원클릭으로
assuring-quality
Use when the user names final QA as the goal — pre-release verification, "上線前驗證 / 最終品質保證".
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
메뉴
Use when the user names final QA as the goal — pre-release verification, "上線前驗證 / 最終品質保證".
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Use when starting project code-improvement work — refactoring, architecture improvement, cleanup of an existing codebase — or when a grilling session should leave docs behind. Runs the interview and records decisions as ADRs and domain terms in CONTEXT.md as they crystallise.
Use when the user requests development work — implementing a feature, debugging an error, designing an approach, or deploying a release. MUST be invoked as the entry point before any implementation, debugging, or design sub-skill.
A relentless interview to sharpen a plan or design.
Grill the user relentlessly about a plan, decision, or idea. Use when the user wants to stress-test their thinking, or uses any 'grill' trigger phrases.
Use when the user asks to check, audit, or review something thoroughly. Builds a check plan, then verifies each item with the investigating skill.
Use when the user asks to research or investigate a question, claim, or topic in depth. Comprehensive investigation with current information, every claim traceable to a source.
| name | assuring-quality |
| description | Use when the user names final QA as the goal — pre-release verification, "上線前驗證 / 最終品質保證". |
Final QA pipeline. Exploratory testing surfaces issues; TDD fixes them; loop until every discovered item is green.
The discovery doc is the single source of truth for progress. If working memory disagrees with the doc, trust the doc.
Locate or create the discovery doc before probing anything.
Find the project's docs directory in priority order: docs/qa/, docs/testing/, docs/.
Create docs/qa/ if none of them exist.
Open or create discover-qa-<YYYY-MM-DD>.md in that directory.
Initialise it with this structure:
# Discover QA: <subject> — <date>
## Scope
<what is being QA'd, version/branch, environment>
## Findings
- [ ] F1: <observed behaviour> — <repro steps>
- [ ] F2: ...
## Resolved
<move completed items here with test name and fix commit>
Walk the primary user flows. Probe edge cases, error paths, boundary inputs, concurrency, permissions, data consistency.
Append every unexpected behaviour as a new finding immediately. Do not filter — record first, judge later.
Phase 1 is complete only when every observation is recorded in the doc and the planned scope is covered.
For the next open finding in the doc:
Do not modify production code in this phase.
No test framework available? Write a minimal reproduction script and treat its non-zero exit as red.
Findings to Resolved with: test name, fix commit hash, one-line summary.Findings.The loop ends only when Findings is empty.
Report at end:
digraph discover_qa {
rankdir=TB;
node [shape=box];
start [label="Start QA", shape=doublecircle];
p1 [label="Phase 1: Exploratory\nrecord findings in doc"];
p2 [label="Phase 2: Infer\nstate root-cause hypothesis"];
p3 [label="Phase 3: Red-light\nfailing test pins the bug"];
p4 [label="Phase 4: Fix\nminimum change"];
p5 [label="Phase 5: Green-light\nnew + full suite pass"];
p6 [label="Phase 6: Update doc\nmove to Resolved"];
check [label="Findings empty?", shape=diamond];
new_finding [label="New failure\nfrom Phase 5?", shape=diamond];
done [label="Report + done", shape=doublecircle];
start -> p1 -> p2 -> p3 -> p4 -> p5 -> new_finding;
new_finding -> p1 [label="yes: append"];
new_finding -> p6 [label="no"];
p6 -> check;
check -> p2 [label="no"];
check -> done [label="yes"];
}