pixel-audit

Pass

Audited by Gen Agent Trust Hub on Jul 24, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill exposes an indirect prompt injection attack surface because it ingests design metadata from external design tools and uses browser automation tools to verify styles.\n
  • Ingestion points: Design metadata and variables are retrieved from Figma MCP nodes in Step 3 of the audit process and referenced in the evidence capture guidelines.\n
  • Boundary markers: The skill lacks explicit delimiters or specific instructions for the agent to disregard any embedded directives that might be present in the imported Figma metadata.\n
  • Capability inventory: The skill is authorized to evaluate JavaScript within a browser environment (getBoundingClientRect, getComputedStyle), perform file system operations to log defects, and use curl to verify asset pipelines.\n
  • Sanitization: There is no mention of sanitizing or validating the contents of the Figma node metadata before it is utilized in the verification and fixing stages.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 24, 2026, 05:28 AM
Security Audit — agent-trust-hub — pixel-audit