Public-safe method kit and AI-OS for responsible bug hunting, evidence review, redaction, and report preparation.
Good security reports are boring in the best way: clear scope, clear evidence, clear impact, and no overclaiming.
A strong impact chain answers:
Evidence should show what happened, which account or role was used, which object was involved, what changed, and what did not happen in the secure control case. Use placeholders in public notes.
Do not claim account takeover, private data exposure, privilege escalation, or financial impact unless the evidence proves it. If the result only affects your own test data, say that clearly.
Discard or reclassify when the behavior is intended, out of scope, not reproducible, self-impact only, missing a boundary crossing, or based on private data you should not access.
For structured AI-assisted classification, see the AI-OS Reportability Rubric.