The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands using standard media utilities, including ffmpeg, ffprobe, and yt-dlp, to analyze video content, measure cuts, and extract audio features.
[EXTERNAL_DOWNLOADS]: Downloads video files from YouTube, Instagram, and TikTok using yt-dlp and a utility named treg. These downloads are central to the skill's purpose of analyzing existing video styles.
[EXTERNAL_DOWNLOADS]: Interacts with ElevenLabs APIs for advanced audio processing.
Evidence: Uses curl to send audio files to api.elevenlabs.io for voice isolation and to generate sound effects based on text prompts.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted media content and metadata from external sources, creating a surface for indirect prompt injection.
Ingestion points: Untrusted content is ingested through URLs provided by the user and processed via yt-dlp and ffmpeg.
Boundary markers: None are explicitly used in the instructions to separate data from instructions during the analysis phase.
Capability inventory: The skill possesses capabilities for network access, file system writes within its workspace, and command execution via the Bash tool.
Sanitization: The skill relies on the inherent parsing safety of mature tools like ffmpeg and yt-dlp to handle potentially malicious media files.

video-steal