figma-mcp-server

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly ingests arbitrary Figma file URLs (e.g., "Copy Figma frame URL → paste in prompt" and the get_design_context/get_screenshot tools that fetch from https://www.figma.com/design/FILE_ID?node-id=...), which are user-generated/public third‑party files that the agent must read and that directly drive tooling and code/write actions, so untrusted content could indirectly inject instructions.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill makes runtime calls to the MCP endpoint (https://mcp.figma.com/mcp) and to user-supplied Figma design URLs (e.g., https://www.figma.com/design/FILE_ID?node-id=123-456), fetching design data and images that are injected into the agent's model context and directly drive generated prompts/output.