remotion-best-practices

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.80). The skill demonstrably fetches and ingests content from arbitrary remote URLs (e.g., SKILL.md and rules/compositions.md's calculateMetadata example fetch(https://api.example.com/video/${props.videoId}) to set duration/props, rules/lottie.md fetching a Lottie JSON, rules/import-srt-captions allowing fetch() of remote .srt, and numerous Mediabunny UrlSource examples), which are untrusted third‑party sources and whose contents are parsed and used to drive metadata, rendering decisions, or tool behavior.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The transcribe-captions rule (see https://www.remotion.dev/docs/install-whisper-cpp) instructs using installWhisperCpp() and downloadWhisperModel() to download whisper.cpp and model files at runtime and then run transcribe(), which clearly fetches remote code/models and executes them.