privilege-sentinel
Pass
Audited by Gen Agent Trust Hub on Jul 30, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is designed as a logic-based analyzer using provided knowledge bases (
citations.md,risk_taxonomy.md,surface_profiles.md). It does not contain any executable code (no Python, Node.js, or shell scripts). - [SAFE]: Instructions explicitly forbid telemetry or external data exfiltration ('Never offer to send the prompt anywhere... This skill runs locally and stays local').
- [SAFE]: The logic requires loading local markdown files to categorize risks based on established legal precedents (e.g., Heppner, Warner). This is a safe instructional pattern.
- [SAFE]: Metadata and frontmatter are consistent with the stated purpose. The skill uses standard markdown for user interaction and does not employ any obfuscation or dynamic execution techniques.
Audit Metadata