figma-implement-design

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

• Third-party content exposure detected (high risk: 0.75). SKILL.md Step 2/3/4 requires calling the Figma MCP server with a user-provided Figma URL/node-id, and the resulting design context/screenshot/asset payload (outsider-authored content from the user’s chosen Figma file) is ingested into the agent’s LLM context, creating an indirect prompt-injection surface.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill declares and requires the Figma MCP tool at runtime using the URL https://mcp.figma.com/mcp, and the agent fetches design context/screenshots/assets from that endpoint which directly control its implementation instructions and behavior.