UXRay WCAG Audit 1.0

Automated WCAG 2.2 testing, measured against the W3C ACT Rules

The accessibility module of UXRay checks a web page against WCAG 2.2 levels A and AA. We test it on the official ACT Rules test cases and publish the full results.

37–38
ACT rules consistent (all examples give the expected outcome)
15–17
more rules partially consistent
1134
official test cases, 91 rules

How it tests

  • Measurements on the rendered page: contrast, text spacing, zoom and reflow, orientation, focus visibility, target size.
  • Interaction with the page: keyboard navigation and focus traps, controls that reveal content, media playback and autoplay, time limits, device motion.
  • A language model for judgements that need context, such as whether a heading or page title describes its content.
  • An embedded axe-core 4.13 for many markup and ARIA rules.

Results that cannot be decided automatically are reported for manual review, not as failures.

Results on the W3C ACT Rules test cases

Three full runs on the official test cases (testcases.json). Consistency is counted as defined in the ACT Rules Format 1.1.

Measured on the same test casesConsistent rules
UXRay WCAG Audit 1.037–38
axe-core 4.13 alone26
Implementation report (EARL) JSON-LD · all 1134 test cases · release 6e30a93

Notes

Parts of the tool use a language model, so a small number of results can differ between runs. Across our three runs, 87 of the 91 rules gave the same verdict every time; the four that moved are all rules decided by the model, about headings, labels and link text. We do not tune the tool for individual test cases — where a rule could not be improved in a general way, it stays inconsistent.