Use this to plan and facilitate a retrospective, whether for your own core team or a client team you are coaching. It picks a format for the team's maturity, builds a timed facilitation script you can run from, and (if you have raw input) synthesizes it into themes and action items. For synthesizing raw retro output after the session, use /retro-synthesize.

Framework attribution: Retrospectives are where a balanced team lives its values -- share experiments, celebrate failure, welcome diverse voices. Kate ran weekly retros as part of the Pivotal team rhythm. See playbook/balanced-teams.

Related resources: retro-facilitation-guide.md -- 4-step Quick Start Guide for running retrospectives (standalone, printable).

How it works

1. You pick the mode (your own team, or a client team), then provide the team name, sprint or period, any raw retro input (feedback, sticky notes, survey data), and optionally a preferred format
2. The skill selects the best retro format for the team's maturity, synthesizes input into themes, and builds a facilitation script
3. It returns a complete facilitation prep doc with discussion script, themed analysis, and prioritized action items

Prompt

You are preparing retrospective facilitation materials. Your job is to give the facilitator a script they can walk into the room with. Before writing, read knowledge/voice-tone-guide.md -- use the internal voice.

Mode (pick one):

• Own team -- the practitioner is planning a retro for her own core delivery team (PM, Design, Eng, no outside stakeholders). Lower ceremony, faster, can lean on shared history. This is the default when the team is internal.
• Client team -- the practitioner is coaching a client team who may have never run an effective retro. More scaffolding, explicit ground rules, format reasoning stated so she can override.

The two modes share the same script structure; the mode shifts the framing, the amount of scaffolding, and whether stakeholders are in the room (own-team retros are squad-only by default, add stakeholders later).

Inputs I will provide:

• Mode: {{MODE}} (own team or client team -- default own team if not stated)
• Team: {{TEAM}} (the team running the retro)
• Sprint or period: {{SPRINT_OR_PERIOD}} (e.g., "Sprint 14", "Q2 2026", "past 6 weeks")
• Retro input (optional): {{RETRO_INPUT}} (raw feedback -- sticky notes, survey responses, Slack threads, previous retro notes)
• Format (optional): {{FORMAT}} (preferred retro format, or leave blank for a recommendation)

Step 1: Select the retro format Reference knowledge/pm-execution-templates.md for the three retro formats. If the user specified a format, use it. If not, recommend one based on these guidelines:

Start/Stop/Continue -- Best for:

• Teams new to retros (lowest cognitive load)
• Teams that need simple, actionable takeaways
• When the practitioner doesn't know the team well yet
• Default choice when in doubt

4Ls (Liked/Learned/Lacked/Longed For) -- Best for:

• Teams with some retro experience who are ready for deeper reflection
• Periods with significant learning or change (new tools, new process, team restructuring)
• When the practitioner wants to surface emotional signals alongside tactical ones

Rose/Thorn/Bud -- Best for:

• Teams whose morale is low or who are coming off a hard stretch (it starts positive)
• Surfacing future opportunities (the "bud") alongside what went well and what hurt
• A quick, accessible format that needs no metaphor priming

Sailboat (Wind/Anchor/Rocks/Island) -- Best for:

• Mature teams comfortable with metaphorical thinking
• Strategic retros (end of quarter, end of project, milestone reviews)
• When the team needs to zoom out from sprint-level details
• When the practitioner wants to connect tactical issues to strategic direction

State your recommendation and reasoning so the practitioner can override if she knows the team better.

Step 2: Synthesize retro input (if provided) If raw feedback was provided:

• Group similar items into 3-5 themes
• Identify the most frequently mentioned topics
• Note sentiment patterns (frustration, energy, confusion, pride)
• Pull representative quotes that capture each theme
• Flag any items that are about individuals rather than process (the practitioner should handle these privately, not in the group retro)

If no raw input was provided, build the facilitation script with placeholder prompts the practitioner can use to elicit feedback live.

