Map every surface displaying the same marker
Choose clearly fictional text and list its possible destinations: input, error, selected label, summary and confirmation. Check URLs or page titles if the implementation puts values there. A rule for one component does not automatically cover a different DOM surface.
A fictional enquiry route might repeat a reference in a next-step summary. Input masking can be correct while the summary lacks the necessary rule. Include both states in the test.
Create a new sequence after configuration changes
Record tool, configuration version and when the rule should take effect. Provider settings can take time to apply. An older recorded sequence does not necessarily show the new rule.
Start a new marked test under the approved setup and repeat the route. Clarity documentation states that masking changes do not apply retrospectively. Handle any earlier exposed evidence under the organisation's agreed process rather than treating it as automatically corrected.
| State | Risk | Check |
|---|---|---|
| Input | Basic rule missing | Marker not visible on protected surface |
| Field error | Value quoted as ordinary text | Feedback follows the same protection decision |
| Summary | Redisplay outside the rule | New surface verified |
| Dynamic component | New selector or nested rule | Current DOM covered |
| New sequence | Old configuration inspected | Version and actual application established |
Check exceptions and surrounding content
Test an allowed public instruction that should remain readable in the observation. Making everything unreadable can hide a broken setup as well as overexposure. Assess intended observation value and protection together.
Review nested masking and unmasking rules and languages with different components. A selector for an old structure may not cover the next version. Do not merely hide the marker in a reviewer's report if it was already transmitted incorrectly.
Keep evidence tied to component changes
Record surface, marker, version, new test sequence and what the reviewer could actually see. State unobservable routes too. Missing access remains a limitation rather than verified masking.
Repeat relevant cases when summaries, feedback, languages or observation settings change. A passed masking test covers its bounded route; it does not replace state, access or data-processing assessment.