qa-systematic
Systematic web application QA testing with issue taxonomy, health scoring, and regression tracking. Triggers on: "QA this", "test the app", "smoke test", "run QA", "systematic test", "regression test", "full QA", "/qa-systematic".
How do I install this agent skill?
npx skills add https://github.com/mathews-tom/armory --skill qa-systematicIs this agent skill safe to install?
- Gen Agent Trust Hubpass
This skill provides a systematic framework for QA testing web applications using browser automation. It analyzes page structure, interactive elements, and console logs. The primary security consideration is the attack surface for indirect prompt injection, as the agent processes untrusted content from the web applications it tests.
- Socketpass
No alerts
- Snykwarn
Risk: MEDIUM · 1 issue
What does this agent skill do?
Systematic QA Testing
Modes
Full (default)
Systematic page-by-page testing, 8-category health score, full issue documentation.
Quick
30-second smoke test of critical paths only: login, main nav, primary action.
Regression
Diff current state against saved baseline, report new/resolved issues.
Browser Automation Detection
Detect available automation in priority order:
- Playwright MCP server — check if Playwright tools are available in the current tool list
agent-browserskill — check if the agent-browser skill is loaded- Direct CLI tools — check for
playwright,puppeteer, orcypressbinaries on PATH - Manual fallback — instruct the user to navigate and report observations
Use the highest-priority method available. State which method is in use at the start of the report.
Workflow
Phase 1: Initialize
- Detect mode from user prompt. Default to full if unspecified.
- Detect application URL:
- Check
references/project-detection.mdfor framework port conventions (e.g., Next.js → 3000, Vite → 5173, Django → 8000). - If not detectable, ask the user.
- Check
- Detect available browser automation method (priority list above).
- If regression mode: load previous baseline from
.qa-reports/.
Phase 2: Authenticate (if needed)
- Navigate to root URL and check if a login wall is present.
- If credentials were provided: authenticate and store session.
- If not: ask the user for test credentials, or skip auth-gated pages and note the gap in the report.
Phase 3: Orient
- Navigate to root URL.
- Map the primary navigation structure — collect all top-level nav links.
- Classify each page: static, form, list, detail, dashboard.
- Build a test plan ordered by page category (forms and dashboards first — highest defect density).
Phase 4: Explore (Full mode)
For each page, run the per-page checklist below. In quick mode, run only the items marked with (Q).
Visual Scan
- (Q) Layout renders correctly — no overlap, no overflow
- Images load — no broken
<img>tags - Typography consistent — no visible font fallbacks
- Responsive: check at desktop (1280px) and mobile (375px) widths
Interactive Elements
- (Q) All buttons and links are clickable and responsive
- Hover states present where expected
- Focus indicators visible for keyboard navigation
- Disabled states visually distinct
Forms
- (Q) Required field validation fires on empty submit
- Error messages display on invalid input
- Success feedback on valid submission
- Form resubmission handled — no duplicate submissions on double-click
Navigation
- (Q) All nav links resolve — no 404s
- Back button works as expected
- Deep links work — direct URL access returns correct page
- Breadcrumbs accurate (if present)
State Management
- Loading states displayed during async operations
- Empty states handled — no blank pages when data is absent
- Error states recoverable — retry or back options present
- Data persists across navigation — no lost form data on back/forward
Console
- (Q) No JavaScript errors in console
- No failed network requests (4xx/5xx)
- No mixed content warnings
- No deprecation warnings in hot paths
Responsiveness
- Mobile layout usable — no horizontal scroll at 375px
- Touch targets >= 44px
- Text readable without zoom (>= 16px body text)
Phase 5: Document
For each issue found, classify using references/issue-taxonomy.md:
- Severity: critical (blocks usage), major (degrades experience), minor (cosmetic/polish)
- Category: functional, visual, accessibility, performance, content, navigation, security, console
- Evidence: screenshot description or reproduction steps
Assign a unique ID: QA-001, QA-002, etc.
Compute health score using the weights defined below and detailed in references/report-template.md.
Phase 6: Wrap Up
- Generate structured report following
references/report-template.md. - Save to
.qa-reports/<YYYY-MM-DD>-<mode>.json. - If full mode: save baseline for future regression comparison.
- Present summary: health score, critical/major/minor counts, top 3 priority fixes.
Health Score
Weighted average across 8 categories, scored 0-100.
| Category | Weight |
|---|---|
| Console errors | 15% |
| Broken links | 10% |
| Functional | 20% |
| UX/Usability | 15% |
| Accessibility | 15% |
| Visual | 10% |
| Performance | 10% |
| Content | 5% |
Scoring per category: start at 100, deduct per issue by severity:
- Critical: -30
- Major: -15
- Minor: -5
Floor at 0. Final health score = weighted sum of category scores.
Quick Mode Behavior
Run only items marked (Q) in the Phase 4 checklist. Skip health score computation — report pass/fail per critical path. Target completion: 30 seconds of actual testing time.
Regression Mode Behavior
- Load the most recent baseline from
.qa-reports/. - Run full mode.
- Diff issues by ID and description similarity.
- Report: new issues, resolved issues, persistent issues.
- Save updated baseline.
How can the creator link this skill?
Add the canonical catalog link to the repository README so users can inspect current installs and available audits. The publishing guide covers the complete discovery path.
<a href="https://skillzs.dev/skills/mathews-tom/armory/qa-systematic">View qa-systematic on skillZs</a>