The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process content from external sources, creating a surface for indirect prompt injection.
Ingestion points: The skill fetches data from user-provided URLs in rules/calculate-metadata.md (props.dataUrl) and rules/compositions.md (props.videoId). Additionally, rules/tailwind.md explicitly instructs the agent to fetch instructions from an external site (https://www.remotion.dev/docs/tailwind).
Capability inventory: The skill has access to command execution (via npx, npm, bun, etc.), file system operations (writing JSON in rules/transcribe-captions.md), and binary installation (installWhisperCpp).
Boundary markers: There are no instructions or markers to prevent the agent from following malicious commands embedded in the fetched data or documentation.
Sanitization: No sanitization or validation of the external content is implemented before it influences the agent's context or script execution.
[EXTERNAL_DOWNLOADS]: The skill downloads various assets and tools from remote servers.
Fetches media assets from remotion.media and lottiefiles.com.
Downloads map styles from maplibre.org.
Installs the whisper.cpp binary and speech-to-text models in rules/transcribe-captions.md.
[COMMAND_EXECUTION]: The skill provides instructions for running numerous shell commands for development and rendering.
Extensive use of npx remotion for project management, previewing, and rendering.
Instructions for package installation across different managers (npm, bun, yarn, pnpm).
Execution of FFmpeg utilities via the Remotion CLI.

remotion-best-practices