Close Campaign interface audit gaps

docs/ACCESSIBILITY_REVIEW.md

@@ -31,6 +31,31 @@ unlabelled icon-only buttons in the message preview, and Campaign-local modal
 implementations that bypass Core's `Dialog`. It complements rather than replaces
 browser and assistive-technology testing.
 
+The guard also verifies that Campaign retains narrow-viewport layouts, visible
+keyboard focus for domain-specific controls, and an explicit reduced-motion
+override. Shared dialog focus trapping and restoration are tested in Core;
+Campaign tests verify that overlays continue to use that shared primitive.
+
+## Release evidence
+
+The feature implementation can be closed once the structural contract, shared
+Core component tests, TypeScript graph, and representative responsive overlay
+tests pass. The seven-step matrix above remains a release-candidate checklist:
+it must be repeated for the exact browser, language packages, theme, density,
+and assistive-technology combination being certified. Closing the implementation
+ticket does not make a general WCAG-conformance claim for future releases.
+
+Implementation closure evidence recorded on 2026-08-03 used headless Chromium
+against the configured development system. It covered all 13 Campaign routes at
+1440 CSS pixels, the list, overview, recipients, and review routes at 320 CSS
+pixels, reduced-motion rendering, semantic accessibility-tree snapshots, HTTP
+500/page-error capture, and 42 sampled Tab stops across desktop and narrow
+Review & Send. The run found no unnamed interactive controls, hidden focus,
+page-level horizontal overflow, titlebar overlap, or unhandled runtime errors.
+Representative list, overview, review, and narrow-review screenshots were also
+inspected. This is implementation evidence; release certification still uses
+the complete matrix above with the selected screen reader and browser versions.
+
 ## Known boundary
 
 Translation keys may be visible in source because Core resolves them at runtime.