The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill advocates for an iterative workflow where a generated image is approved and then used as a reference_image for subsequent generations to maintain consistency. This creates a data flow where the output of one AI task is used as input for the next, which is a common attack surface for indirect prompt injection.
Ingestion points: reference_image and pose descriptions (SKILL.md).
Boundary markers: The skill instructs to specify strict grid sizes (e.g., 32x32) and palette limits (e.g., 16-colors) to constrain output.
Capability inventory: Image generation, pack_sprite_sheet, and split_sprite_sheet utilities.
Sanitization: The skill does not explicitly describe input sanitization; it relies on model parameters like negative_prompt to filter unwanted attributes.
[COMMAND_EXECUTION]: The documentation includes a command npx skills add layerai/skills intended for environment setup. This targets the author's own repository to provide additional tools necessary for the pixel art workflow.

layer-pixel-art