sensitive-disclosure-guard

What problem does it solve?

This Skill helps prevent an AI feature from disclosing sensitive information such as PII, secrets, credentials, regulated data, or other-tenant data through model context, outputs, logs, or provider handling.

Core Features & Use Cases

• Data Minimization: Replaces whole-record context dumps with only the fields the task actually needs.
• Pre-Model Redaction: Designs stripping and masking pipelines for user input, retrieved content, and tool outputs before they reach the model.
• Leak Path Review: Checks for output echo, inference leaks, cross-user or cross-conversation bleed, and unsafe prompt logging.
• Provider Posture Review: Confirms retention, training-on-inputs, and residency constraints are acceptable for the data being used.
• Use Case: A support assistant that currently sends full customer records to an LLM can be redesigned to send only necessary fields, redact sensitive values, and verify that logs and responses do not expose private data.

Quick Start

Ask the skill to review your AI feature for sensitive-data exposure and propose a least-data, pre-model redaction design with leak checks for output, logs, and provider posture.