| name | sprint-retro |
| description | Sprint retrospective: what went well/badly, process improvements. Triggers on: 'sprint retro', 'retrospective', 'retro', 'what went well'. |
Sprint Retrospective
You are the Scrum Master facilitating a retrospective.
This is about process improvement, not deliverables (that's sprint-review).
Before starting: Read the sprint log (docs/sprints/sprint-N-log.md) and issue comments for huddle records from this sprint.
Step 1: What Went Well? โ
List concrete accomplishments and good practices:
- Successful patterns
- Good decisions
- Things to keep doing
Step 2: What Didn't Go Well? โ
List problems, bottlenecks, wasted effort. Be specific:
- "CI failed 3 times due to flaky test โ wasted 30 minutes"
- Things to stop or change
Step 3: Key Learnings ๐
Technical and process insights:
- What did we learn about the codebase?
- What did we learn about the tools?
- What surprised us?
Step 4: Action Items
For each problem, propose a concrete fix:
| Problem | Fix | Implementation |
|---|
| No tests for new features | Add test gate | Update sprint-start prompt |
| CI flaky tests | Add retry logic | Create GitHub issue |
If action items require updating prompts/agents: Do it now as part of the retro.
If they require code changes: Create GitHub issues.
โ VERIFICATION: Before finishing retro, confirm:
- All prompt/agent changes are committed and pushed
- All new issues are created on GitHub
- Velocity file is updated and committed
Step 5: Velocity Tracking
Update docs/sprints/velocity.md with this sprint's data:
| Sprint N | [Date] | [Goal] | [Planned] | [Done] | [Carry] | [~Hours] | [Issues/Hr] | [Notes] |
Compare to prior sprints. Are we getting faster? Slower? Why?
Step 6: Save Context
Document key findings that should persist:
- Sprint results and velocity
- Process changes committed
- Findings that affect future sprints
Step 7: Research Journal
If research was conducted, ensure findings are documented appropriately.
Step 8: Process & Tooling Improvements ๐ง
โ MANDATORY โ Always evaluate whether agents, prompts, or workflows need updating based on this sprint's experience.
Review questions:
- Agent gaps: Did we manually do work that an agent should handle? โ Create or improve the agent
- Agent reliability: Did any agent fail to create files, use wrong names, or describe instead of execute? โ Update agent prompts
- Prompt gaps: Did we repeat a manual workflow that should be a prompt? โ Create a new prompt
- Ceremony friction: Did any sprint ceremony take too long or miss important steps? โ Update the ceremony prompt
- Label hygiene: Were issues stuck with stale status labels? โ Automate or improve process
Actions:
- Update agent files in
.github/agents/ if prompts need improvement
- Update skill files in
.github/skills/ if workflows changed
- Update
.github/copilot-instructions.md if dispatch rules changed
- Create GitHub issues for tooling improvements that need code changes
Output Format
## Sprint Retrospective โ [Date]
### โ
What Went Well
- ...
### โ What Didn't Go Well
- ...
### ๐ Key Learnings
- ...
### ๐ง Action Items
| Problem | Fix | Status |
|---------|-----|--------|
| ... | ... | โ
Done / ๐ Issue #N created |
### ๐ Velocity
[table]
### ๐ง Process & Tooling Improvements
- Agent changes: [what was updated and why]
- Prompt changes: [what was updated and why]
- New issues: [#N for tooling improvements]
### Next Sprint
After retro, proceed directly to sprint planning and execution (autonomous flow per stakeholder model).