design-references

Pass

Audited by Gen Agent Trust Hub on Sep 26, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructions direct the agent to download high-resolution images from dynamic URLs returned by Mobbin search tools. These files are stored in a local scratchpad folder to facilitate a visual catalog. This is a legitimate functional requirement for a design reference skill.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests metadata and images from an external database (Mobbin), which represents a potential surface for indirect instruction injection. However, the instructions prioritize visual analysis, and the risk is considered low for this use case.
  • Ingestion points: Search tool outputs for screens, flows, and sections in SKILL.md.
  • Boundary markers: The skill does not employ specific delimiters to isolate external search data from the agent's instructions.
  • Capability inventory: The agent can write files (image downloads) and access the network via the Mobbin MCP server.
  • Sanitization: No explicit sanitization or validation of the retrieved metadata is specified before it is presented to the user or processed by the agent.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 26, 2026, 05:47 AM