Auto Video Editing

This skill appears to do what it claims: local video/audio processing + Whisper-based transcription + FFmpeg-based editing. Before installing or running: - Expect large model downloads and network access (HuggingFace / mirrors, Google Fonts / jsDelivr) when you run transcription or when fonts are missing. This is normal for Whisper-based tools but can consume significant bandwidth and disk space. - The package asks you to pip-install faster-whisper/openai-whisper manually; review those packages and run them inside a virtualenv. Faster-whisper will download models from HuggingFace by default — you can choose mirrors if needed. - The repository owner/homepage are not provided; if provenance matters, consider auditing utils.py (it handles font/model downloads and platform detection) and running the skill in an isolated environment (VM or container) on sample files first. - The skill does not request secrets or remote endpoints for uploading your videos, but it will print transcript text (generate_cover may print TRANSCRIPT_FOR_TITLE_GENERATION to stdout) and may send network requests only to fetch models/fonts. If you want to avoid any external network activity, do not run the transcribe or font-download steps without reviewing and pre-provisioning models/fonts locally. If you want, I can also: (a) scan utils.py for exact network calls and mirror configuration, or (b) produce step-by-step safe install/run instructions (virtualenv, offline model use).