Threat-informed guardrails map to the frameworks your auditor cites.
VibeReview is a control engine for AI-assisted code. The threat model identifies the risks. Guardrails enforce policy at prompt time. PR reviews log evidence on every change, and mapping reports translate that activity into the vocabulary your assessor uses.
VibeReview reads the repo and identifies entry points, trust boundaries, data flows, and AI failure modes. VibeReview maps each finding to the controls in your framework that govern it. Auditors get scope without a workshop.
ENFORCE
Guardrails → policy at the prompt.
Each control becomes a guardrail. Rules load in the IDE before the agent writes a line. Policy enforcement happens upstream of the bug, not in the SOC 2 report two quarters later.
EVIDENCE
PR review → control evidence on every change.
Every PR runs the same guardrails. Every fire writes to the audit log. You don't reconstruct evidence for the assessor. You export the log.
REPORT
Mapping report → audit-ready artifact.
Per-framework reports cite controls by name and show which guardrails, which threat-model findings, and which PRs supplied the evidence. Hand it to the assessor on day one of fieldwork.
FRAMEWORK COVERAGE
Nine frameworks. One control engine.
WEB APPS
OWASP ASVS
ASVS V1 to V14 covered, with Level 1 to Level 3 mapping per control.
Threat model identifies in-scope controls per repo
Guardrails enforce ASVS verifications at prompt time
Coverage report cites the verification IDs by hand