What quality measurement covers
How Ballmac UI measures every item in a real browser, what ballmac check does and doesn't tell you, and where the limits are.
On this page
Every item in the catalog is measured before it ships, and the results are public on the quality page, including failures. ballmac check holds your own app to the same bar.
What it is, and is not
- Conservative. A rule reports only when it can see the problem in the code. If a value comes from a variable it can't follow, it stays quiet. A clean result means none of these specific mistakes was found, not that the app is accessible.
- Not a replacement for testing. Automated checks find a subset of problems. Use a keyboard and a screen reader on the real app as well.
- Honest about severity. Errors are things that break for someone, such as an image with no alt or a button with no name. Warnings are likely gaps. You decide what fails the build.
What each item is measured on
Every item page has a scorecard measured in a real browser: an axe-core accessibility scan in light and dark, no sideways scrolling at 360, 390 and 1280 px, reduced motion, a right-to-left mirror, a clean console, a keyboard pass, states, tests and size.
The measurement is honest about its limits: a loading spinner is kept turning on purpose, WCAG 2.2's 24 px target size is reported as a note rather than a failure, and some items aren't mirrored by design (code, charts, keyboard shortcuts), each with its reason shown.
Limits of ballmac check
- It reads TypeScript and JSX. Other file types and styles written in separate stylesheets are only partly covered (motion in CSS is).
- Components from other libraries are treated as able to render what they need. Only plain elements and Ballmac components are judged.
- Contrast is measured only by
--url(axe), because it depends on the final rendered colors.
Found a false positive? Tell us and include the line.