layout-system
Pass
Audited by Gen Agent Trust Hub on Sep 10, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill directs the agent to visit and inspect multiple production websites, including Supabase, Linear, Stripe, and Grafana. These external references are used as industry-standard examples for extracting design patterns. All provided URLs target well-known and reputable technology services, and the interaction is limited to visual and structural inspection for design purposes.
- [INDIRECT_PROMPT_INJECTION]: The skill has a vulnerability surface for indirect prompt injection because it instructs the agent to browse and ingest data from external production sites. • Ingestion points: Hardcoded reference URLs pointing to external dashboards and marketing pages (e.g., supabase.com, grafana.org, plane.so). • Boundary markers: There are no explicit instructions or delimiters defined to isolate the content of these external pages from the agent's core instructions. • Capability inventory: The agent utilizes browser tools to observe bounding boxes, computed styles, and grid definitions. • Sanitization: No specific sanitization or filtering logic is provided for the structural data extracted from these sites, although the agent's focus is restricted to technical UI attributes.
Audit Metadata