advanced-frontend-uiux
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied parameters, such as website type, target audience, and visual style, to generate complete frontend and backend code. This ingestion of untrusted data represents a potential attack surface for indirect prompt injection where a user might embed malicious instructions in their design request.
- Ingestion points: User requests analyzed during the "Input Processing" phase (SKILL.md).
- Boundary markers: None. The skill does not provide explicit delimiting or instructions to ignore commands embedded within the user data.
- Capability inventory: The agent is instructed to generate modular code across multiple files (HTML, CSS, JS, PHP) and assets.
- Sanitization: The skill proactively instructs the agent to generate code that sanitizes form inputs and uses HTTPS (Section 8), which mitigates some downstream risks of the generated output, though it does not sanitize the input prompts themselves.
- [EXTERNAL_DOWNLOADS]: The skill references several well-known and industry-standard third-party libraries for 3D rendering and animations, such as Three.js, GSAP, Babylon.js, and Particles.js. These resources are intended to be loaded via CDN links in the generated frontend code, which is standard practice for web development.
Audit Metadata