Accessibility

If you work in education or government, this is usually the first question you have to answer, and most vendors do not answer it. Here is everything we actually know, including the parts that do not look good.

Read this first. What follows is automated checks only. Machine testing catches roughly a third of real accessibility problems. It cannot tell you whether alt text is meaningful, whether a screen reader announces something sensibly, or whether an interaction can really be completed without a mouse. We have not had a human accessibility audit, we do not have a VPAT, and we have not tested with screen reader users. Do not treat this page as a conformance claim, because it is not one.

The measurement changed in v0.23.0. The script that produces these numbers was rewritten, and the set of checks it runs is not the same one behind any figure published before then. Several old results turned out to be wrong in both directions: it read colours through a stylesheet without accounting for transparency, so it invented contrast failures on three courses that never had one; it rejected valid labelling; and it counted headings in pages that build their headings after load, marking four packages as missing a main heading when each of them has one. Scores from before v0.23.0 are not comparable with these. A number that moved is at least as likely to mean we fixed the ruler as that we fixed the product — where a package genuinely improved, its row below says what it now passes, and the fixes are in the changelog rather than implied by the total.

What we check

Static analysis of the HTML and CSS each package ships, run against 18 of the 18 packages that have a live demo. Nothing here is measured in a running browser, which is a real limit: it is why the list below is short and why the caveats under it matter.

The page declares a language, and has a title
There is exactly one main heading, and heading levels do not skip
Images have alt text, or are explicitly marked decorative
Form fields have labels
Buttons and links have something a screen reader can announce
A visible focus style is defined, so keyboard users can see where they are
Sampled visible text meets the WCAG AA contrast ratio of 4.5:1

Why the totals differ between packages. A check that cannot apply is left out of a package’s total rather than counted as a pass — a page with no form fields is not credited for labelling them. A check is also left out when the answer cannot be read from the files: several packages build their whole interface in JavaScript, and for those we can see the code but not the page it produces. That is why one package is scored out of five and another out of eight. A missing check is not a passed check.

Two limits worth naming. Contrast is only sampled where a rule sets a text colour and a background together, and it is not sampled at all on translucent panels, because what sits behind them is not knowable from a stylesheet. And the 4.5:1 threshold we apply is the one for normal text; we do not model the looser 3:1 allowance for large text, so a heading can be reported as failing when it would in fact pass.

What fails most often

Across all 18 packages we checked. This is our fix list, in priority order.

Exactly one main heading	1 of 18 packages

The most common remaining failure is exactly one main heading, on 1 of 18 packages. Keyboard focus styling used to head this list on most of the catalog; it was fixed in v0.23.0 and every package now defines a visible focus ring.

Every package, worst first

Nothing is hidden or sorted to flatter. If a package you need is low on this list, ask us before you buy.

Package	Automated checks	What failed
Inbox Rush	7/8	Exactly one main heading
Auto Course Trailer	5/5	nothing
Block Course (Rise-style)	5/5	nothing
Brewing Methods	5/5	nothing
Crossword	7/7	nothing
Decision Tree	5/5	nothing
Espresso Foundations	6/6	nothing
Illustrated Slide Course	3/3	nothing
Microlearning module	6/6	nothing
Paginated Reader	6/6	nothing
Scrollytelling lesson	4/4	nothing
Self-Talk reflection	3/3	nothing
Semantic-Zoom Canvas	3/3	nothing
Slide Player	5/5	nothing
The Institute Framework (SL2)	7/7	nothing
The Science Lab	5/5	nothing
Transitions library	7/7	nothing
Widgets library	5/5	nothing

If you need more than this

A buyer with a legal obligation needs more than automated checks, and we would rather say so than sell you something that fails a review. Write to hello@actyra.com and tell us which packages matter to you. We will tell you honestly what has and has not been tested, and what it would take to get a package to the standard you need.