The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates processing untrusted data from external sources, which creates a vulnerability surface for indirect prompt injection.
Ingestion points: Untrusted data enters the agent context via fetch calls in calculateMetadata.md (retrieving data from props.dataUrl) and lottie.md (retrieving animation data from external LottieFiles URLs).
Boundary markers: The provided code templates do not include boundary markers or explicit instructions to the AI to ignore embedded commands in the fetched content.
Capability inventory: The skill involves powerful capabilities including shell command execution via npx remotion ffmpeg (ffmpeg.md), network operations via fetch (voiceover.md, lottie.md, calculateMetadata.md), and file system writes in the suggested generate-voiceover.ts script (voiceover.md).
Sanitization: There is no evidence of sanitization, validation, or escaping logic for external content before it is processed or used to influence the video rendering pipeline.
[COMMAND_EXECUTION]: The skill instructs the user or agent to execute shell commands using the Remotion CLI and Node.js.
Evidence: The documentation provides numerous examples of command execution, such as npx remotion add, npx remotion studio, npx remotion still, npx remotion ffmpeg, npx remotion ffprobe, and node --strip-types generate-voiceover.ts.
[EXTERNAL_DOWNLOADS]: The skill references and fetches resources from several external domains and services.
Evidence: It suggests fetching TTS audio from ElevenLabs (api.elevenlabs.io), Lottie animations from LottieFiles (assets4.lottiefiles.com), and sound effects from a third-party GitHub repository (github.com/kapishdima/soundcn). Standard vendor resources are fetched from remotion.media and official package registries.

remotion-markup