The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes external command-line tools to process audio and video files.
Evidence in SKILL.md: Uses subprocess.run to call whisper for transcription and ffmpeg for audio extraction.
Implementation detail: Commands are invoked using list-based arguments, which is a secure practice to prevent shell injection.
[PROMPT_INJECTION]: The skill is designed to process external, untrusted data (interview transcripts) which represents a surface for indirect prompt injection.
Ingestion points: The extract_quote function and transcription functions handle external text and media.
Boundary markers: None identified in the provided scripts.
Capability inventory: The skill can execute local shell commands (whisper, ffmpeg) and write files to the local file system.
Sanitization: No explicit sanitization or filtering of transcript content is implemented.
[SAFE]: No patterns of data exfiltration, credential harvesting, obfuscation, or unauthorized persistence were found. The skill's functionality is consistent with its stated purpose of interview management.

interview-transcription