The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses ffmpeg and ffprobe to process local video files, extract frames, and isolate audio. These are standard tools for media processing and consistent with the skill's purpose.
[DATA_EXFILTRATION]: Audio data is sent to external Whisper backends, such as OpenAI or Groq, for transcription. This involves network operations to recognized services for functionality.
[PROMPT_INJECTION]: The skill processes potentially untrusted content from video files into a report, creating a potential surface for indirect prompt injection.
Ingestion points: Audio and visual content from the local video file provided as input.
Boundary markers: The observation.md output does not specify the use of delimiters or warnings to separate transcribed content.
Capability inventory: Uses ffmpeg, ffprobe, and external Whisper APIs.
Sanitization: No sanitization or filtering of transcribed text or visual observations is mentioned.

watch