plan-secrets-audit
Pass
Audited by Gen Agent Trust Hub on Jul 24, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill functions as a template for a security engineering persona focused on credential auditing. All operations described are local to the user's environment.
- [DATA_EXFILTRATION]: While the skill searches for sensitive credentials (API keys, tokens, etc.), it contains explicit guardrails: 'Never print the secret. Type + location + last 4 chars at most.' There are no network operations or external data transfer commands identified.
- [PROMPT_INJECTION]: The skill identifies a surface for indirect prompt injection by scanning untrusted files and git history. However, it specifically restricts its output to a planning phase ('Plan only. No rotation, history rewriting, or env edits until each phase is approved') and concludes the turn after report generation, minimizing the risk of a multi-step attack.
Audit Metadata