The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/scrape_youtube_outliers.py utilizes subprocess.run to call the yt-dlp command-line utility. While the command is constructed as a list to prevent shell injection, the skill relies on the presence and behavior of this external executable to process search keywords.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes untrusted third-party content (YouTube transcripts) which is then summarized by an LLM.
Ingestion points: Transcripts are retrieved from YouTube videos using an Apify scraper in scripts/scrape_youtube_outliers.py and scripts/update_transcripts.py.
Boundary markers: The transcript text is interpolated directly into the prompt without delimiters or instructions to ignore embedded commands.
Capability inventory: The skill has the ability to make network requests to Apify, Anthropic, and Google APIs, and it can write data to Google Sheets.
Sanitization: The skill truncates the transcript length but does not perform any sanitization, filtering, or escaping of the content to prevent the LLM from following instructions embedded within the transcript.

youtube-outliers