figma-implement-design

Pass

Audited by Gen Agent Trust Hub on Mar 26, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • Indirect Prompt Injection Surface: The skill translates external design data into code, which introduces a potential surface for indirect instructions. (1) Ingestion Point: Figma design data is retrieved via get_design_context in SKILL.md. (2) Boundary Markers: The instructions do not specify explicit delimiters for untrusted content. (3) Capability Inventory: The agent can write and modify application code in the user's repository. (4) Sanitization: No explicit sanitization of Figma text content is described in the instructions. This is a common pattern for design-to-code tools and is handled by the agent's core safety measures.
  • External Resource Usage: The skill utilizes official Figma MCP servers and assets endpoints for design implementation. (1) Ingestion Point: agents/openai.yaml references https://mcp.figma.com/mcp. (2) Evidence: Step 4 of SKILL.md describes asset retrieval from the server's localhost endpoint. This is a standard and expected behavior for design integration tools provided by the vendor.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 26, 2026, 10:41 AM