The Agent Skills Directory

[COMMAND_EXECUTION]: Several scripts in the skill's directory automate the spritesheet creation process by invoking each other via subprocess.run. For example, finalize_pet_run.py executes extraction, composition, and validation steps. The skill also invokes ffmpeg in render_animation_videos.py to create previews. These calls use specific argument lists and target local or known system binaries.
[EXTERNAL_DOWNLOADS]: The fallback generation script generate_pet_images.py performs network requests to api.openai.com using curl. This is used to interact with OpenAI's image generation services when the primary generation tool is unavailable. This is a legitimate functional dependency on a well-known service.
[DATA_EXPOSURE]: The skill utilizes the OPENAI_API_KEY environment variable to authenticate requests to the OpenAI API. This key is transmitted to official OpenAI endpoints. The skill also manages project assets within the ${CODEX_HOME:-$HOME/.codex} workspace, which is consistent with its stated purpose.
[PROMPT_INJECTION]: The skill generates complex, multi-part prompts for image generation that include technical constraints and style guidelines. While it processes user-provided descriptions, it uses structured templates that help maintain boundaries. No evidence of malicious prompt injection or safety filter bypass attempts was found.

hatch-pet