ADR 015 — Accessibility (WCAG 2.2 AA)

Status

Accepted — Track B of the Migration Plan. The standard applies to the rewritten Nuxt app (ADR 011); it is built in during the rewrite, not retrofitted.

Context

The brief calls for a large accessibility overhaul — aria-label, aria-hidden, keyboard navigation, and more. The rewrite is the moment to build accessibility in rather than bolt it on. Two facts shape the work: Ant Design Vue is not a headless / WAI-ARIA-first library (ADR 013), so accessibility needs deliberate manual hardening; and the app is EU/French-facing while the European Accessibility Act is in force, so conformance may become a legal obligation.

Decision

Target WCAG 2.2 Level AA across the application, built into the rewrite, and structured so a formal EN 301 549 / EAA conformance package can be added later (accessibility statement, independent audit, remediation log) without rework — but without running that formal process now.

Practice areas

• Semantics & ARIA. Native elements first; aria-label / aria-labelledby for icon-only controls; aria-hidden on decorative pictograms/illustrations (ADR 013 assets); landmark regions; aria-live for async notifications and validation summaries.
• Keyboard navigation. Every interactive element reachable and operable by keyboard, logical focus order, visible focus indicators never removed, and no keyboard traps.
• Focus management. Move focus into dialogs/drawers/menus on open and restore it on close; trap focus within modals (Ant Design Vue overlays are the main risk area); a skip-to-content link; and deliberate focus handling on route change (SPA/SSR navigation — ADR 012).
• Forms. Programmatic label association, aria-describedby for hints/errors, aria-invalid, an error summary that takes focus, and validation announced via a live region — mapping the backend's cross-field validators / ErrorDto to accessible messages.
• Colour & contrast. Meet AA contrast in both light and dark themes — leveraging the per-mode primary override (ADR 013) — and never signal by colour alone (pair with icon/text).
• Images & media. Meaningful alt where informative; empty alt / aria-hidden when decorative (ADR 013 assets).
• Motion. Honour prefers-reduced-motion.
• Language. Set and switch the lang attribute with i18n (en/fr); localized accessible names.
• WCAG 2.2 additions specifically. Cover the criteria new in 2.2 — Target Size (Minimum), Focus Not Obscured, Dragging Movements, Consistent Help, Redundant Entry, and Accessible Authentication (the last pairs with the OIDC/BFF login — ADR 022).

Ant Design Vue hardening

Audit each Ant Design Vue component in use for ARIA correctness and keyboard behaviour, wrap or patch where the defaults fall short, and keep a per-component accessibility checklist so the gaps are tracked rather than assumed absent.

Testing (with ADR 021)

• Automated: @axe-core assertions in component and e2e tests, run in CI as a gate.
• Manual: keyboard-only passes and screen-reader testing (NVDA / VoiceOver) on the critical journeys — the part automation cannot cover.
• A conformance checklist maps features to WCAG 2.2 AA criteria, so an accessibility statement and external audit can be produced on demand (the formalize-later hook).

Consequences

Positive

• Built in, not retrofitted. Doing this inside the rewrite window is far cheaper than a later remediation pass, and it directly delivers the brief's aria-* / keyboard goals.
• Enforceable in CI. axe gates keep regressions out; the per-mode contrast work reuses the theming decision (ADR 013).
• Ready to formalize. The conformance checklist means EAA/EN 301 549 compliance can be produced without re-doing the work.

Negative

• Real manual effort. Because Ant Design Vue is not ARIA-first, per-component hardening plus manual screen-reader/keyboard audits are unavoidable and cannot be fully automated.
• Ongoing discipline. Every new component must pass the checklist, or conformance erodes; the WCAG 2.2 additions add specific, easily-missed criteria.

Why not best-effort AA only

Fixing what axe finds without the checklist and structure would leave any future EAA formalization as a from-scratch effort. Structuring for it now is cheap insurance given the EU exposure.

Why not full EAA formalization now

The published statement, independent audit, and periodic re-audits are real overhead that is not warranted until a legal trigger applies. Building to AA and maintaining the checklist captures most of the value at a fraction of the process cost, and leaves the formal package a short step away.