The Agent Skills Directory

[PROMPT_INJECTION]: The skill uses a state-forcing technique in the 'FINAL STEP' section of SKILL.md by instructing the agent to act as if the user has already provided specific feedback: 'The user ALREADY said "It isn't perfect enough. It must be pristine, a masterpiece if craftsmanship..."'. While intended for quality improvement, this technique overrides the agent's current state with simulated user input.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface where untrusted user input is ingested to define the art topic and philosophy without explicit sanitization or boundary markers.
Ingestion points: User input for art requests, posters, or design philosophies (found in SKILL.md).
Boundary markers: None identified in the instructions to separate user-provided concepts from the system's execution logic.
Capability inventory: The skill is authorized to create files (.md, .pdf, .png) and is instructed to 'refine/polish' code, implying access to code execution tools (found in SKILL.md).
Sanitization: No input validation or escaping mechanisms are described for the user-supplied conceptual foundations.
[DYNAMIC_EXECUTION]: The skill encourages the agent to generate and refine code at runtime to produce design artifacts. In SKILL.md, the 'FINAL STEP' instructs the agent to 'Go back to the code and refine/polish further', which involves the dynamic assembly and modification of executable content to generate the final PDF/PNG output.

canvas-design