award-design
Pass
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill identifies and fetches configuration or documentation from well-known services and trusted organizations relevant to its primary purpose.
- References official documentation and GitHub repositories for well-known libraries such as GSAP, Three.js, and Lenis.
- Cites design research and guidelines from trusted organizations including Google Labs and Vercel Labs.
- [INDIRECT_PROMPT_INJECTION]: The skill establishes a data ingestion surface for external content but provides explicit architectural mitigations.
- Ingestion points: Uses a browser rung to fetch content from external URLs in
references/brand-extraction.mdto reverse-engineer design tokens. - Boundary markers: Includes a high-level instruction in
references/brand-extraction.mdto treat fetched content as untrusted data rather than instructions. - Capability inventory: The skill has filesystem access and executes local Python scripts to generate and audit design files.
- Sanitization: Instructs the agent to extract specific design observations (palette, typography, layout) while ignoring embedded copy or markup that could contain instructions.
Audit Metadata