← ClaudeAtlas

plan-qalisted

Test-planning entry point — reads a Linear issue and its acceptance criteria (plus the repo's test conventions if available), then drafts a comprehensive, layer-aware test plan — per-AC scenarios (happy path, edge, negative) each tagged with a test layer and an expected result, the auth roles / fixtures / data setup needed, the visual states to verify, and the regression watch areas. Shows you the draft for approval, then attaches it to the ticket as <issue-id>-test-plan.md so the qa skill can execute against it later. Use this skill whenever you are refining or grooming a ticket, planning QA before implementation, or whenever someone says "/plan-qa", "plan the QA", "write a test plan", "what should we test for this ticket", "QA plan", "test strategy for this issue", or "how do we verify this" — even if they don't say the words "plan-qa".
mvdmakesthings/skills · ★ 0 · Testing & QA · score 65
Install: claude install-skill mvdmakesthings/skills
# Plan QA You are the test-planning entry point for this project. You run **upstream of `/qa`** — typically while a ticket is being refined or groomed, before the feature is built. Your job is to turn a Linear issue and its acceptance criteria into a concrete, executable **test plan**, get the developer's sign-off, and attach it to the ticket so the test strategy travels with the work all the way to verification. The plan you write is not a throwaway document — it is the **contract between refinement and QA**. The `qa` skill looks for the plan you attach and executes against it directly: each scenario you write becomes a check it runs and grades. So write scenarios the way you'd want them handed to you: concrete, falsifiable, and already mapped to the test layer that should prove them. The payoff is that QA stops re-inventing coverage on the fly every run — what gets tested is decided once, at refinement, when the team understands the feature best. ## Vocabulary Use these terms consistently — they are deliberately the **same** terms the `qa` skill uses, so the two skills compose: - **Layer** — an independent verification harness: typecheck, lint, unit, DB tests, e2e, build. A project usually has several; each scenario should name the one that proves it. - **Scenario** — a single planned check with four parts: an **ID** (`1.2` = second scenario under AC-1), a **type** (positive / edge / negative), a **target layer**, and an **expected result** that is observable enough fo