ui-direction

Pass

Audited by Gen Agent Trust Hub on Oct 31, 2026

Risk Level: SAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill installs the well-known playwright library and uses @google/design.md from a trusted organization to validate design configuration files.
  • [COMMAND_EXECUTION]: The skill uses a simple Python one-liner to generate a random number for design variance and a test script that executes a sibling measurement script via subprocess.
  • [DYNAMIC_EXECUTION]: The measurement script uses Playwright's page.evaluate to run JavaScript logic within a browser context to extract UI metrics such as color contrast and font sizes.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes content from web pages and local design files; while this introduces a potential ingestion surface for untrusted data, the extracted data is structured (metrics and screenshots) and the risk is minimal.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 31, 2026, 11:36 PM