The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides Python scripts (e.g., video_convert.py, image_scale.py) that interface with FFmpeg and ImageMagick. Analysis of the scripts shows they correctly use subprocess.run() with argument lists and the default shell=False setting. This implementation ensures that user-provided file paths or parameters cannot be used to perform command injection attacks.
[EXTERNAL_DOWNLOADS]: The skill's configuration requires the installation of ffmpeg and imagemagick using apt-get. These are highly reputable, well-known multimedia processing tools maintained by trusted open-source communities and delivered via official package repositories.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied media files as input for its conversion scripts. Because the skill treats these files as binary data for processing by specialized external utilities and employs safe execution wrappers, there is no viable path for malicious instructions within the media files to influence the agent's behavior.
Ingestion points: The input_path parameter in all processing scripts (scripts/*.py).
Boundary markers: None (standard for binary media processing).
Capability inventory: Subprocess execution of ffmpeg and magick for media manipulation.
Sanitization: Input validation is performed by the underlying system utilities, and shell injection is prevented by the script's architecture.

media-processor