ad-creative
Pass
Audited by Gen Agent Trust Hub on Oct 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from customer reviews and comments to inform ad generation and strategy. Evidence:
- Ingestion points: Data is read from the 'inputs/reviews/' and 'inputs/comments/' directories (SKILL.md).
- Boundary markers: Prompt instructions do not specify delimiters or instructions to ignore embedded commands in the ingested text.
- Capability inventory: The agent can execute advertising platform CLI tools to create and manage campaigns (SKILL.md).
- Sanitization: Generated review pages implement HTML escaping and localized image source validation (references/creative-review-page.md).
Audit Metadata