commercial-ad
Pass
Audited by Gen Agent Trust Hub on Sep 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from user-supplied briefs and external video analysis outputs which are used to generate content via batch processing tools.
- Ingestion points: External data enters the context via the initial brief grounding phase (Workflow Step 1) and the
$video-analysistool output in Review Mode. - Boundary markers: The instructions define 'Claim Boundaries' and 'Continuity Locks' to separate product facts from creative output, though there is no explicit instruction to sanitize or ignore instructions embedded within the user data.
- Capability inventory: The skill is empowered to invoke
$image-batch-runnerand$video-batch-runnerto create production assets. - Sanitization: The skill incorporates strong behavioral guardrails, such as prohibiting the invention of claims and the imitation of trademarked material, which mitigates the risk of malicious input driving the generation process.
Audit Metadata