figma-code-connect

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

• Third-party content exposure detected (high risk: 0.70). This skill’s required workflow instructs the agent to “load Figma’s current figma-code-connect-components skill from the official MCP guide distribution” (SKILL.md:9), which implies runtime ingestion of externally authored documentation/skill text from an outsider source (Figma’s public GitHub guide/MCP distribution) into the agent’s LLM context.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill explicitly instructs at runtime to "load Figma's current figma-code-connect-components skill" from the official MCP guide (https://github.com/figma/mcp-server-guide), which means it fetches remote instructions that directly control agent behavior and is required by the procedure.