educational-ad
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill accepts and analyzes user-supplied media files and ad concepts, which creates an attack surface where instructions hidden within that media could attempt to subvert the agent's logic.
- Ingestion points: External media and scripts are ingested and analyzed during the 'Review Mode' workflow (SKILL.md).
- Boundary markers: The instructions provide a structured 'Review Rubric' (references/review-rubric.md) that mandates evidence-first reporting, which helps anchor the agent to objective observations rather than following potentially malicious embedded instructions.
- Capability inventory: The skill has the ability to route data and commands to production tools including '$image-batch-runner' and '$video-batch-runner' (references/workflow-handoff.md).
- Sanitization: The use of structured audit scales and required evidence-matching for every claim provides a functional validation layer that limits the impact of untrusted input.
Audit Metadata