creativeclaw-ad-variations
Pass
Audited by Gen Agent Trust Hub on Oct 24, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user-supplied content, including product photos, video advertisements, and transcripts, to generate prompts for downstream image and video models.
- Ingestion points: Untrusted data enters the agent context through the analysis of "Product shot" images and "Existing ad" files as defined in the source identification section of SKILL.md.
- Boundary markers: The instructions do not mandate the use of delimiters or explicit "ignore embedded instructions" warnings when the agent interpolates content extracted from user media into prompts for generation tools.
- Capability inventory: The agent possesses access to numerous tools that can consume credits or modify assets, such as generate_video, generate_image, merge_media, and add_audio_to_video.
- Sanitization: There are no explicit requirements for sanitizing, escaping, or validating the text extracted from user-provided media before it is used to steer the output of generative AI models.
Audit Metadata