Install any skill in seconds. Free to start, no credit card required.
Get Started Free →WCAG AAA conformance specialist. Audits web content against WCAG 2.2 Level AAA success criteria that go beyond the standard AA target. Covers enhanced contrast (7:1), extended audio descriptions, sign language, reading level, abbreviations, pronunciation, and focus appearance.
.claude/skills/community-access-wcag-aaa/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 12% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 116% | 0% |
| case-21 | ✗→✓ | ▲ Improved | 81% | 0% |
| case-03 | ✓→✓ | = Same ✓ | 16% | 0% |
| case-04 | ✓→✓ | = Same ✓ | 189% | 0% |
Derived from .claude/agents/wcag-aaa.md. Treat platform-specific tool names or delegation instructions as Codex equivalents.
You audit web content against WCAG 2.2 Level AAA success criteria. AAA represents the highest level of accessibility conformance.
Important: AAA conformance is not required by most legal frameworks. Always confirm the user has achieved AA conformance first — AAA builds on top of AA.
| SC | Name | AAA Requirement | AA Baseline | |----|------|----------------|-------------| | 1.2.6 | Sign Language (Prerecorded) | Sign language for all prerecorded audio | Captions (1.2.2) | | 1.2.7 | Extended Audio Description | Extended audio desc for prerecorded video | Audio desc (1.2.5) | | 1.2.8 | Media Alternative (Prerecorded) | Full text transcript for prerecorded media | Captions + audio desc | | 1.2.9 | Audio-only (Live) | Text alternative for live audio | None at AA | | 1.4.6 | Contrast (Enhanced) | 7:1 for normal text, 4.5:1 for large text | 4.5:1 / 3:1 (1.4.3) | | 1.4.7 | Low or No Background Audio | Speech: no background, or 20dB lower | None at AA | | 1.4.8 | Visual Presentation | User-selectable colors, max 80 chars/line | None at AA | | 1.4.9 | Images of Text (No Exception) | No images of text at all (logotypes only) | Images of text with override (1.4.5) |
| SC | Name | AAA Requirement | |----|------|----------------| | 2.1.3 | Keyboard (No Exception) | ALL functionality via keyboard, no exceptions | | 2.2.3 | No Timing | No time limits at all | | 2.2.4 | Interruptions | User can postpone all interruptions | | 2.2.5 | Re-authenticating | No data loss on session timeout | | 2.2.6 | Timeouts | Warn about inactivity data loss | | 2.3.2 | Three Flashes | No flashing content, period | | 2.3.3 | Animation from Interactions | Motion can be disabled | | 2.4.8 | Location | Breadcrumb or location indicator | | 2.4.9 | Link Purpose (Link Only) | Link text alone conveys purpose | | 2.4.10 | Section Headings | Content organized with headings | | 2.4.12 | Focus Not Obscured (Enhanced) | Focused element fully visible | | 2.4.13 | Focus Appearance | Focus: 2px perimeter, 3:1 contrast |
| SC | Name | AAA Requirement | |----|------|----------------| | 3.1.3 | Unusual Words | Mechanism for jargon definitions | | 3.1.4 | Abbreviations | Mechanism for expanded forms | | 3.1.5 | Reading Level | Lower-secondary level or supplement | | 3.1.6 | Pronunciation | Mechanism for ambiguous words | | 3.2.5 | Change on Request | Context changes only on user request | | 3.3.5 | Help | Context-sensitive help available | | 3.3.6 | Error Prevention (All) | All submissions: reversible, checked, confirmed | | 3.3.9 | Accessible Auth (Enhanced) | No cognitive function test at all |
Confirm AA conformance is achieved or in progress.
Read targets and check each applicable AAA criterion.
List findings by WCAG principle with criterion, description, location, and remediation.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-03 | pass→pass | 34,433 | 34,992 | +2% | 1 | 1 | 0% | 6,219 | 7,228 | +16% | 0 | 0 | — |
case-01 | fail→pass | 35,919 | 33,549 | -7% | 1 | 1 | 0% | 6,227 | 7,003 | +12% | 0 | 0 | — |
case-02 | fail→fail | 34,002 | 31,784 | -7% | 1 | 1 | 0% | 6,230 | 6,722 | +8% | 0 | 0 | — |
case-04 | pass→pass | 3,628 | 4,679 | +29% | 1 | 1 | 0% | 594 | 1,718 | +189% | 0 | 0 | — |
case-05 | pass→pass | 12,789 | 13,399 | +5% | 1 | 1 | 0% | 2,393 | 3,494 | +46% | 0 | 0 | — |
case-06 | pass→pass | 16,860 | 20,682 | +23% | 1 | 1 | 0% | 2,957 | 4,604 | +56% | 0 | 0 | — |
case-07 | pass→pass | 8,056 | 11,236 | +39% | 1 | 1 | 0% | 1,457 | 3,190 | +119% | 0 | 0 | — |
case-08 | fail→pass | 5,721 | 6,288 | +10% | 1 | 1 | 0% | 972 | 2,104 | +116% | 0 | 0 | — |
case-09 | pass→pass | 12,849 | 7,656 | -40% | 1 | 1 | 0% | 2,079 | 2,231 | +7% | 0 | 0 | — |
case-10 | pass→pass | 12,158 | 11,857 | -2% | 1 | 1 | 0% | 2,109 | 2,973 | +41% | 0 | 0 | — |
case-11 | fail→fail | 10,637 | 10,394 | -2% | 1 | 1 | 0% | 1,741 | 2,861 | +64% | 0 | 0 | — |
case-12 | pass→pass | 12,076 | 9,445 | -22% | 1 | 1 | 0% | 1,855 | 2,564 | +38% | 0 | 0 | — |
case-13 | pass→pass | 12,691 | 11,031 | -13% | 1 | 1 | 0% | 2,023 | 2,774 | +37% | 0 | 0 | — |
case-14 | pass→pass | 13,478 | 11,720 | -13% | 1 | 1 | 0% | 2,130 | 2,804 | +32% | 0 | 0 | — |
case-15 | pass→pass | 7,288 | 6,865 | -6% | 1 | 1 | 0% | 1,241 | 2,218 | +79% | 0 | 0 | — |
case-16 | pass→pass | 9,052 | 7,679 | -15% | 1 | 1 | 0% | 1,532 | 2,391 | +56% | 0 | 0 | — |
case-17 | pass→pass | 6,568 | 7,763 | +18% | 1 | 1 | 0% | 1,022 | 2,224 | +118% | 0 | 0 | — |
case-18 | pass→pass | 12,463 | 11,225 | -10% | 1 | 1 | 0% | 1,954 | 2,784 | +42% | 0 | 0 | — |
case-19 | pass→pass | 6,017 | 7,027 | +17% | 1 | 1 | 0% | 907 | 2,085 | +130% | 0 | 0 | — |
case-20 | pass→pass | 9,929 | 11,275 | +14% | 1 | 1 | 0% | 1,655 | 2,897 | +75% | 0 | 0 | — |
case-21 | fail→pass | 8,713 | 9,506 | +9% | 1 | 1 | 0% | 1,525 | 2,757 | +81% | 0 | 0 | — |
case-22 | pass→pass | 13,012 | 12,705 | -2% | 1 | 1 | 0% | 1,975 | 3,053 | +55% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +14 percentage points is the difference between those two pass rates over the 22 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.