organize
Pass
Audited by Gen Agent Trust Hub on Nov 2, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to write self-contained HTML files to the local working directory (e.g.,
sitemap-<topic>.html) and open them to provide visualizations for the user. - [EXTERNAL_DOWNLOADS]: The skill contains instructions to interact with external design platforms, specifically Figma and Pencil.dev, using the Model Context Protocol (MCP) to create or modify documents.
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data provided by users, such as content inventories and research data, which represents a potential injection surface. However, the skill is focused on structural design and documentation rather than executing logic from that data.
- Ingestion points: Untrusted user-supplied product information, content inventories, search logs, and research data.
- Boundary markers: Not explicitly defined in the instructions for user data interpolation.
- Capability inventory: File system writes (HTML) and external tool interactions (Figma, Pencil).
- Sanitization: None explicitly mentioned; the skill relies on the agent's generative process to structure the output.
Audit Metadata