canvas-design

Pass

Audited by Gen Agent Trust Hub on Jun 21, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill employs a behavioral conditioning technique by claiming the user has already requested a 'masterpiece,' which is used to override default operational constraints and force a specific high-effort output style.
  • [EXTERNAL_DOWNLOADS]: The instructions command the agent to 'Download and use whatever fonts are needed,' which involves performing network operations to retrieve assets from unspecified external sources.
  • [COMMAND_EXECUTION]: The skill requires the agent to iteratively write and refine code to generate visual artifacts such as .pdf and .png files, representing a surface for dynamic code execution.
  • [PROMPT_INJECTION]: The skill processes user-provided conceptual references or 'subtle inputs' as the foundation for design creation. This ingestion point lacks explicit boundary markers or sanitization, creating a surface where embedded instructions in the user's input could influence the agent's file-writing or code-execution tasks.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 21, 2026, 06:50 PM
Security Audit — agent-trust-hub — canvas-design