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 reads outsider-authored free text because it ingests Figma frame/layer links and then calls get_design_context, get_metadata, and get_screenshot based on that selection, where the selection content originates from user/organization-authored Figma designs.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill declares and uses the streamable Figma MCP server URL https://mcp.figma.com/mcp (in the config and agent dependency) at runtime to fetch design context and generated code/snippets that directly drive the agent's outputs, so it is a required external runtime dependency that can control prompts/code.