remotion-best-practices

Pass

Audited by Gen Agent Trust Hub on Mar 6, 2026

Risk Level: SAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill guides users to install various official and standard packages required for video production, including @remotion/three, @remotion/media, mapbox-gl, zod, and mediabunny.
  • [COMMAND_EXECUTION]: Provides instructions for using Remotion's CLI tools (e.g., npx remotion add, bunx remotion ffmpeg) to manage project dependencies and perform video processing tasks.
  • [REMOTE_CODE_EXECUTION]: Utilizes the @remotion/install-whisper-cpp library which programmatically installs the Whisper.cpp binaries and models. This is a documented and intended functionality for audio transcription within the Remotion ecosystem.
  • [DATA_EXFILTRATION]: Instructs users on how to manage sensitive environment variables like ELEVENLABS_API_KEY and REMOTION_MAPBOX_TOKEN. These are used for legitimate authentication with established third-party services (ElevenLabs and Mapbox).
  • [INDIRECT_PROMPT_INJECTION]: The skill establishes several data ingestion points where external, potentially untrusted data is processed, such as fetching Lottie JSON animations, SRT subtitles, and caption files.
  • Ingestion points: rules/lottie.md (Lottie JSON), rules/import-srt-captions.md (SRT files), rules/display-captions.md (Caption JSON).
  • Boundary markers: Not explicitly defined in code snippets, though this is typical for asset-loading logic.
  • Capability inventory: Uses fetch(), installWhisperCpp(), and extractFrames() (via Mediabunny).
  • Sanitization: Standard React rendering and Mediabunny parsing are used, which provide baseline protection, though developers should remain cautious of untrusted asset sources.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 6, 2026, 03:24 PM