The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill defines workflows for analyzing external video content and social media posts, creating an ingestion surface for untrusted data. * Ingestion points: External video transcripts and sampled frames (via watch-video) and social media post metadata (via social-fetch) in references/edit-anatomy.md. * Boundary markers: The instructions do not specify explicit delimiters or "ignore" instructions for the processed transcript data when extracting the "beat sheet." * Capability inventory: The agent is authorized to generate code (HTML/CSS/JS for Hyperframes or React for Remotion) and execute package management commands (npm install, npx) in SKILL.md. * Sanitization: No explicit sanitization or filtering is defined for the external transcript content before processing.
[EXTERNAL_DOWNLOADS]: The skill instructs the agent to install the hyperframes package from the npm registry and use the create-video starter for Remotion in SKILL.md. These are standard dependencies for the programmatic video workflows described.

video