The deliverable, section by section

Section	What it says
1. Verdict	STUB / PARTIAL / REAL / UNKNOWN, counts by severity, how many files were scanned, analysed, not analysed, skipped as tests
2. What was audited	The manifest digest the audit is bound to, and every excluded file with its reason
3. Method	The six groups of checks, and how severities are set
4. Scope and limitations	The engine's own scope sentence verbatim, what a REAL grade does not mean, and the measured accuracy, including the 3-of-12 blind result
5. Findings by blast radius	Files ranked by severity and closeness to real actions
6. Detail	Every finding with line, explanation and code; a note when more exist than fit; the list of files with no findings
7. Reviewer notes	A person's review, signed with the rest; or, in bold, that no person reviewed it
8. Remediation priorities	What to fix first; Fix + Prove per item
9. Evidence and verification	How to verify the bundle without trusting the key inside it

The signed record also says whether a person reviewed it

reviewed_by_person and reviewer are fields of the signed record, so a report without review cannot later be presented as reviewed. The command refuses a reviewer name without notes.

FIRME AI CORPORATION · CAIN-42 is a Byzantine trust fabric. Numbers marked "our own check" were run by us, not by an independent party. Nothing on these pages is a certification.