Storytime Retrospective
Run a retrospective on a completed Storytime spec by reconvening the original team and comparing outcomes against the plan.
Arguments
The topic to retrospect: $ARGUMENTS
Process
- Load the original spec:
- Read
specs/.storytime/sessions/<topic>/team.md— the original team - Read
specs/.storytime/sessions/<topic>/plan.md— what was planned - Read
specs/.storytime/sessions/<topic>/icebreaker.md— original constraints
- Read
- Survey current state:
- Launch Explore agent to check what was actually implemented
- Compare implementation against the plan's success criteria
- Check if non-goals are still valid or need revisiting
- Load personas:
- Reconvene the original team (from cohort + specialist files)
- Each persona evaluates from their domain perspective
- Run the retrospective conversation:
- What went as planned?
- What diverged and why?
- What did we learn?
- What should change for next time?
- Timestamp audit:
- Check all session documents for universal frontmatter (
type,created,session) - If timestamps are missing (common for pre-timestamp-principle docs), backfill from git history and filesystem evidence
- Mark inferred timestamps with confidence level (git-derived, approximate, estimated)
- See
${CLAUDE_PLUGIN_ROOT}/docs/timestamps.mdfor backfill rules - Review
last_reviewedon any decisions from this session — update if still valid
- Check all session documents for universal frontmatter (
- Update artifacts:
- Add a chapter to
specs/.storytime/sessions/<topic>/changelog.md - Update persona files with lessons learned (include
evolved[]entry with date) - Update persona
last_activedates - Flag any non-goals that should now become goals
- Update decision
last_revieweddates for decisions confirmed during retro
- Add a chapter to
Output
Write specs/.storytime/sessions/<topic>/retrospective.md with the team's analysis.
Include universal frontmatter (type: retrospective, created, session).
Update persona files and decision log.
Fill out the evaluation scorecard — see
${CLAUDE_PLUGIN_ROOT}/skills/storytime/references/evaluation-scorecard.md.
Present the card at the end of the retro for the user to review/fill in.
The scorecard is the structured self-check that builds the confidence
corpus across runs.