Audit comparison
See what changed.
Compare two completed reports for the same target and settings. The comparison separates changed checks from missing observations. Try the fictional example or enter report links from your own audits.
Comparison method and limits
Compare like with like
A comparison requires the same starting address, final redirect destination, device, page limit, accessibility setting, pass threshold and result schema. Even then, a bounded crawler may reach different pages. A finding observed in only one run is labelled not observed in the other; it is never silently treated as repaired. Changed measurements with the same status remain distinct from improvements or regressions.
Share deliberately
Real audit report reads are public to anyone who knows the report identifier. Unlisted means absent from public history, not protected by a login. Sharing a report link or exported card gives others its recorded findings. Edit tokens are for managing a report and must stay separate. The sample contains no token; its public values are safe examples.
Repeat only with an owner
The API kit provides a manual CI workflow and a commented weekly option for a site you are authorised to audit. It does not enable a recurring job on this website. The owner chooses cadence, artifact retention and repository notifications. Server reports keep the existing service retention policy; a downloaded CI baseline is a separate copy with its own retention and access settings.
Verify the repair independently
A changed status is a reason to inspect the underlying finding, not a substitute for retesting the intended behaviour. Review the exact page and build, confirm that the rule is still applicable and keep the manual investigation separate. A changed header may improve an automated observation while a permission flaw remains outside this crawler’s scope.