Route-level semantics
Catch duplicated landmarks, heading gaps, unnamed controls, and DOM relationships after composition.
Component tests matter, but routes can still combine into inaccessible pages. Scan deployed output, fix the source, and check it again after release.
Catch duplicated landmarks, heading gaps, unnamed controls, and DOM relationships after composition.
Review forms, dialogs, navigation, and loading states with automated evidence plus keyboard and screen-reader checks.
Schedule scans of saved routes and compare trends when design-system or feature changes ship.
Use targeted re-checks, CI, comparable-run diffs, and authenticated journeys to keep concrete evidence near engineering work.
Run the shipped GitHub Action or GitLab CLI with an explicit new-finding policy.
Re-scan the changed route and keep the rule, selector, source snippet, and affected-page evidence together.
Compare completed runs only when the page set and standard are comparable.
Model repeatable checkpoints for product states that a public crawl cannot reach.
Rendered-rule checks, configured CI policy, reached-page counts, comparable-run changes, and checkpoint findings.
Interaction quality, task completion, assistive-technology behavior, content meaning, and a conformance assessment.
The sample is a real report view: severity, affected element, rule, and a practical direction for the fix. Automated coverage is clearly separated from manual review.
