WCAG 2.1 A and AA · 48 of 50 success criteria covered · 1.2.3 and 1.2.4 not covered

Requirement matrix with evidence levels

This page shows what the system can prove automatically, what has partial machine support, and what still needs human review. We cover 48 of the 50 A/AA success criteria, with the evidence level shown per criterion. Machine findings are based on rules, DOM, CSS, HTTP and measured browser behavior, not AI.

Test a website

48Success criteria covered

21Deterministic machine findings

23Partial machine support

4Manual checks only

92%Machine-supported review before expert review

Status explanations

Deterministic machine findings

When the tool finds an issue here, the finding should be technically verifiable in the DOM, CSS, HTTP or actual browser behavior. Judgement-based parts are marked for expert review.

Partial machine support

The machine finds concrete signals or performs a bounded measurement, but the whole requirement cannot be decided safely without expert review. These are shown as 'needs review', not red machine evidence.

Manual checks only

The requirement needs human judgement or domain/media understanding, for example media, flashing or legal/financial transactions.

This runs in every scan

Several engines look at the same page. They never count the same violation twice - if two of them find the same thing, that strengthens the evidence instead.

axe-core

The industry standard for automated rule testing.

Our own rules and behaviour tests

We tab through the page in a real browser: visible focus, 200 % text, 320 px reflow and target size are measured, not guessed.

QualWeb (W3C ACT rules)

The official ACT rules run independently and corroborate our findings. Corroborated findings are marked in the report.

v.Nu (Nu Html Checker)

HTML validation. Catches parser errors such as duplicate ids and unclosed elements - errors the other layers do not see.

veraPDF (PDF/UA)

PDFs you link to are checked for tagging, language, title and text layer. Reading order and content still need a human.

Shadow DOM guard

Says so when part of the page is closed to inspection, instead of letting it pass in silence.

The scan uses no language model, and your page is not sent to any third party.

1.1.1 A Non-text content

✓ Machine evidence

Checks: Machine evidence: missing alt via axe/custom.

1.2.1 A Audio-only and video-only (prerecorded)

○ Manual

Checks: Coverage of the text alternative = manual

No automated check. Requires manual review.

1.2.2 A Captions (prerecorded)

◐ Partial

Checks: Behaviour: is there a <track kind=captions>? Quality/sync = manual

Manual review recommended before conclusion.

1.2.5 AA Audio description (prerecorded)

○ Manual

Checks: Coverage of the audio description = manual

No automated check. Requires manual review.

1.3.1 A Info and relationships

✓ Machine evidence

Checks: Machine evidence: heading/list/table/ARIA semantics via axe/custom.

1.3.2 A Meaningful sequence

◐ Partial

Checks: Behaviour: CSS reorders (order/*-reverse) AND the geometry confirms the text sequence actually changed.

Manual review recommended before conclusion.

1.3.3 A Sensory characteristics

◐ Partial

Checks: Deterministic: an instruction pointing to colour/shape/position alone, except where the element is named.

Manual review recommended before conclusion.

1.3.4 AA Orientation

✓ Machine evidence

Checks: Behavior: does the page lock orientation (CSS/JS)?

1.3.5 AA Identify input purpose

✓ Machine evidence

Checks: Behavior: correct autocomplete on name/email/phone fields.

1.4.1 A Use of color

◐ Partial

Checks: Behaviour: text links with no underline or other distinction than colour (color-only-link)

Manual review recommended before conclusion.

1.4.2 A Audio control

✓ Machine evidence

Checks: Behavior: audio autoplays for >3s without pause/control

1.4.3 AA Contrast (minimum)

✓ Machine evidence

Checks: axe: text contrast 4.5:1 / 3:1.

1.4.4 AA Resize text

✓ Machine evidence

Checks: Behavior: 200% text zoom → clipping/loss of content.

1.4.5 AA Images of text

◐ Partial

Checks: SVG text is read directly (inline and <img src=*.svg>, fetched outside CORS): a visible sentence drawn as an image with no other graphics = violation; text inside a chart/illustration = should be checked; logo/wordmark and hidden text exempt. Raster images with text: model layer from the screenshot only.

Manual review recommended before conclusion.

1.4.10 AA Reflow

✓ Machine evidence

Checks: Behavior: real 320 px test → horizontal scroll?

1.4.11 AA Non-text contrast

◐ Partial

Checks: Measures border/surface contrast for CUSTOM controls (ui-component-contrast, 3:1) + forced-colors. axe 4.10 has no 1.4.11 rule. Graphics/icons are not measured yet

Manual review recommended before conclusion.

1.4.12 AA Text spacing

✓ Machine evidence

Checks: Behavior: inject WCAG text-spacing CSS → clipping/overlap?

1.4.13 AA Content on hover or focus

◐ Partial

Checks: Behavior: measured by hovering every focusable element – new content must be hoverable, persistent and dismissible (Escape or documented keys) when it covers other content. Without scripts on the page a missing dismissal is proven; with scripts = should be checked. title attribute = should be checked (the browser's own tooltip is exempt). JS tooltips delayed more than 180 ms are not caught.

Manual review recommended before conclusion.

2.1.1 A Keyboard

✓ Machine evidence

Checks: Behavior: real Tab walkthrough: is all interactive content operable without a mouse?

