The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external media and metadata, creating a standard surface for indirect prompt injection via untrusted user-supplied content.
Ingestion points: Files rules/display-captions.md, rules/import-srt-captions.md, rules/videos.md, and rules/audio.md describe loading external captions (JSON/SRT), audio, and video assets.
Boundary markers: The instructions do not define explicit delimiters or instructions for the agent to ignore potential commands embedded within data fields of ingested files (e.g., subtitle text).
Capability inventory: The skill utilizes subprocess calls for FFmpeg (rules/ffmpeg.md) and network operations via fetch for ElevenLabs TTS and Mapbox services (rules/voiceover.md, rules/mapbox.md).
Sanitization: There are no specific instructions for sanitizing or validating the content of ingested captions or external metadata.
[COMMAND_EXECUTION]: The skill utilizes standard Remotion CLI tools through npx for project scaffolding, studio previewing, and video processing with FFmpeg. These operations are within the expected scope of the skill's functionality and involve standard development commands.
[EXTERNAL_DOWNLOADS]: The skill references and installs several standard Node.js packages from the Remotion ecosystem and well-known libraries such as mapbox-gl, zod, and mediabunny. It also fetches data from established services like Mapbox, ElevenLabs, and LottieFiles, which is routine for the intended use case.

remotion-best-practices