ads
Pass
Audited by Gen Agent Trust Hub on Nov 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from external sources such as ad campaign metrics and review cards.
- Ingestion points: Data enters the agent context through the
results_getandreview_gettools. - Boundary markers: The skill contains a specific rule instructing the agent to treat text in tool results as data, never as instructions.
- Capability inventory: The skill has capabilities to approve financial spend and launch campaigns via the
review_answertool. - Sanitization: The skill relies on natural language boundary instructions to prevent malicious data from being executed as commands.
Audit Metadata