Use this skill to close a delivery with objective verification, remaining risks, and next steps.
Initial context received via slash: $ARGUMENTS
If $ARGUMENTS is filled, use as reference (e.g., plan, story, issue).
If empty, try to identify the active plan or ask.
Objective
- Consolidate what was delivered against the original plan
- Record executed verifications and actual results
- Make remaining risks and pending items explicit
- Close the delivery with clear handoff
Process
1. Identify the delivery
- Which plan, story, or epic is being closed?
- Read the plan and compare with current state
2. Compare plan vs result
- What was delivered: completed tasks (reference the checklist)
- What was left pending: incomplete tasks and reason
- Scope changes: what changed during implementation
3. Execute verifications
Run and record actual results:
bun run lint— resultbun run typecheckortsc --noEmit— resultbun test— result (if tests exist)- Manual validation — what was verified and how
4. Record risks and next steps
- Remaining risks (what could go wrong)
- Next steps (what still needs to be done)
- Handoff (who needs to know about this delivery)
Where to save
planning/<initiative>/post-impl.mdorplanning/<initiative>/post-impl-YYYY-MM-DD.md- If it's a standalone plan: present inline
Chaining
- If there are next steps that become new stories: suggest
/storyor/plan - If the period ended: suggest
/retro - If part of an epic: update story status in the epic
Reference template
Use ~/.agents/templates/post-implementation-report.md as base.
Rules
- Always run lint, typecheck, and tests. Don't assume they pass.
- Report actual results, not intention. If lint failed, say it failed.
- If tasks were left pending, explain why.
- The report must allow another person to understand the delivery state without additional context.
Relationship with the flow
flowchart LR
A[plan] --> B[execution]
B --> C[post-impl]
C --> D[retro]