The Agent Skills Directory

[COMMAND_EXECUTION]: The workflow in SKILL.md (Steps 1 and 2) instructs the agent to construct and execute shell commands using user-provided text. Specifically, user captions are interpolated into a prompt string which is then passed as a positional argument to python3 scripts/minimax_image.py "<prompt>". If a user provides a caption containing shell metacharacters (e.g., backticks, $(), or ;), it could lead to arbitrary command execution on the host system.
[PROMPT_INJECTION]: The skill processes untrusted user input in the form of custom captions. These captions are used to fill templates in assets/image-prompt-template.txt and assets/video-prompt-template.txt. This creates an indirect prompt injection surface where a user could attempt to influence the behavior of the image/video generation models or the agent's command execution logic.
[SAFE]: The Python scripts (minimax_image.py and minimax_video.py) correctly use the requests library for network operations and do not perform suspicious file system actions beyond saving the generated assets. The use of subprocess.run in scripts/convert_mp4_to_gif.py follows best practices by passing arguments as a list rather than a single shell string.

gif-sticker-maker