Protect Sensitive Data Before It Leaves the Boundary
The PII & Privacy Leak Gate is an adversarial review skill that scans text - prompts, logs, code, or AI outputs - for personal data, secrets, and confidential information. It acts as a final check before you paste sensitive data into a third-party AI or ship agent-generated content downstream.
How It Works
The skill runs a structured privacy review, analyzing your input against six categories of sensitive data, including direct identifiers, government and financial IDs, credentials, and confidential business info. It does not just block; it explains the risk and returns a safer version of your text.
- Risk Assessment: Categorizes findings into High, Medium, and Low severity.
- Adversarial Logic: Assumes destinations are untrusted, so it errs toward catching more rather than less.
- Automatic Redaction: Returns a clean rewrite of your text with clear placeholders like [REDACTED_API_KEY] or [REDACTED_EMAIL].
- Clear Verdict: Every review ends in a single PASS, FLAG, or BLOCK decision with the reasoning shown.
Why Pay for This?
Prompting an AI to "be careful" is not enough. This skill applies a structured adversarial framework that forces an explicit PASS/FLAG/BLOCK verdict, so findings are surfaced consistently instead of slipping through. It is a practical guardrail that helps your team catch accidental data exposure, support your privacy and compliance efforts, and avoid credential leaks before text leaves a trusted boundary.
Built for developers, support and ops teams, and anyone who pastes text into AI tools or ships agent output and wants a last review first.