The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes automation and utility scripts that execute shell commands, such as ffmpeg for video processing in examples/kern-calibration/assets/generate-i2v.py and git for repository consistency checks in tools/check-consistency.mjs. These operations are well-scoped and essential for the tool's functionality.
[EXTERNAL_DOWNLOADS]: Examples within the skill fetch libraries like three.js, gsap, and model-viewer from well-known CDNs including unpkg.com and jsdelivr.net. It also supports optional asset generation via the fal.ai service. All external references target reputable technology providers.
[CREDENTIALS_UNSAFE]: The skill utilizes a billable API key (FAL_KEY) for AI generation tasks. It implements strict safeguards, instructing users to store keys in gitignored .env.local files and providing a consistency scanner (tools/check-consistency.mjs) that proactively checks for accidentally leaked secrets.
[DYNAMIC_EXECUTION]: The compile-choreography.mjs script provides a core feature that compiles a declarative JSON schema into runnable GSAP animation code. This process is highly structured and does not involve eval() or other high-risk patterns for executing untrusted data.
[SAFE]: The skill demonstrates a high level of security awareness, employing Subresource Integrity (SRI) for versioned CDN scripts and maintaining clear boundaries between local development and production deployment.

cinematic-scroll