design-system-builder
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from user descriptions, screenshots, and Figma links to generate structured metadata and instructions for downstream AI coding agents. Ingestion points: Figma metadata and user-provided style descriptions in SKILL.md. Boundary markers: The generated design-tokens.json and design-system.md files lack explicit delimiters or instructions to ignore embedded commands. Capability inventory: The skill writes JSON, CSS, Markdown, and HTML files to the working directory. Sanitization: No specific filtering or sanitization logic for ingested design text is described.
- [EXTERNAL_DOWNLOADS]: The preview templates in the examples folder reference external assets from well-known services. Evidence: HTML files in the examples directory fetch typography from Google Fonts (fonts.googleapis.com).
Audit Metadata