2.1.2 A No keyboard trap

✓ Machine evidence

Checks: Behavior: can focus always leave each component?

2.1.4 A Character key shortcuts

◐ Partial

Checks: Heuristic: global single-character shortcuts with no visible off/on/remap must be confirmed

Manual review recommended before conclusion.

2.2.1 A Timing adjustable

◐ Partial

Checks: Behaviour: detects timeouts/meta-refresh; adjustability = partly manual

Manual review recommended before conclusion.

2.2.2 A Pause, stop, hide

✓ Machine evidence

Checks: Behavior: measured in the DOM (Web Animations API) and confirmed with two screenshots: a running animation that actually changes position/visibility of information, over 5 s or forever, with no pause/stop/hide control = violation. Stops under reduced motion = should be checked. <marquee> = infinite. JS-driven movement (setInterval) is not measured.

2.3.1 A Three flashes or below threshold

○ Manual

Checks: Flash analysis of video/animation = manual

No automated check. Requires manual review.

2.4.1 A Bypass blocks

✓ Machine evidence

Checks: axe/behavior: analysed in the rendered DOM across pages – a visible block repeated on 2 or more pages of the same site without a skip link with an existing target, main landmark or heading (h1–h6) = violation. A single page alone: skip link/landmarks = should be checked.

2.4.2 A Page titled

✓ Machine evidence

Checks: Machine evidence: <title> exists, via axe.

2.4.3 A Focus order

◐ Partial

Checks: Behavior: measured on the real Tab sequence (wrap-around from autofocus is separated from interleaving) – focus leaving a fieldset/form group and coming back = violation; the same pattern in section/nav = should be checked. Positive tabindex = should be checked (not a violation in itself).

Manual review recommended before conclusion.

2.4.4 A Link purpose (in context)

◐ Partial

Checks: Deterministic: the same link name pointing to different targets, and vague link text without distinguishing context.

Manual review recommended before conclusion.

2.4.5 AA Multiple ways

◐ Partial

Checks: Behaviour: are there at least two ways to the content (menu + search or sitemap)

Manual review recommended before conclusion.

2.4.6 AA Headings and labels

◐ Partial

Checks: axe: structure + deterministic: the same label on two fields in the same form, and placeholder headings.

Manual review recommended before conclusion.

2.4.7 AA Focus visible

✓ Machine evidence

Checks: Behavior: real keyboard focus gives a visible indicator.

2.5.1 A Pointer gestures

◐ Partial

Checks: Behaviour: gesture-based navigation without a simple alternative; edge cases manual

Manual review recommended before conclusion.

2.5.2 A Pointer cancellation

◐ Partial

Checks: Behaviour: action on the down-event (should be up/undoable)

Manual review recommended before conclusion.

2.5.3 A Label in name

✓ Machine evidence

Checks: axe label-content-name-mismatch (experimental, enabled explicitly). Machine evidence only when visible text and name share no word; otherwise should be checked (hand-verified 15.09)

2.5.4 A Motion actuation

◐ Partial

Checks: Behaviour: motion-actuated function; alternative/off switch = partly manual

Manual review recommended before conclusion.

3.1.1 A Language of page

✓ Machine evidence

Checks: Machine evidence: <html lang> exists and is valid, via axe.

3.1.2 AA Language of parts

◐ Partial

Checks: Offline language detection (py3langid) per paragraph, never within Scandinavian.

Manual review recommended before conclusion.

3.2.1 A On focus

◐ Partial

Checks: Behaviour: focus alone triggers a change of context; edge cases manual

Manual review recommended before conclusion.

3.2.2 A On input

◐ Partial

Checks: Behaviour: typing/selecting changes context unexpectedly; edge cases manual

Manual review recommended before conclusion.

3.2.3 AA Consistent navigation

✓ Machine evidence

Checks: Cross-page: main menu order across pages (deterministic).

3.2.4 AA Consistent identification

✓ Machine evidence

Checks: Cross-page: same link = same text across pages (deterministic).

3.3.1 A Error identification

◐ Partial

Checks: Behaviour: aria-live/role=alert present, and an error message without a suggestion (error-message-without-suggestion)

Manual review recommended before conclusion.

3.3.2 A Labels or instructions

◐ Partial

Checks: axe: field has a label + deterministic: a requirement with no explanation (input-constraint-without-hint)

Manual review recommended before conclusion.

3.3.3 AA Error suggestion

◐ Partial

Checks: Deterministic: a pattern requirement that is never explained, and an error message without a suggestion (read from validationMessage/aria-errormessage).

Manual review recommended before conclusion.

3.3.4 AA Error prevention (legal, financial, data)

○ Manual

Checks: Undo/confirm on important transactions = manual

No automated check. Requires manual review.

4.1.1 A Parsing

✓ Machine evidence

Checks: axe duplicate-id/duplicate-id-active enabled EXPLICITLY (marked obsolete in axe, but the Norwegian supervisory authority profile lists the criterion). Versioned profile choice, see _AXE_NORSK_PROFIL

4.1.2 A Name, role, value

✓ Machine evidence

Checks: Machine evidence: name/role/value and ARIA validity via axe + screen reader basis (role, state and broken relationships).

4.1.3 AA Status messages

◐ Partial

Checks: Behaviour: form without an aria-live region for status messages; whether the region covers all updates = manual

Manual review recommended before conclusion.