landing-page-information-structure
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs the agent to retrieve media assets, videos, and design references from external domains including
goodcase.ai,motionsites.ai, and Amazon CloudFront. It also incorporates font and styling resources fromfonts.googleapis.comanddb.onlinewebfonts.com. These resources are used as templates for generating the requested user artifact. - [DYNAMIC_EXECUTION]: The skill generates implementation-ready UI code, including React components, Three.js 3D configurations, and Framer Motion interaction logic, at runtime based on the provided reference templates and user requirements.
- [INDIRECT_PROMPT_INJECTION]: The skill processes external design prompts and case study excerpts to guide the generation of page narratives and hierarchies. While this is an intended feature for template-based design, it represents an attack surface for indirect prompt injection from the external sources.
- Ingestion points: Case evidence files and external media URLs defined in
references/cases.md. - Boundary markers: The workflow utilizes a mandatory "reference contract" (Preserve/Replace/Avoid) to explicitly delimit how external traits are incorporated into the output.
- Capability inventory: Generation of HTML, CSS, and JavaScript artifacts; no shell execution or direct file system write capabilities are identified for the agent itself.
- Sanitization: The skill includes safety rules that restrict unauthorized publishing, purchases, or account changes without explicit user approval.
Audit Metadata