human-gate-designer

Pass

Audited by Gen Agent Trust Hub on Mar 9, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill content is restricted to markdown documentation and Mermaid diagrams describing workflow design.
  • [SAFE]: No scripts, binary files, or external software dependencies are included or referenced for installation.
  • [SAFE]: The analysis found no evidence of credential usage, network operations, or persistence mechanisms.
  • [PROMPT_INJECTION]: The skill describes a pattern for injecting human feedback into subsequent prompts, which is an indirect prompt injection surface.
  • Ingestion points: 'Human feedback' field in the 'Feedback Injection' section of SKILL.md.
  • Boundary markers: Uses simple quotation marks and brackets for template interpolation.
  • Capability inventory: None (design documentation only).
  • Sanitization: Not mentioned in the implementation guidelines.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 9, 2026, 06:13 PM