The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses subprocess.run to execute FFmpeg for audio extraction and video editing. These commands are implemented using argument lists rather than shell strings, which is a secure practice that prevents shell injection vulnerabilities.
[EXTERNAL_DOWNLOADS]: The skill transmits audio data to the OpenAI Whisper API for transcription. This is a core functionality of the skill and utilizes a well-known and reputable service provider.
[CREDENTIALS_UNSAFE]: The skill correctly instructs users to manage the OpenAI API key through an environment variable (OPENAI_API_KEY). This avoids hardcoding secrets and follows industry standards for secure credential management.
[PROMPT_INJECTION]: The skill transcribes audio from user-provided video files, which represents an ingestion point for untrusted data that could potentially contain indirect prompt instructions.
Ingestion points: transcribe_video.py generates transcript files in .json and .md formats from arbitrary audio content.
Boundary markers: No explicit markers or instructions are included in the output files to separate transcribed content from system instructions.
Capability inventory: The skill has permissions to read and write files and to execute system commands through FFmpeg.
Sanitization: The transcription output is stored directly without being filtered or sanitized for potential executable instructions.

video-cleaning