원클릭으로
qa-verification
Phase 3 of code review: run tests, verify behavior, check edge cases, confirm no regressions.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
메뉴
Phase 3 of code review: run tests, verify behavior, check edge cases, confirm no regressions.
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
| name | qa-verification |
| description | Phase 3 of code review: run tests, verify behavior, check edge cases, confirm no regressions. |
Verify the implementation actually works, not just that the code looks correct.
Work from narrow to broad:
If a full suite is too slow or unavailable, report: command not run, reason, risk, and narrower evidence collected.
For components with user interaction, QA verification includes browser-based testing via Chrome DevTools MCP or agent-browser:
Click Navigation
Keyboard Navigation
Mobile/Touch
Accessibility Audit
mcp_chrome_devtoo_lighthouse_audit with mode="snapshot"## Browser Interaction Results
- Click navigation: [PASS/FAIL with screenshot paths]
- Keyboard navigation: [PASS/FAIL with test sequence]
- Mobile layout: [PASS/FAIL at 375px]
- Lighthouse a11y: [score]
PASS or FAIL with specific test results and evidence.
## QA Results
- Test suite: [X passed, Y failed, Z skipped]
- Edge case coverage: [list from design doc, each PASS/FAIL]
- Regressions: [none / list]
- Error path verification: [PASS/FAIL with evidence]
Use this skill after problem framing to produce a locked architecture document. Creates data flow diagrams, state machines, error paths, test matrices, and bite-sized task breakdowns.
Design + accessibility skill for ASU UI work (WCAG 2.1 AA). Use when designing, building, or reviewing any ASU component or page in Unity or Webspark. Covers semantics/landmarks, keyboard + focus, contrast, reflow at 320px, forced-colors mode, labels/forms, and a pre-delivery checklist. Prefer Unity/Bootstrap components over hand-rolled div+ARIA.
Evaluate code structure, maintainability, and engineering standards alongside adversarial review.
Use during REVIEW to compare a rendered UI component against its design references (screenshots in .intake/ and/or the Figma frame). Drives a running Storybook (Unity) or local Webspark page with agent-browser to screenshot the build at multiple viewports, then reports P1/P2/P3 visual-fidelity findings to .pipeline/visual-report.md. Used by the acp-visual agent.
Use at the END of the Unity stage to enforce the PR-ready gate and write the cross-stage handoff. Defines the gate (build+lint, Storybook story + interaction/a11y tests, HTML-parity) and writes handoff.json (+ markup.html) to the shared peer handoff/ directory, keyed by Jira key or component slug, for the Webspark stage to consume. Validate with .kiro/scripts/validate-handoff.sh.
Use at pipeline INTAKE to turn a component request into an enriched intake brief. A component name is ALWAYS required; a Jira key, a Figma URL, and/or one-or-more screenshots in .intake/ are optional enrichment. Reads Jira via mcp-atlassian (read-only), design specs/tokens via the Figma Dev Mode MCP, and screenshots via vision. Ask clarifying questions when inputs are sparse or conflicting. Produces .pipeline/intake-brief.md for the architect.