canvas-design

Pass

Audited by Gen Agent Trust Hub on Apr 21, 2026

Risk Level: SAFE
Full Analysis
  • Refinement Directives: The skill employs specialized instructional techniques to ensure the highest output quality. This includes a section that simulates user feedback requesting a 'masterpiece,' which is used to steer the agent towards a second pass of refinement and polish. This is a common performance-enhancing pattern for creative AI tasks.
  • Asset Management: The instructions guide the agent to use a local directory (./canvas-fonts) for typography. While it mentions downloading additional fonts if needed, this is part of its core functionality as a design tool and typically relies on the environment's standard font-handling capabilities.
  • Subtle Conceptual Mapping: The skill instructs the agent to deduce 'subtle conceptual threads' from user requests. This creates an indirect data processing surface, but given the skill's specialized focus on visual PDF/PNG output, the risks associated with this pattern are naturally constrained.
  • Instructional Intensity: The use of strong markers like 'CRITICAL' and 'NON-NEGOTIABLE' is focused entirely on maintaining artistic standards and craftsmanship rather than attempting to bypass safety protocols or modify core agent behaviors.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 21, 2026, 11:51 AM