≈693 tokens on demand. Your agent reads this file only when SKILL.md points to it.
Retrospective Voice
Purpose: Use this mode when Launch should turn PR and release data into a human, developer-friendly narrative without changing the underlying facts.
Contents
- When to apply
- Voice styles
- Output lengths
- Event-to-narrative mapping
When To Apply
Use when the report should feel like a retrospective, release commentary, internal newsletter, or culture note rather than a pure metric dump.
Voice Styles
| Style |
Tone |
Best for |
| Pragmatic |
Grounded, direct, practical |
Sprint retros, refactoring, hard weeks |
| Passionate |
High-energy, celebratory |
Feature launches, morale posts |
| Analytical |
Structured, data-forward |
Metrics reviews, trend commentary |
Output Lengths
| Length |
Use when |
Shape |
| Short |
Slack or chat update |
1-2 sentences plus one metric |
| Medium |
Team update |
short narrative + numbers + wins + next |
| Long |
Retrospective document |
full story + lessons learned + forward look |
Event-To-Narrative Mapping
| Git event |
Narrative angle |
| Large refactor |
"The Cleanup" |
| Many small PRs |
"The Machine" |
| Hotfix after release |
"The Save" |
| First-time contributor |
"The New Voice" |
| Long-lived branch merged |
"The Marathon" |
| Dependency major upgrade |
"The Migration" |
Guardrails
- Do not invent sentiment that the data cannot support.
- Keep the facts, dates, and metrics accurate.
- Use narrative to explain, not to mask weak signals.
Retrospective Framework Catalogue (2026)
| Framework |
Lens |
Best fit |
| Start / Stop / Continue (Keep/Drop/Start variant) |
Behavior change |
Default low-friction sprint retro; every note maps to a commitment |
| Mad / Sad / Glad |
Emotion |
Surfacing morale and frustration after intense sprints |
| 4Ls (Liked / Learned / Lacked / Longed for) |
Positive-framed learning |
Project end, milestone, or onboarding retro |
| Plus / Delta |
Lightweight pair lens |
Quick post-meeting or release-checkpoint review |
| Sailboat (Wind / Anchor / Rocks / Island) |
Goal-and-blocker narrative |
Cross-functional or quarterly retros |
| Speed Boat |
Hazard-focused variant of Sailboat |
Identifying drag and risks before a launch |
Selection guidance: per agile-coach surveys for 2025, Keep/Drop/Start, Mad/Sad/Glad, and Speed Boat remain the most-cited defaults (kollabe.com/posts/retrospective-formats-compared, retroteam.ai 2025). When narrative voice is layered onto Launch data, pick the framework that matches what the PR record actually shows: a heavy-refactor week pairs with Sailboat (rocks), a launch week pairs with Plus/Delta, and a high-defect-rate week pairs with Mad/Sad/Glad.
