The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes multiple local JavaScript and shell commands to manage the video production pipeline.
Orchestrates local scripts such as scripts/video-explainer.mjs and scripts/verify-shorts.mjs for diagnostics, review, rendering, and verification.
Utilizes system tools like ffmpeg and ffprobe for media processing.- [EXTERNAL_DOWNLOADS]: The installation process involves downloading external resources from the vendor repository.
Fetches the claude-video-kit from the author's GitHub repository using npx skills add and installs dependencies via npm ci.- [PROMPT_INJECTION]: The skill has an indirect prompt injection surface as it processes external research briefs into executable video scripts.
Ingestion point: The script.json generation process in brief-to-script.md accepts research briefs or scripts from the agent context.
Boundary markers: The skill implements a mandatory 'review gate' workflow defined in review-gate.md, which requires an independent review and a SHA-256 bound receipt before rendering is permitted.
Capability inventory: The skill calls subprocesses for video rendering and verification via local scripts.
Sanitization: The manual/independent review receipt serves as the primary validation step for untrusted content.

video-explainer