The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides instructions to run npx remotion commands for dependency management and video rendering. It specifically guides the user to use commands like npx remotion add to install ecosystem packages and npx remotion render with specific flags for processing.
[EXTERNAL_DOWNLOADS]: The skill documents methods to download various @remotion ecosystem packages and third-party libraries such as mapbox-gl, zod, and @turf/turf from the NPM registry. It also fetches assets from known services like Lottiefiles and official media repositories for use in video compositions.
[INDIRECT_PROMPT_INJECTION]: The skill documents ingestion of external data which constitutes a potential attack surface for indirect prompt injection.
Ingestion points: The skill fetching patterns retrieve JSON data, SRT subtitles, and Lottie assets from remote URLs provided in component props or fixed endpoints (documented in rules/calculate-metadata.md, rules/lottie.md, and rules/import-srt-captions.md).
Boundary markers: The provided code examples do not demonstrate the use of delimiters or explicit instructions for the agent to ignore or isolate potentially malicious content within the processed external data.
Capability inventory: The skill environment supports significant capabilities including shell command execution via the npx CLI, network operations via the fetch API, and configuration via environment variables.
Sanitization: There is no documentation of data validation, escaping, or sanitization steps for the text or metadata fetched from external sources before it is rendered or used in logic.

remotion-best-practices