The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on several shell-based tools including yt-dlp, ffmpeg, and python3 to perform its primary functions. It specifically references a hardcoded local script path at /Users/charliedeist/Desktop/New Root Docs/scripts/vtt_to_markdown.py to convert transcript files. This creates a dependency on an external script not provided within the skill package.
[INDIRECT_PROMPT_INJECTION]: The skill exhibits a vulnerability surface for indirect prompt injection as it is designed to ingest and analyze YouTube transcripts.
Ingestion points: Transcript files (.vtt and converted .md) are downloaded from YouTube and processed by the agent in Phase 2.
Boundary markers: The instructions do not define clear delimiters or warnings for the agent to ignore potentially malicious instructions embedded within the transcript text.
Capability inventory: The agent has the capability to execute shell commands (ffmpeg, yt-dlp, rm) and write files to the local system.
Sanitization: There is no evidence of sanitization or filtering of the transcript content before it is provided to the agent for narrative analysis.
[EXTERNAL_DOWNLOADS]: The workflow involves downloading video and subtitle assets from YouTube's infrastructure using the yt-dlp command-line utility. The skill also suggests installing dependencies via the Homebrew package manager.

youtube-clip-extractor