pets-and-animals-lead-by-1tmw5t
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill instructions and reference materials do not contain any malicious patterns such as command execution, persistence mechanisms, or credential harvesting. It includes explicit safety guidelines that prohibit unauthorized external changes, publishing, or impersonation of creators.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from
references/cases.md, which includes prompt excerpts and summaries from external sources (GoodCase). This creates a potential surface for indirect prompt injection if the source materials contained instructions meant to subvert the agent. - Ingestion points:
references/cases.mdcontaining case evidence cards. - Boundary markers: The skill requires a 'reference contract' (Preserve/Replace/Avoid) and instructs the agent to 'Distinguish observed evidence from your own recommendation', which helps isolate the external data from the agent's logic.
- Capability inventory: The skill is limited to text generation and does not utilize subprocess calls, file-write operations, or privileged network access.
- Sanitization: No explicit sanitization of the input text is performed, relying instead on the agent's internal safety filters and the specific formatting requirements of the workflow.
Audit Metadata