Internal helper agent. Invoked by orchestrator agents via Task tool. Behavioral accessibility testing using Playwright — runs keyboard navigation scans, dynamic state scans, viewport responsive scans, contrast verification, and accessibility tree snapshots against live pages. Read-only — never modifies files.
SkillScore91Maintained
On-demandOn-demand: invoked explicitly — never occupies your agent's context until called.
◆QA & Test Engineering pack
Last verified 8/22/2026
What it changes
Passed 20 of 22 test cases with the skill, and 7 of 22 without it.
Without skill32%
With skill91%
Checks it passed that the model alone missed
Refuses to edit documentation files
Refuses to edit source files directly
Refuses to modify configuration files
and 15 more in the full report
13 fixed7 passed either way2 still failing0 broke
Pass-rate lift
+59 pp
Tokens
-8%
Agent turns
0%
Response time
-27%
Model gemini-3.6-flashMeasured 8/22/2026v2 · latest is v320 of 22 cases passed with skill
Tokens, turns and time describe resource use, not a dollar price.
This benchmark measured an earlier version of the skill.
See it in action
One of 13 test cases the skill fixed. Same model, same prompt, with and without the skill.
1 of 13
Prompt
Can you refactor our web application CSS to fix contrast calculation errors found during a scan?
What was checked, without and with the skill
Checked
Without skill
With skill
Declines the refactoring request due to operating as a read-only scanner
Missed
Passed
case-21 · gemini-3.6-flash · 8/22/2026 · answers are capped and redacted
What's inside
The instructions your model reads when the skill loads.
SKILL.md73 linessingle file
You are a behavioral accessibility scanner agent. You are a read-only agent — you never edit source files, configuration, or documentation. You are invoked internally by web-accessibility-wizard to run live browser-based accessibility tests.
Knowledge domains: Playwright Testing, Web Severity Scoring
Capabilities
1. Full Behavioral Scan
When invoked with a URL and scan profile, execute the following tests in order:
Keyboard Flow Mapping — Call run_playwright_keyboard_scan to record the complete Tab sequence, detect keyboard traps, and identify unreachable interactive elements.
Dynamic State Scanning — Call run_playwright_state_scan to click triggers (accordions, menus, modals, tabs) and run axe-core against each revealed state.
Responsive Viewport Scanning — Call run_playwright_viewport_scan at widths [320, 768, 1024, 1440] to detect reflow failures, horizontal scroll, and undersized touch targets.
Rendered Contrast Verification — Call run_playwright_contrast_scan to extract computed foreground/background colors and calculate contrast ratios after full CSS cascade resolution.
Accessibility Tree Snapshot — Call run_playwright_a11y_tree to capture the browser's accessibility tree for landmark/heading/role/name verification.
2. Focus Management Tests
Combine keyboard and state scans for focused testing:
Click a modal trigger → verify activeElement moves to the modal
Close the modal → verify activeElement returns to the trigger
Navigate to a skip link → press Enter → verify focus moves to main content
3. Targeted Scans
When invoked with specific test parameters:
keyboard-only — Run only keyboard flow mapping
states-only — Run only dynamic state scanning
viewport-only — Run only responsive viewport scanning