Proof-Backed Findings: What a PandaONE Finding Looks Like
Every alert that ships includes safe exploit evidence — or it does not ship. How the Finalizer and proof-forward finding detail work after an AI pentest.
You open a scanner export with four hundred rows and no reproduction steps. PandaONE’s rule is the opposite: if we cannot attach safe exploit evidence, the finding does not ship. After an engagement, a Finalizer normalizes agent output into records the dashboard can show without raw tool noise.
What a finding contains
- Title and severity, grouped into customer categories — Injection, Authentication, Exposure, Configuration, Vulnerability, Infrastructure, Information (not internal tool names).
- Proof — request/response or other safe evidence that shows the issue is real.
- Impact and remediation — what broke, how to fix it, and status from open through retest to fixed.
Why proof comes first
Triage time scales with alert volume, not severity. A single curl replay beats a paragraph of risk scoring. Domain verification and MASVS mapping only matter if the underlying alert is reproducible.
Run Scan Now on a host you control, or follow the AI Security Engineer walkthrough for how each pipeline phase feeds the Finalizer.