figma-implement-design
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted external data from Figma design files via the Figma MCP server tools (
get_design_context). It lacks specific boundary markers or instructions to isolate or ignore malicious text commands embedded within design layers or node text. This data directly influences the agent's code generation capabilities as it updates local components, representing an indirect prompt injection surface.
Audit Metadata