visual-validation

Pass

Audited by Gen Agent Trust Hub on Jul 9, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to run a local script, scripts/pixel_compare.py, to perform pixel-level similarity testing on component screenshots. This command is scoped to the specific task of visual comparison using relative paths for input/output.
  • [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface because it evaluates untrusted external data (screenshots and design tokens).
  • Ingestion points: Component screenshots and JSON token records in ./components/ directories.
  • Boundary markers: None explicitly defined for the visual inspection phase, though evaluation criteria are structured in tables.
  • Capability inventory: Subprocess execution for pixel comparison and file system updates for token remediation.
  • Sanitization: No validation or sanitization is specified for the input data before processing.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 9, 2026, 01:18 AM
Security Audit — agent-trust-hub — visual-validation