design-mobile-apps
Pass
Audited by Gen Agent Trust Hub on Nov 3, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to use shell commands like
curlto fetch API responses and write them directly to disk. This is used to handle large HTML/CSS documents from the Sleek API and to fetch SVG assets from Iconify. - [EXTERNAL_DOWNLOADS]: The skill connects to
https://sleek.designfor design services andhttps://api.iconify.designfor icon assets. These are well-known or service-specific endpoints necessary for the skill's primary function. - [INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection by ingesting and processing external HTML/CSS data.
- Ingestion points: External design data is fetched from the Sleek API via
GET /api/v1/projects/:id/components/:componentIdas described inimplementing.md. - Boundary markers: The instructions warn that HTML content is a mockup/placeholder and encourage visual verification against screenshots, but do not provide explicit structural delimiters for the agent to ignore instructions potentially embedded in the design data.
- Capability inventory: The agent is authorized to write new React Native source files, modify existing project navigation structures, and manage Node.js/Expo packages.
- Sanitization: No specific sanitization or filtering of the external HTML content is mentioned before the agent uses it to generate implementation code.
Audit Metadata