Platform Security & Guardrails

Every feature that accepts input or routes contact is wrapped by a centralized security guardrail. This is visible, auditable, and non-negotiable.

Security Guardrail Monitor

Every feature action is audited for misuse & PHI exposure

No events yet. Submit a lead to see guardrails in action.

Guardrails active: rate limits • PHI term detection • input sanitization • heuristic + AI risk scoring (Google AI Studio path ready)

Rate limiting per session for lead submissions
Strict sanitization of all free-text fields
Real-time pattern detection for clinical/PHI language (diagnosis, QSI, medication lists, patient narratives, etc.)
Heuristic + optional Google AI Studio risk scoring (0–100)
Full immutable audit log (feature, user, risk, decision, reason)
Hard blocks surface clear messages and never store prohibited content

This demo log is in-memory and resets on refresh. Production writes to Neon with Clerk user correlation and alerting.