remotion-best-practices

Pass

Audited by Gen Agent Trust Hub on Nov 14, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill suggests the installation of numerous official Remotion packages and third-party libraries (such as mapbox-gl, @turf/turf, and zod) from the npm registry to support video composition features.
  • [COMMAND_EXECUTION]: The skill utilizes the Remotion CLI and Node.js execSync to invoke ffmpeg and ffprobe for essential media operations including video trimming, transcoding, and audio extraction.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data such as JSON/SRT captions, Lottie animations, and remote API responses fetched during the metadata calculation phase. This data is used to drive video composition properties and subprocess calls.
  • Ingestion points: Remote API fetching in calculateMetadata (rules/calculate-metadata.md), external caption file loading (rules/display-captions.md, rules/import-srt-captions.md), and remote asset retrieval for Lottie animations (rules/lottie.md).
  • Boundary markers: Not explicitly implemented in the provided rule snippets.
  • Capability inventory: Subprocess execution via FFmpeg and file writing to the public/ directory for audio transcription outputs.
  • Sanitization: No explicit content sanitization or validation steps are defined for external data before its interpolation into the rendering pipeline.
  • [DYNAMIC_EXECUTION]: The skill provides a workflow for audio transcription that involves the dynamic installation of the whisper.cpp engine and the runtime download of AI models to local storage for processing (rules/transcribe-captions.md).
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 14, 2026, 08:43 AM