figma

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

• Third-party content exposure detected (medium risk: 0.30). The required runtime workflow uses a selection-based Figma MCP flow where the user provides a Figma frame/layer link, the client extracts the node ID, and then the agent calls get_design_context/get_screenshot to ingest the resulting Figma-provided text/data for code generation (i.e., outsider-authored content can enter via the user-supplied Figma node selection rather than being limited to first-party vendor docs).

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

• Potentially malicious external URL detected (high risk: 0.80). The skill requires contacting the Figma MCP server (https://mcp.figma.com/mcp) at runtime to fetch design/context and generated React+Tailwind code that directly drives the agent's implementation, so the external content controls agent output.