The Agent Skills Directory

COMMAND_EXECUTION (HIGH): The skill relies on a local Python script (scripts/video_processor.py) that executes shell commands (FFmpeg). There is a significant risk of shell injection if the script does not rigorously sanitize the file paths provided as CLI arguments.
INDIRECT_PROMPT_INJECTION (HIGH): The 'transcribe' functionality processes external audio content and returns it to the agent's context. This creates a surface for indirect prompt injection where malicious instructions embedded in the audio (e.g., spoken commands) could manipulate the agent's subsequent actions.
Ingestion points: video_processor.py reads user-provided video and audio files.
Boundary markers: None specified in the instructions to delimit transcript content from agent instructions.
Capability inventory: The skill can execute subprocesses (FFmpeg), write files to the local system, and generate transcripts that influence agent reasoning.
Sanitization: No evidence of sanitization for the content of transcripts or validation of input file metadata.
EXTERNAL_DOWNLOADS (MEDIUM): The skill requires manual installation of system-level binaries (ffmpeg) and Python packages (openai-whisper) from external sources that are not within the explicitly defined trust scope for this analysis.

Video Processor