The Agent Skills Directory

[COMMAND_EXECUTION]: Several scripts (isometric_pipeline.py, video_to_spritesheet.py, generate_walk_video.py, image_to_ascii.py) utilize subprocess.run to interface with system utilities. Specifically, the skill invokes ffmpeg for video processing, magick (ImageMagick) for image manipulation, jp2a for ASCII conversion, and rembg for background removal. All calls use argument lists rather than shell strings, which follows security best practices to prevent command injection.
[REMOTE_CODE_EXECUTION]: The isometric_pipeline.py script depends on an external skill, nano-banana-pro, and attempts to execute its edit_image.py script from a standard installation path (~/.claude/skills/nano-banana-pro/scripts/edit_image.py). This is a documented inter-skill dependency required for the pipeline's image-to-image generation steps.
[DATA_EXFILTRATION]: No network exfiltration or unauthorized data access patterns were detected. The skill references well-known services and repositories (GitHub, ayotomcs.me) for research and documentation purposes without transmitting sensitive data.
[PROMPT_INJECTION]: The skill uses hardcoded prompt templates for image generation in isometric_pipeline.py. It does not attempt to bypass agent safety filters or override system instructions.
[DYNAMIC_EXECUTION]: animation_builder.py generates JavaScript code for GSAP animations and can embed it into a standalone HTML file. This behavior is confined to generating front-end code for browser rendering based on predefined animation presets and does not execute dynamic code within the agent's environment.

sprite-forge