claude2figma-design-system-harness

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.80). The skill parses a Figma file URL from CLAUDE.md and programmatically loads that file's user-generated assets (e.g., via extractFileKey(CLAUDE_MD_CONTENT), loadDesignSystemAssets(fileKey), figma.getFileAsync/getLocalComponentsAsync) and then uses those components/tokens to drive design actions, so arbitrary third-party Figma content could supply instructions that materially influence behavior.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.80). The skill's settings.json runs "npx @anthropic-ai/figma-mcp" and the README lists the package URL (https://www.npmjs.com/package/@anthropic-ai/figma-mcp), which means remote package code from that npm package is fetched and executed at runtime and is a required dependency.