The model identifies what should be redacted, hedged, or removed before publication. It
does not explain how an adversary would weaponize a finding. Output stays defensive: what
is wrong, how to fix it.
REMINDER: Output is an analytical aid, not a security clearance. The scan can miss findings or flag patterns that are intentional. Treat as a redline. Do not paste classified material; this is a public web app.