figma-core-workflow-a

Pass

Audited by Gen Agent Trust Hub on Sep 9, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external Figma files (style names, typography properties, and descriptions) that could potentially contain malicious instructions or unexpected content designed to influence the agent.
  • Ingestion points: Figma API responses fetched in SKILL.md via client.getFile() and client.getFileNodes().
  • Boundary markers: None present in the code snippets to distinguish between instructions and ingested data.
  • Capability inventory: File system write access (writeFileSync), network access (curl), and shell execution (Bash(npm:*)).
  • Sanitization: Token names are sanitized for CSS variable usage using .toLowerCase().replace(/[\s/]+/g, '-') in SKILL.md.
  • [DATA_EXPOSURE]: The skill handles sensitive credentials (FIGMA_PAT) to authenticate API requests. It follows best practices by retrieving these from environment variables rather than hardcoding them. The requests are directed to the official api.figma.com endpoint.
  • [EXTERNAL_DOWNLOADS]: The skill utilizes curl to fetch data from api.figma.com. This is the official API endpoint for a well-known service, and the data fetched is used specifically for the skill's stated purpose of design token extraction.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 9, 2026, 03:45 AM
Security Audit — agent-trust-hub — figma-core-workflow-a