Step 3: Build the facilitation script

Facilitation Script

Opening (5 min)

Optional warm-up (2-3 min): Before diving into the retro format, consider a brief opener to shift the room's energy:

• One-Word Check-In: Each person shares one word for how they're arriving. No explanation. Surfaces real energy and gives quiet voices an early win.
• Rose/Thorn (standing): Everyone stands. Share one thing that went well (rose) and one that was hard (thorn). The standing format raises energy and signals this is active, not passive.
• Count to 20: The group counts to 20, one person per number, no system. If two people speak at once, restart from 1. Requires listening and yielding -- exactly what a good retro needs.

Related skills: See /workshop-warm-up for the full warm-up library.

Psychological safety primer (2 min, before any silent writing): Explicitly name that half-formed ideas are welcome, and model imperfection: "This is my first stab, it is probably wrong, and I need your help." The inner critic gets louder as people gain seniority, so the most senior people in the room are often the ones most likely to self-censor. Name that dynamic so the room gives itself permission to think out loud. This matters most in own-team mode, where shared history can make people assume the problems are already understood.

• Set the tone: "We're here to improve how we work, not to assign blame"
• Ground rules (one speaker at a time, assume good intent, focus on process not people)
• Quick context: what happened this sprint/period (key milestones, challenges, wins)
• If this team has never done a good retro, add the practitioner's framing: "A retro isn't a status meeting. We're not reporting what happened -- we're deciding what to change."

Gathering Input (15 min)

[Format-specific prompts and questions for each category]

• Suggested time per category
• Facilitation tips for getting quiet team members to contribute
• How to handle dominant voices without shutting them down

Theming & Discussion (15 min)

[If input was provided, present the themed analysis] [If not, instructions for live clustering]

• Discussion questions for each theme
• How to move from symptoms to root causes
• When to go deeper vs. when to move on

Action Items (10 min)

• Limit to 2-3 items maximum (more won't get done)
• Each action item must have: specific action, owner, deadline, success metric
• Review previous retro actions if available (were they completed?)

Closing (5 min)

• Quick round: one word for how this retro felt
• Confirm action items and owners
• Set the next retro date

Themed Analysis

[If retro input was provided]

Suggested Action Items

Facilitator Notes

• Things to watch for with this specific team
• Sensitive topics to handle carefully
• Carry-over items from previous retros (if known)
• Suggested follow-up cadence for action items

Step 4: Run it live (board-native and AI-assisted)

The script above is tool-agnostic. In 2026 most retros run on a shared board with AI doing the first pass, so adapt:

• Miro/FigJam board template: Set up columns matching the chosen format (Start/Stop/Continue, the four Ls, or the Sailboat zones) before the session. Use a sticky color legend with one color per category plus a distinct color for action items. Pre-label a "Previous actions" parking area and an "Action items" column so nothing gets lost in the noise.
• AI first-pass clustering: Let the board AI (Miro's Agile Coach Sidekick or FigJam's sort-into-themes) group the stickies while the room watches, then curate the clusters together. The AI proposes themes, the team confirms or splits them. This is the theming step accelerated, not replaced. The discussion about why a theme exists is still the team's work.
• AI notetaker for capture: Bring a notetaker into the call so action items, owners, and decisions are captured and pushed to the tracker before people leave. Review the captured actions on screen before closing. The principle holds: human-led, tech-enhanced.

For a live run-sheet with facilitator cues and difficult-dynamic handling, point the practitioner at the Facilitator's Kit: pick "retrospective," flag the dynamics expected (a dominant voice, a skeptic, hard news), and print or present the run-sheet.

Step 5: Set the follow-through cadence

A retro that produces action items but no follow-through teaches the team that retros do not matter. Before closing:

• Assign each action an owner and a due date, and put it where the team already looks (the tracker, not the retro board).
• Add the actions to the next few standups as a recurring check ("two of three from last retro are done").
• Open the next retro by reviewing the previous actions first. Completion rate is the real retro metric, not sticky-note volume.

For post-engagement retrospectives (the practitioner's own review of an engagement), use the engagement-retro skill. This skill is for facilitating client team retros.