The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/extract_beats.py executes external system binaries (ffmpeg, ffprobe) to process media files.
Evidence: The script uses subprocess.run in functions _probe_duration, _is_video, and _extract_audio to interact with audio and video streams.
Context: These calls are necessary for the skill's primary function of rhythmic analysis. The implementation correctly uses the list-based argument format to pass parameters, which is a security best practice that prevents standard shell injection vulnerabilities.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-provided audio and video files, which represents a potential surface for indirect injection if malicious metadata is processed by the underlying analysis libraries.
Ingestion points: The source parameter in scripts/extract_beats.py accepts file paths for processing media content.
Boundary markers: None. The skill analyzes the binary and structural content of the media files directly to generate a beat grid.
Capability inventory: The skill has file-write capabilities (writing beat-grid.json and manifest.json) and invokes external system commands (ffmpeg and ffprobe).
Sanitization: The script performs basic file existence validation and uses established libraries like librosa. It avoids shell interpolation in subprocess calls.

extract-beat-grid