figma-pixel-implement

Pass

Audited by Gen Agent Trust Hub on Aug 27, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external UI design data from Figma URLs and nodes, which constitutes an indirect prompt injection surface where malicious design metadata could theoretically influence agent behavior.
  • Ingestion points: The skill ingests data through Figma design-context tools, metadata fetchers, and asset export APIs as described in SKILL.md and reference.md.
  • Boundary markers: No explicit string delimiters are defined for the design payload, but the skill includes instructions to treat fetched content as a reference for adaptation rather than raw code to be executed or pasted directly.
  • Capability inventory: The skill possesses the ability to read design context and assets via MCP and write code or assets to the local project filesystem.
  • Sanitization: The workflow mitigates risk by requiring the agent to map Figma properties to existing project tokens and variables, and by defining specific white/blacklists for asset types in reference.md.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 27, 2026, 12:15 PM
Security Audit — agent-trust-hub — figma-pixel-implement