canvas-design

Pass

Audited by Gen Agent Trust Hub on Sep 22, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill interprets user-provided aesthetic inputs to create visual art, creating a theoretical attack surface for indirect prompt injection.
  • Ingestion points: The skill body in SKILL.md instructs the agent to gather 'subtle input or instructions by the user' to serve as a creative foundation.
  • Boundary markers: There are no explicit delimiters or markers to separate user-supplied data from the skill's own instructions, though it does emphasize maintaining creative freedom over literal compliance.
  • Capability inventory: The skill is authorized to generate and write multiple file types (.md, .pdf, .png) and search local file directories (./canvas-fonts).
  • Sanitization: The skill does not provide instructions for sanitizing or validating the conceptual inputs provided by the user.
  • [EXTERNAL_DOWNLOADS]: The skill instructions allow for the acquisition of external design resources to complete its tasks.
  • Evidence: SKILL.md contains the directive to 'Download and use whatever fonts are needed to make this a reality.'
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 22, 2026, 03:50 AM