Help

Two kinds of question land here, and it's worth knowing which one you have before you start digging: "why did the engine do that" has an answer below, and "what will it never do" is answered just as directly — stated upfront, not left for you to discover the hard way in production.

FAQ

The questions that come up most, answered plainly: why a selection of tags and rule ids can run no rule and still look clean, why a web component comes back cantTell, what changes in your results after upgrading to core 1.15, why the title and language rules are notApplicable in a frame, why each WCAG criterion reads cantTell even when every check passed, why a geometry-dependent rule reports notApplicable under jsdom, and what a clean scan actually does and doesn't mean. Fifteen minutes here saves most of the debugging most people end up doing anyway. See FAQ.

Known limitations

The known boundaries of this engine, named explicitly rather than discovered by surprise: what one scan of one page state can't see, what depends on where you run it (jsdom, a browser, the page's own scripts), and what it deliberately declines to guess at because a wrong guess would be worse than an honest cantTell, each with what to do about it. Knowing these up front is what lets you trust the rest of the results. See Known limitations.