QA Review
Evaluate implementation against technical plan, coding standards, and acceptance criteria.
Inputs
- •
technical-context.md - •
technical-plan.md - •Code change (diff/commit(s)/staged)
Output
{ide-folder}/{outputFolder}/task/{epicNumber}-EPIC-{epicName}/US-{usName}-{usNumber}/qa-{qaNumber}.md
Evidence Policy
- •If you ran checks, list commands + summarize results
- •If not, state why + what should run in CI
- •For calculation stories: Run at least one real-world scenario end-to-end and compare to reference
Required Structure
yaml
Epic ID: EPIC-{epicNumber}
User Story ID: US-{usNumber}
Review ID: QA-{qaNumber}
Status: Pass | Pass-with-issues | Fail
Owner: QA
Reviewed commit(s)/diff: (describe)
Last Updated: (ISO timestamp)
| Section | Content |
|---|---|
| Scope of review | What you looked at |
| Traceability Check | MANDATORY - For each AC-*: Pass/Fail + pointer to tests/code |
| Verification Matrix adherence | Required tests present? Missing? |
| Reference Comparison | MANDATORY when ref exists - Scenario / Expected / Actual / Match |
| Switch Statement Exhaustiveness | MANDATORY for calculation code - List all switches, verify exhaustive |
| Pipeline Test Coverage | MANDATORY for data transformation - Tests use raw inputs? |
| Findings | Functional, code quality, architecture, test quality |
| Issues list | Blocker/Major/Minor/Nit with description, impact, location, recommendation |
| Acceptance recommendation | Accept / Request changes / Reconsider design |
| Follow-ups | Non-blocking suggestions |
Red Flags for Calculation Code
- •Tests only use pre-normalized inputs (not raw formats like CRUSHED_ICE, UNIT)
- •Switch statements with
default: return value(silent passthrough) - •No reference comparison against known-correct values