Step 1: preserve the original finding
Capture the original scanner output, location, version, and context. Without the original evidence, later verification cannot compare outcomes. The finding is the baseline.
Step 2: identify the fix artifact
Record the commit, build, or deployment that contains the claimed fix. The artifact must be identifiable and reproducible. In multi-repository vulnerability remediation, the fix may span several repos and services.
Step 3: re-evaluate with independent evidence
Run the original test or scanner against the new artifact. Where possible, add a second independent source such as runtime telemetry or reviewer attestation. Two findings from the same scanner do not double the confidence.
Step 4: record one of three outcomes
The only honest outcomes are: risk absent, risk present, or indeterminate. Indeterminate must be escalated, not promoted to fixed.
Step 5: bind to policy and release
The verified evidence is evaluated against the policy version active at release time. The decision record links the original finding, the fix artifact, the verification evidence, and the release outcome.