Loading
You already build accessible components without being asked, and that is the part this course skips. Establishing the standard across a portfolio is a different job: it means naming the criterion by number, knowing which of the WCAG 2.1 AA criteria actually fail in production code, being able to say how much of that a scanner can see, and ending a disagreement with a spec reference instead of a strongly held view about contrast. That is a citation skill, not a craft skill, and nobody acquires it by continuing to do good work quietly.
Start lesson one: What conformance actually claims →
The whole twelve-lesson spine is registered so the numbering never shifts, but only the first module is written. Work through it, fill in the conformance note, and ask your teaching agent for the next module.
What WCAG 2.1 AA actually obliges, which criteria fail in real components, and how much of it a scanner can see.
The four patterns you keep rebuilding, and enough screen reader to catch your own mistakes.
Where the published guidance runs out: streaming output, long runs, interrupts, and traces.
The first two modules are catch-up: precise versions of things you mostly already do. The third is the part almost nobody has written down. A live region pointed at a token stream re-announces itself into uselessness; a keyboard user has no obvious way to cancel a run that has been going for four minutes or to jump to the review gate when it finally appears; and the W3C’s own working draft on the accessibility of generative AI, as of its March 2026 editor’s draft, does not address any of it. That gap is the reason this course sits next to the streaming and observability courses rather than standing on its own.
Every criterion number on these pages links to the success criterion or its Understanding document. If a number or a level looks wrong, check the resource list and tell your teaching agent — a course about citing the spec correctly has no excuse for citing it wrongly.