fashion-campaign-production
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external data such as brand briefs, campaign truth packets, and garment truth sheets. This ingestion of untrusted content represents an attack surface where malicious instructions could be embedded within product descriptions or campaign requirements to influence agent behavior.
- Ingestion points:
SKILL.mddefines an "Intake" phase that collects brand briefs and garment truth sheets from potentially external or user-provided sources. - Boundary markers: The skill suggests using "garment truth locks" and "negative constraints" within prompts to ensure accuracy, which serves as a prompt-based boundary to prevent hallucination or override.
- Capability inventory: The skill is intended to plan, prompt, and localize imagery and video, and it interacts with downstream AI generation tools as part of a production workflow.
- Sanitization: The workflow includes a "Fashion QA checklist" and a "claim ledger" to verify objective claims against references, though it does not provide technical sanitization for the input data strings.
Audit Metadata