plan-data-integrity
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is explicitly designed for defensive security auditing. Its primary purpose is to map paths to data loss and recommend system boundaries (such as RBAC and approval gates) rather than relying on soft prompt-based instructions.
- [SAFE]: The skill includes strict 'plan-only' guardrails that prevent the agent from executing migrations, editing tokens, or running destructive commands, even for verification purposes. It mandates that no changes be made until manual approval is obtained.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data such as database migrations and source code. However, the risk is mitigated by the 'plan-only' restriction, the explicit instruction to use system boundaries instead of prompt instructions, and the requirement that the agent only produce a markdown report as its deliverable.
- [CREDENTIALS_UNSAFE]: While the skill mentions tokens and credentials, it does so only to advise on their security (e.g., identifying overprivileged tokens). No hardcoded credentials or unsafe secret handling practices were found.
Audit Metadata