",
description_html: "From Sprint N retrospective.
",
priority: "high",
labels: [""],
assignees: [""],
target_date: ""
})
7. Save retro notes:
create_project_page({
project_id: "",
name: "Retro — Sprint N (YYYY-MM-DD)",
description_html: "Sprint Metrics
...Start
...Stop
...Continue
...Action Items
..."
})
```
## Sprint Close Workflow
### Step-by-step
```
1. Complete the sprint review (above)
2. Handle incomplete items:
Option A: Transfer to next sprint
- Create next cycle first (if not exists):
create_cycle({
project_id, name, owned_by, start_date, end_date
})
- Transfer:
transfer_cycle_work_items({
project_id,
cycle_id: "",
new_cycle_id: ""
})
Option B: Move back to backlog
- Remove from cycle (items return to backlog)
- Don't carry over items that weren't started — reprioritize
3. Archive the sprint:
archive_cycle({
project_id: "",
cycle_id: ""
})
4. Record velocity:
Note completed_points for velocity tracking
(Use velocity-metrics skill for historical analysis)
```
### Decision: Transfer vs Return to Backlog
| Situation | Action |
|-----------|--------|
| Item was in-progress, nearly done | Transfer to next sprint |
| Item was not started | Return to backlog, reprioritize |
| Item was blocked all sprint | Return to backlog, resolve blocker first |
| Item scope changed significantly | Return to backlog, re-estimate |
## Action Items Best Practices
1. **Limit to 2-3 actions per retro** — more won't get done
2. **Assign each to a specific person** — "the team" is nobody
3. **Set a deadline** — by end of next sprint
4. **Track completion** — review at start of next retro
5. **Make them specific** — "Improve code reviews" → "Every PR must have review within 4 hours"
6. **Label them** — use "retro-action" label for easy filtering
## Retro Health Metrics
Track these across sprints:
| Metric | Healthy | Warning |
|--------|---------|---------|
| Sprint completion rate | ≥ 80% | < 70% |
| Action items completed | ≥ 80% | < 50% |
| Velocity trend | Stable or improving | Declining 3+ sprints |
| Scope changes mid-sprint | ≤ 10% | > 20% |
## Best Practices
1. **Review before retro** — discuss what was built, then how to improve
2. **Time-box the retro** — 45 min for 1-week sprint, 60 min for 2-week
3. **Everyone speaks** — go around the room, no silent members
4. **Focus on process, not people** — "the deploy process failed" not "John broke the deploy"
5. **Celebrate wins** — acknowledge completed items and improvements
6. **Follow through** — action items without follow-up erode trust in the process