reviewing-test-plan — SKILL.md

Variant: standard · When to use: judging a finished QA/verification test plan as an acceptance gate — checking a tester can execute it, every upstream behavior has a traceable case, and the catalog is risk-weighted not combinatorial, then emitting VERDICT: approve|revise with actionable findings.

Overview

This skill is the review half of an authoring/judging test-plan pair. Loaded by a reviewer who holds a finished QA/verification test plan — the document stating what to test, at what level, in what environment, to what done-criteria, plus the case catalog to run — it judges that document against one question: can a tester execute this plan with no questions, does every behavior in the handed-in upstreams have at least one traceable test case, and is the catalog sized by risk (not the full input-permutation cross-product, not skeletal)? It applies a fixed coverage + testability checklist — the same bar a test-plan author produces to, so the produce-bar and the review-bar do not drift — then emits a single machine-parseable verdict plus findings the author can act on in one revision pass. It is an acceptance gate: it does not author, fix, or rewrite the plan; it judges and returns findings, and the producer revises.

The single most load-bearing checks are coverage completeness (enumerate every upstream behavior/operation/error and confirm each has >=1 traceable case — a behavior with no case is the highest-impact defect) and catalog sizing (risk-weighted, not a combinatorial blow-up and not thin/padded). Both are easy to miss when the plan looks thorough.

When to activate

• A finished QA/verification test plan needs an accept/revise decision before testing begins or before it feeds a release runbook's exit criteria.
• You are the independent reviewer / gate for a test plan a producer just authored, and you have the upstreams (feature-spec / api-spec / PRD) to check coverage against.
• Re-judging a revised test plan after a prior revise verdict.

Do NOT activate when: