The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill documents patterns for ingesting and processing external data, which introduces a theoretical surface for indirect prompt injection.
Ingestion points: Loading external Lottie JSON files (rules/lottie.md), fetching data for composition metadata (rules/calculate-metadata.md), and importing subtitle files (rules/display-captions.md).
Capability inventory: The skill leverages subprocess execution via FFmpeg (rules/ffmpeg.md), file system writes (rules/transcribe-captions.md), and network requests to external APIs (rules/voiceover.md).
Sanitization: No specific boundary markers or sanitization logic are defined for the handled external content.
[EXTERNAL_DOWNLOADS]: The documentation identifies several third-party dependencies and external tools required for specific functionalities.
Packages: Recommends installing mediabunny for media inspection and mapbox-gl for map animations.
Binaries: The transcription workflow (rules/transcribe-captions.md) involves downloading the Whisper.cpp binary via the official @remotion/install-whisper-cpp vendor package.
Assets: References external sound effect repositories on GitHub (rules/sfx.md).
[COMMAND_EXECUTION]: Provides instructions for executing CLI tools within the development environment.
Media Tools: Includes examples for running bunx remotion ffmpeg and bunx remotion ffprobe for video trimming and metadata inspection (rules/ffmpeg.md).

remotion-best-practices