product-ad-campaign
Pass
Audited by Gen Agent Trust Hub on Nov 7, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection as it ingests untrusted data from user-supplied product briefs, logos, and advertising copy (SKILL.md).
- Ingestion points: Product brief, logo references, and marketing copy provided by the user.
- Boundary markers: Explicit boundary markers for external data are absent in the prompt templates.
- Capability inventory: Media generation via
fal-mediaendpoints and manifest logging (SKILL.md, assets/plan.json). - Sanitization: The skill includes instructions to "Never invent claims, discounts or testimonials," which acts as a behavioral guardrail against hallucinated or injected content.
- [EXTERNAL_DOWNLOADS]: The skill describes a mechanism to retrieve external guides and version-pinned references using the
fal-mediaskill (SKILL.md). These are used for specialist guidance and model discovery within the author's own ecosystem. - [COMMAND_EXECUTION]: The skill provides an optional terminal-only workflow that references a local production-plan guide (
../genmedia/references/production-plans.md) and a local JSON template (assets/plan.json). No arbitrary or dangerous command execution was detected.
Audit Metadata