Security Retesting: How to Confirm a Vulnerability Is Fixed
The relevant question is whether the original attack path is actually closed.
Fixing code is not the finish line#
A patch can change one endpoint while leaving a second path vulnerable. An access-control fix can work for reads but fail for exports or background jobs.
Recreate the original conditions#
Identify the original actor, permissions, object, request and expected impact. Recheck the same trust boundary after the fix in a representative environment.
Probe nearby variants#
Within the agreed scope, consider alternate verbs, object identifiers, roles, endpoints and state transitions that exercise the same control.
Record what changed#
Distinguish verified resolution, partial remediation, still reproducible and cannot verify due to environment constraints. Preserve evidence without unnecessary sensitive data.
Prevent regressions#
Turn the root cause into automated authorization and security regression tests where appropriate. Independent retesting complements, rather than replaces, engineering tests.