The Agent Skills Directory

[DYNAMIC_EXECUTION]: The script uses pickle.load() in yt_playlist.py to read the authentication token from token.pickle. This is an unsafe deserialization pattern; if an attacker manages to modify the token.pickle file, they could achieve arbitrary code execution when the skill is loaded or run.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from the YouTube API, creating a vulnerability surface for indirect prompt injection.
Ingestion points: Functions list_playlists, list_playlist_videos, get_watch_history, and get_subscriptions in yt_playlist.py fetch titles and descriptions from YouTube.
Boundary markers: None. The fetched content is printed directly to the console/agent context without delimiters or warnings.
Capability inventory: The skill uses subprocess execution via python3 commands as defined in SKILL.md.
Sanitization: There is no evidence of filtering or escaping the text retrieved from YouTube APIs before it is presented to the agent.

youtube-playlists