The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides terminal commands for project scaffolding via npx create-video, studio previewing via npx remotion studio, and static frame rendering with npx remotion still.
[EXTERNAL_DOWNLOADS]: It facilitates the installation of various Remotion plugins and external libraries such as mediabunny, maplibre-gl, and zod. The transcription module specifically handles the download of the Whisper.cpp binary and pre-trained AI models.
[DYNAMIC_EXECUTION]: Documentation describes creating Node.js scripts for AI voiceover and transcription that manage file writing and remote API calls. The html-in-canvas rule utilizes browser APIs for frame-by-frame rendering logic in a canvas context.
[INDIRECT_PROMPT_INJECTION]: The skill contains patterns for ingesting external data sources that could influence composition behavior.
Ingestion points: calculateMetadata fetches from remote APIs; Lottie components load animation JSON; parseSrt processes subtitle files.
Boundary markers: None specified in the instructional text.
Capability inventory: Filesystem writing via fs.writeFileSync, network operations via fetch, and command execution through the Remotion CLI and Whisper binaries.
Sanitization: Not explicitly detailed in the provided coding examples.

remotion-best-practices