The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes system commands using subprocess.run to perform audio conversion via ffmpeg and to execute the local parakeet-cli transcription tool. The implementation follows security best practices by passing arguments as a list and avoiding the use of shell=True, which prevents common command injection vulnerabilities.
[INDIRECT_PROMPT_INJECTION]: The skill transcribes external audio and video files. This introduces a surface where malicious instructions spoken in the audio could be transcribed into text and potentially influence downstream AI processes if the resulting transcript is not handled with caution.
Ingestion points: External media files provided to transcribe.py and pauses.py for processing.
Boundary markers: The skill outputs results into text and JSON files but does not wrap the transcription in specific security delimiters for downstream agent consumption.
Capability inventory: The skill can read media files, write transcript files (.txt, .json, .srt), and execute specific local binaries (ffmpeg, parakeet-cli).
Sanitization: There is no semantic sanitization of the transcribed text to filter for potential adversarial instructions hidden in the audio.

transcribe