Skip link, landmarks, hierarchical headings, and a focusable main area.
Accessibility is a verification register, not a footer intention.
This self-assessment states what is implemented in the showcase, what needs retesting, and what still has no evidence. It is neither a VPAT nor a WCAG certification.
Last self-assessment
Scope: public site and local demos. Target: WCAG 2.2 AA. Third-party audit: none.
Observable state and expected evidence.
‘Implemented’ describes the current code; it does not mean exhaustive conformance.
Navigation, menus, forms, and principal demos retain visible focus.
Principal mobile actions are at least 44 × 44 px.
Animations are removed when reduced motion is preferred.
The palette targets AA; every change still needs a complete manual pass.
No complete VoiceOver, NVDA, or JAWS matrix has been published.
No independent audit or conformance document is claimed.
The review protocol.
Every significant change must revisit the same scenarios, not merely pass an automated audit.
- 01Complete keyboard navigation, focus order, and menu exit.
- 02200% zoom, narrow reflow, and enlarged text without lost actions.
- 03Contrast for text, states, focus, and interactive surfaces.
- 04Names, roles, errors, and announcements in forms and demos.
- 05Reduced motion and comprehension without animation.
- 06VoiceOver and NVDA on critical paths before publication.
Known limits of this register.
Simulated Atlaskit components evolve with their dependency. Internal prototypes are outside the public promise. No accessibility test with disabled users is documented yet.
A report must stay actionable.
- 01URL and affected step.
- 02Assistive technology and version.
- 03Expected and observed results.
- 04Task impact and possible workaround.
- 05No personal data or secret in the screenshot.
Publishing the gap begins the work; it does not finish it.
Use support to prepare a reproducible report. A private channel and response target will be added before commercial publication.