The Agent Skills Directory

[COMMAND_EXECUTION]: The skill requires the agent to search and read font-related files from the local directory path ./canvas-fonts to incorporate specific typography into visual designs.
[EXTERNAL_DOWNLOADS]: Instructions in SKILL.md direct the agent to download and use external fonts if required for artistic expression. No specific or untrusted URLs are hardcoded; the capability is intended for functional design purposes.
[PROMPT_INJECTION]: The skill uses simulated conversation history (e.g., 'The user ALREADY said...') as a prompt engineering technique to force the model into a refinement state and enforce high craftsmanship standards.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by processing untrusted user input to derive the 'soul' or 'conceptual DNA' of the artwork. Evidence Chain:
Ingestion points: User input referenced in the 'DESIGN PHILOSOPHY CREATION' and 'DEDUCING THE SUBTLE REFERENCE' sections of SKILL.md.
Boundary markers: Absent; there are no clear delimiters or instructions to ignore embedded commands within the user-provided topic.
Capability inventory: Reading local files (./canvas-fonts), writing multiple file types (.md, .pdf, .png), and potential network operations for downloading fonts.
Sanitization: Absent; no validation or escaping of the user-provided reference is performed before it influences the model's creative and code-based output.

canvas-design