skills/jihnma/skills/fe-design-diff/Gen Agent Trust Hub

fe-design-diff

Pass

Audited by Gen Agent Trust Hub on May 17, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Fetches design assets from Figma's official API (api.figma.com) and its associated image CDN. These operations are necessary for the skill's primary function and target a well-known service.
  • [COMMAND_EXECUTION]: Invokes the bundled vrt.mjs Node.js script and manages Playwright browser instances to capture and compare screenshots of the local Storybook environment.
  • [REMOTE_CODE_EXECUTION]: The helper script vrt.mjs dynamically resolves and imports Node.js libraries (playwright, sharp, pixelmatch, pngjs) from the project's local node_modules directory. This is a common and expected pattern for developer utilities to ensure compatibility with the host project's environment.
  • [PROMPT_INJECTION]: The skill identifies and manages an indirect injection surface through the ingestion of external Figma data and local Storybook rendered content.
  • Ingestion points: Figma API image responses and rendered Storybook HTML/CSS content.
  • Boundary markers: Documentation in SKILL.md explicitly flags Figma content as untrusted and points to a shared security policy.
  • Capability inventory: Local script execution, file system cache access, and browser-based screenshotting.
  • Sanitization: The logic parses specific ID formats for Figma nodes and restricts Playwright navigation to localhost routes.
Audit Metadata
Risk Level
SAFE
Analyzed
May 17, 2026, 12:24 PM
Security Audit — agent-trust-hub — fe-design-diff