sprint-retrospective

Facilitate effective sprint and team retrospectives that produce actionable insights and genuine improvement. Covers multiple retrospective formats (Start/Stop/Continue, 4Ls, Mad/Sad/Glad, sailboat),

SKILL.md

GitHub

Sprint Retrospective

Facilitate retrospectives that produce real change — not just a list of complaints that gets forgotten.

For production incident post-mortems, use the incident-retrospective skill instead.
This skill is for team/sprint/quarterly retrospectives.


What Makes a Retro Valuable

A retrospective is only worth the action items it produces. Common failure modes:

  • Same themes surface every sprint, nothing changes
  • Action items have no owner or deadline
  • Dominant voices crowd out quieter team members
  • Facilitator skips synthesis — raw sticky notes ≠ insight
  • No check-in on last sprint's actions (accountability vacuum)

The measure of a good retro: Did behavior change?


Before the Retro (Facilitator Prep)

  • Review last sprint's retro action items — which were completed? which weren't?
  • Check sprint metrics: velocity, scope changes, blocked items, on-call load
  • Choose a format that fits the team's current mood and last retro's format
  • Book room / video call with 60-75 min timebox (< 8 people) or 90 min (8-12 people)
  • Set up collaborative board: Miro, FigJam, Retrium, or EasyRetro

Retrospective Formats

Choose based on team context:

Start / Stop / Continue

Best for: new teams, post-process changes, when you want clear directional actions

Column Prompt
Start What should we begin doing that we're not doing now?
Stop What are we doing that isn't working or creates friction?
Continue What's working well that we should keep doing?

4Ls (Liked / Learned / Lacked / Longed For)

Best for: end-of-quarter, after a big delivery, or when team growth is the focus

Column Prompt
Liked What did you enjoy or appreciate this sprint?
Learned What did you learn — about the codebase, the team, yourself?
Lacked What was missing — tools, clarity, support, process?
Longed For What do you wish we had? What would have made this sprint better?

Mad / Sad / Glad

Best for: after a difficult sprint, when morale is low, when emotional acknowledgment matters

Column Prompt
Mad What frustrated you or felt unfair?
Sad What disappointed you or didn't go as hoped?
Glad What are you proud of or grateful for?

Sailboat (Winds / Anchors / Rocks / Sun)

Best for: quarterly planning, when team wants to look forward as well as back

Element Meaning
Wind (🌬️) What's helping us move forward?
Anchor (⚓) What's slowing us down or holding us back?
Rocks (🪨) What risks or obstacles are ahead?
Sun (☀️) What's our goal / what are we sailing toward?

Running the Session

Opening (5 min)

  1. Check in on last sprint's action items: "We committed to X — did we do it?"
  2. Set the tone: "This is a safe space. We improve systems, not blame people."
  3. State the timebox and format

Gather (10-15 min)

  • Silent individual writing first — 5-7 min for everyone to add notes
  • No discussion during this phase; prevents groupthink
  • Encourage specifics: "Deploys were slow" is better than "things were bad"

Group & Theme (10 min)

  • Cluster similar items together (facilitator or team)
  • Dot-vote on themes if too many to discuss: each person gets 3-5 votes
  • Focus discussion on highest-voted clusters

Discuss (20-30 min)

  • One theme at a time — facilitator keeps focus
  • Ask "why" for complaints: understand root cause, not just symptom
  • Draw out quieter voices: "What do you think, [name]?"
  • Time-box each theme: max 8-10 min per topic

Action Items (10-15 min)

  • For each discussed theme, ask: "What's one concrete thing we can change?"
  • Every action item needs:
    • What — specific action, not vague aspiration
    • Who — single named owner (not "the team")
    • By when — specific date or sprint
    • Success criteria — how will we know it's done?

Close (5 min)

  • Read back the action items; get verbal confirmation from each owner
  • Rate the retro: 1-5 thumbs (was this a good use of time?)
  • Thank the team

AI-Assisted Theme Analysis

Paste your raw sticky notes into an agent with this prompt:

Here are the raw retrospective notes from our sprint retro.
Please:
1. Group them into 3-5 themes with descriptive names
2. For each theme, summarize what the team is experiencing
3. Identify the top 2-3 themes by volume/sentiment
4. For each top theme, suggest 1-2 concrete, actionable improvements
   the team could try next sprint
5. Flag any items that indicate a systemic issue needing escalation

Notes:
[paste sticky notes here]

Action Item Template

## Retro Action Items — Sprint [N] — [Date]

| # | Action | Owner | Due | Success Criteria | Status |
|---|---|---|---|---|---|
| 1 | [specific action] | [name] | [sprint N+1 / date] | [measurable outcome] | ⬜ Open |
| 2 | | | | | ⬜ Open |

### Review at next retro:
- Item 1: [completed / in progress / dropped — with reason]

Action Item Quality Check

Before the retro ends, test each action item:

Test Bad example Good example
Is it specific? "Improve communication" "Add a #deploys Slack channel; post before + after every prod deploy"
Does it have one owner? "The team will..." "Priya will..."
Is it bounded? "Eventually fix the flaky tests" "Priya will fix the top 3 flaky tests by end of next sprint"
Can you tell if it's done? "Be more proactive" "Run a mid-sprint health check every Wednesday at standup"

Retrospective Health Over Time

Track these signals across retros to detect patterns:

Signal Healthy Warning
Action item completion rate > 70% each sprint < 50% consistently
New themes surfacing Yes — shows psychological safety Same 3 themes every sprint = systemic block
Participation Everyone contributes 1-2 people dominate every time
Retro rating (1-5) 3.5+ average < 3 average = retro format needs change
Items escalated to manager Rare Frequent = problems beyond team's control

When Action Items Repeat

If the same theme shows up 3+ sprints in a row:

  1. Acknowledge it explicitly: "We've flagged this 3 times. It's not getting better with small fixes."
  2. Escalate or time-box: Either escalate to leadership (if it's systemic/resourcing) or time-box a dedicated fix sprint
  3. Consider a focused working session: A 2-hour deep dive on one chronic problem > 6 sprints of 10-min discussion

Quick Formats (< 30 min)

For busy periods or teams that need a lighter touch:

Rose / Thorn / Bud (async-friendly)

  • 🌹 Rose: Something that went well
  • 🌵 Thorn: Something that was difficult
  • 🌱 Bud: Something to try or explore next sprint

One-word check-in + one action Each person: one word describing the sprint + one thing to change. Facilitator synthesizes into 2-3 actions.


References

MCP Integration

When Jira MCP and Confluence MCP are configured, close the loop between retro insights and team execution.

Query completed sprint stories to ground the retrospective in data:

jira_search_issues(jql: "project = PROJ AND sprint in closedSprints() AND sprint = '<sprint_name>'", max_results: 50)

Create action items as Jira tickets:

jira_create_issue(project: "PROJ", issue_type: "Task", summary: "Retro action: <action>", description: "<what, why, success criteria>", assignee: "<owner_username>", labels: ["retro-action"])

Write the retrospective outcomes to Confluence:

confluence_create_page(space_key: "TEAM", title: "<Sprint N> Retrospective", content: "<formatted retro summary with themes and actions>", parent_page_id: "<retros_page_id>")

Check previous retro action items for carry-overs:

jira_search_issues(jql: "labels = 'retro-action' AND status != Done AND project = PROJ")

Graceful fallback: Produce the retrospective summary and action items as Markdown; instruct the team to paste into Confluence and create Jira tickets manually.