The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use a shell environment to execute the yt-dlp utility for downloading video content when provided with a URL.
[EXTERNAL_DOWNLOADS]: The skill performs network operations to fetch video files from user-supplied external links using standard command-line tools.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data in the form of video files and URLs, creating a potential surface for indirect instructions.
Ingestion points: Video files provided as conversation attachments or remote URLs downloaded via shell.
Boundary markers: The skill implements a mandatory 'Hard Gate' requiring technical analysis and reporting of the source video characteristics before any generation or transformation logic is permitted.
Capability inventory: Access to the host shell (yt-dlp), network access for downloads, and integration with the hailuo-h3-max video generation model.
Sanitization: The workflow relies on the agent's internal video-analysis capabilities to describe subject, style, and metadata, which serves as a validation layer before proceeding with user-requested variants.

video-fission