product-ad-shot-design-by-1cw2bs8
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to process content from
references/cases.mdand external URLs (e.g.,goodcase.ai,x.com) to extract design traits and prompts. - Ingestion points: The agent reads the local
references/cases.mdfile and is directed to use browser or media tools to inspect content from external URLs provided in that file. - Boundary markers: The skill provides structural instructions (Preserve/Replace/Avoid) to scope the agent's output, but does not implement cryptographic verification or strict sanitization of the external reference content.
- Capability inventory: The agent uses inspection tools (image/video/browser) and text processing to generate video prompts and storyboards.
- Sanitization: No explicit sanitization or filtering of the remote content is performed before interpolation into the workflow.
- [EXTERNAL_DOWNLOADS]: The skill contains several URLs for reference media (MP4 videos, JPG posters) and case summaries.
- Evidence: Links to
goodcase.ai,media.goodcase.ai,x.com, andmp.weixin.qq.comare provided for visual reference. - Context: These are used for analytical inspection by the agent's tools to verify visual fidelity and do not involve the execution of remote scripts, package installations, or shell commands.
Audit Metadata