The Agent Skills Directory

[SAFE]: The skill executes transcription tasks via subprocess.run using list-based arguments, which is a secure method that prevents shell injection vulnerabilities.
[SAFE]: External resources, specifically machine learning models, are downloaded from Hugging Face's official repositories. As a well-known service, these downloads are considered standard functionality for the skill's purpose.
[SAFE]: File operations are handled appropriately, using standard temporary directories (/private/tmp/) for staging audio and storing results, without evidence of unauthorized data access or exfiltration.
[SAFE]: The use of the uvx package runner to execute mlx-whisper is a modern and legitimate practice for managing tool dependencies in a local environment.

audio-transcription