visual-style
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted data from external sources to generate structured design systems and derivative content.
- Ingestion points: External data enters the agent's context through URL fetching and screenshot analysis (
references/extractors/from-website.md), PDF parsing (references/extractors/from-pdf.md), and video keyframe analysis (references/extractors/from-video.md). - Boundary markers: The instructions lack explicit boundary markers or safety delimiters to isolate content extracted from external sources from the agent's operational instructions.
- Capability inventory: The skill generates complex outputs including standalone HTML files, generation prompts for HeyGen Video Agent, and TypeScript interfaces for Figma plugins. It utilizes capabilities for web fetching and vision-based analysis.
- Sanitization: There is no evidence of sanitization, filtering, or escaping for the external content before it is interpolated into prompts or generated code.
Audit Metadata