klimkit-checklister

Klimkit Checklister

Use this before implementation work. The output is an agent-authored task note with observable acceptance criteria, not a loose plan.

Workflow

1. Read the user request, repo instructions (AGENTS.md), relevant docs/work/ LOGs, relevant docs, and enough nearby code to understand the real acceptance surface.
2. Resolve the active work folder. If the repo has no docs/work/ layout, use klimkit-setup first.
3. Create or update docs/work/<NNN-DDMMYY-slug>/<phase>/NNN-acceptance-checklist.md inside the current work folder's active phase.
4. Add an Acceptance Checklist section using checkbox bullets.
5. Make every item observable: a file exists, a command passes, a behavior is visible, a release points to a commit, a proof report renders, or an explicit risk is named.
6. Include scope boundaries and skipped work. For prototypes, require the prototype question, run command, and deletion-or-absorption decision.
7. Include verification expectations that match the blast radius:
   • UI: screens, states, screenshots/video, report path, and report URL expectations.
   • Backend/API: request/response shape, validation, auth, errors, logs, and tests.
   • Persistence/sync: files, database rows, caches, queues, services, restart behavior, and cross-machine effects.
   • Security-sensitive work: require klimkit-security-auditor.
8. Include reflection/final-review expectations for non-trivial work: klimkit-reflector after verification and klimkit-final-reviewer before completion.
9. End with the checklist path and any unresolved scope questions.

Treat checklist items as blocking unless the user explicitly changes scope.