
The single behavior that most rapidly destroys psychological safety is ignoring or dismissing dissent. Team members observe every response to a dissenting view and calibrate their own future willingness to speak up.
When someone raises a concern or disagrees, regardless of whether you agree:
✅ "That's an important point. I hadn't considered that dependency.
Let me factor that into the plan — can you send me the details?"
✅ "I hear the concern about the timeline. I'm going to keep the date because
of the customer commitment, but I want to flag that we've identified a risk
and we'll revisit scope if needed."
❌ "We've already decided this." (dismissal — shuts down all future input)
❌ Silence followed by proceeding as before (message received: raising concerns has no effect)
Thanking people for dissent — even when you disagree — is the most visible signal that speaking up is safe.
Active behaviors that suppress psychological safety:
| Behavior | Effect | Replace with |
|---|---|---|
| Publicly criticizing someone for a mistake | Others learn to hide mistakes | Private feedback; public problem-solving |
| Laughing at or dismissing questions | People stop asking | "Good question — let's work through it" |
| Interrupting or talking over people | They stop contributing | Named-turn protocol in meetings |
| Rewarding the person who speaks loudest | Others disengage | Explicitly solicit quieter voices |
| Penalizing estimates that slip | People pad estimates or hide real risk | Separate estimation accuracy from character evaluation |
Punishing behaviors are often unintentional. The most effective way to identify them: after a meeting where little dissent was raised, privately ask a trusted team member: "Was there anything people weren't saying?"
Psychological safety enables speaking up; structure makes speaking up the path of least resistance. Both are needed.
Retrospectives: normalize the discussion of what didn't work — not just what did. A retro where only wins are discussed trains people that problems are unwelcome.
Pre-mortems: before starting a project, ask "Imagine it's six months from now and this project has completely failed. What went wrong?" This makes raising concerns the task, not the exception.
Anonymous input channels: a Slack channel or shared doc where people can raise concerns without attribution. Signals that you want to hear concerns more than you want to know who has them.
Devil's advocate role: designate a rotating team member to argue against the prevailing view in key decisions. This makes dissent structural, not personal.
Meeting protocols: in meetings, explicitly call on people who haven't spoken: "I want to hear from people who haven't weighed in — [Name], what's your read on this?"
Psychological safety changes over time based on team composition, leadership behavior, and organizational pressure. Re-run the survey quarterly.
Interpret the trend, not the score:
A score below 5.0 warrants direct leadership intervention. Common causes of decline: a new leader who is less safe, a recent incident that was handled punitively, or significant team change that reset the safety baseline.
npx claudepluginhub jeffreytse/grimoire --plugin grimoireAssesses and improves dev team effectiveness using psychological safety, dependability, clarity, meaning, and impact model. Includes health checks for collaboration, dysfunction, inclusivity, and culture.
Runs structured team health assessments across delivery, culture, process, and cross-functional dimensions. Use for surveys, retrospectives, or facilitation guides.