QA Engineer

The qa-engineer agent provides structured audit methodology and quality standards that work across any platform. It reads the active platform agent for platform-specific checks.

activate_agent("qa-engineer")

What It Covers

Code review methodology — severity-based review (Critical, Warning, Suggestion, Praise) across 10 categories
Accessibility — WCAG 2.1 AA audit methodology (keyboard nav, ARIA, contrast, motion, forms, semantics, touch targets)
Performance — Core Web Vitals assessment (LCP, CLS, INP thresholds)
SEO — meta tags, structured data, Open Graph, heading hierarchy, image alt text
Visual QA — playwright-cli comparison methodology

How It Works With Platform Agents

The QA Engineer provides the methodology — how to structure a review, grade findings, and present results. The active platform agent provides the specifics — what Liquid patterns to check, what React anti-patterns to flag, what validation command to run.

QA Engineer provides	Platform agent provides
Severity levels (C/W/S/P)	Platform-specific anti-patterns
Review category structure	Component creation rules to check against
Audit workflow (baseline, audit, present, fix, verify)	Validation command (`validate_theme` or `tsc --noEmit`)
WCAG checklist	Platform-specific accessibility patterns
Core Web Vitals thresholds	Platform-specific performance patterns

Code Review

For PR and code reviews, the QA Engineer uses the code-review skill:

get_skill("code-review")

This loads the full review methodology with 10 categories, severity guidelines, output format with clickable file_path:line_number references, and verdict rules.

Commands

Task	Command
Full accessibility audit	`accessibility`
Full performance audit	`performance`
Full SEO audit	`seo`
Comprehensive code + site audit	`audit`
Visual QA against design reference	`refine`
PR / code review	Load `get_skill("code-review")`