The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides numerous instructions for the agent to run CLI commands, including npx create-video for project initialization, npx remotion studio for development previews, and npx remotion add for installing ecosystem-specific packages. It also uses npx remotion ffmpeg and npx remotion ffprobe for media processing.
[EXTERNAL_DOWNLOADS]: The skill fetches assets and configuration files from several external domains. This includes official Remotion media assets from remotion.media, animation data from lottiefiles.com, and map style definitions from maplibre.org. Additionally, it provides a script for downloading and installing the whisper.cpp binary via the @remotion/install-whisper-cpp package to facilitate audio transcription.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process and render data from external, potentially untrusted sources such as JSON captions, SRT subtitle files, Lottie animations, and dynamic API metadata. This introduces a surface where malicious instructions could be embedded in media assets to influence the agent's behavior.
Ingestion points: Data is loaded from external JSON and SRT files in rules/display-captions.md, rules/import-srt-captions.md, and rules/lottie.md, and from dynamic URLs in rules/calculate-metadata.md.
Boundary markers: There are no explicit instructions or delimiters defined to separate untrusted external data from the agent's primary instructions.
Capability inventory: The skill possesses capabilities for shell command execution (via npx and ffmpeg), file system writes (for generating audio and caption files), and network access (for fetching assets).
Sanitization: Content from external files is parsed (e.g., as JSON or SRT) but not otherwise sanitized before being used in the rendering logic or the agent's context.

remotion