The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted third-party textual data (such as video titles, descriptions, comments, and tags) retrieved dynamically via the YouTube Data API during metadata processing and search workflows.
Ingestion points: SKILL.md (Channel Analysis, Video Performance, Search, and YouTube SEO Analysis steps fetching metadata via curl).
Boundary markers: Absent. The skill does not instruct the model to treat external content or API data boundaries as untrusted text strings.
Capability inventory: Employs standard curl commands in SKILL.md for querying standard HTTP services.
Sanitization: Absent. No specific filtering or escaping mechanisms are mandated for processing text returned from YouTube endpoints.
[COMMAND_EXECUTION]: Executable shell syntax is showcased using backtick-wrapped curl examples designed to invoke raw command line tools dynamically with explicit shell concatenation logic (key=${YOUTUBE_API_KEY}).

youtube-analytics