The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes the yutu CLI to manage YouTube watermarks through the YouTube Data API.
[EXTERNAL_DOWNLOADS]: Installation instructions reference the @eat-pray-ai/yutu NPM package and the eat-pray-ai/yutu GitHub repository, which are official resources from the skill author.
[CREDENTIALS_UNSAFE]: The skill requires YouTube API access via OAuth credentials and tokens stored in local files (client_secret.json, youtube.token.json) or environment variables (YUTU_CREDENTIAL, YUTU_CACHE_TOKEN).
[PROMPT_INJECTION]: The skill incorporates user-supplied parameters, such as channel IDs and file paths, directly into CLI commands, which constitutes an indirect prompt injection surface. Ingestion points: Parameters like --channelId and --file in the operation reference files. Boundary markers: No explicit markers or delimiters are used in the command templates. Capability inventory: The skill executes subprocesses via the yutu binary. Sanitization: No input validation or sanitization logic is explicitly detailed in the documentation.

youtube-watermark