canvas-design
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is a creative framework that guides the agent to act as a professional graphic designer, using aesthetic manifestos and quality-focused prompts to generate original visual art.
- [INDIRECT_PROMPT_INJECTION]: The skill identifies a potential attack surface by ingesting user input to deduce conceptual threads for design.
- Ingestion points: User instructions are used as a 'foundation' for the design philosophy (SKILL.md).
- Boundary markers: The skill explicitly instructs the agent that user input should 'not constrain creative freedom' and encourages 'artistic interpretation'.
- Capability inventory: The skill allows the writing of .md, .pdf, and .png files.
- Sanitization: None explicitly defined, but the limited output formats (static media) provide inherent mitigation against complex injection attacks.
- [DYNAMIC_EXECUTION]: The instructions guide the agent to 'refine/polish' its work, which typically involves generating and executing internal rendering code (e.g., SVG or image-generation scripts) to produce visual artifacts. This is the primary intended function of the skill and does not involve untrusted data sources.
- [EXTERNAL_DOWNLOADS]: The skill mentions downloading fonts but does not provide external URLs or automated commands to do so. It includes a directory of local font license agreements, suggesting the use of pre-approved or standard design assets.
Audit Metadata