aesthetic-analysis
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the processing of untrusted data by instructing the agent to navigate to user-provided URLs and 'analyze deeply' to extract design elements. This creates a surface for indirect prompt injection where a malicious website could contain instructions intended to override the agent's primary task.
- Ingestion points: External websites accessed via Playwright MCP in the 'Capture Reference Design' step.
- Boundary markers: Absent. The instructions do not include delimiters or specific warnings to the agent to ignore executable instructions or natural language commands found within the design reference material.
- Capability inventory: The skill includes instructions to read existing project files and write new implementation code to the local filesystem (Step 3).
- Sanitization: Absent. There is no mention of filtering or sanitizing the data extracted from reference URLs before it is used to inform the implementation logic.
- [EXTERNAL_DOWNLOADS]: The skill utilizes the Playwright MCP server to interact with external web content. It references documentation and tools from established sources including official Expo and React Native repositories, as well as the Playwright MCP tool provided by ExecuteAutomation.
Audit Metadata