What problem does it solve? Sensitive values such as credentials, user content, identifiers, and recovery material often leak into logs, analytics events, crash reports, traces, and support exports, violating privacy claims like E2EE or zero-knowledge. This Skill provides a structured audit workflow to detect and document those leaks with evidence. ## Core Features & Use Cases - Sink Inventory and Field Mapping: Enumerate SDKs (Sentry, Crashlytics, Datadog, Amplitude, Segment, and more), logging wrappers, crash hooks, and support tools, then map every emitted field to its sink, readers, retention, and redaction point. - Sentinel-Based Runtime Testing: Generate synthetic sentinel values per data class, exercise normal, error, support, and consent flows, and search collected artifacts for exact, encoded, hashed, or transformed sentinel forms. - Structured Findings and Reports: Emit findings with TCLS-prefixed IDs, severity, redacted evidence, and retest steps using the provided JSON schema and Markdown report template. - Use Case: Before shipping a feature that claims "we never log message content", run this audit to trace message bodies through loggers, crash breadcrumbs, and analytics properties, and produce a report proving or disproving the claim. ## Quick Start Ask the agent to audit the current repository's logging, crash reporting, analytics, and support export paths for sensitive data leakage and report findings with file locations and redacted evidence.
