Video Subtitle Downloader

Download and convert subtitles from YouTube, Bilibili, and 1000+ platforms into SRT, JSON, or TXT with automatic timestamps and GPU acceleration support.

What to consider before installing: - The code you get is simple: it runs yt-dlp to fetch subtitles and saves converted files. There is no built-in transcription pipeline in the provided scripts despite the README/SKILL.md claiming GPU-accelerated transcription with faster-whisper. If you need audio->text transcription, expect to implement or verify that functionality yourself. - The skill recommends pip installing faster-whisper (and other heavy packages). Faster-whisper and similar models can download large model files and consume GPU/CPU and disk space; only install them if you trust the source and need that feature. - yt-dlp will contact the video hosting sites you provide and may download data depending on options. Review yt-dlp behavior and ensure you comply with copyright and site terms. - There is no network exfiltration in the scripts provided, and no credentials are requested. Still, if the final distribution you install differs from these files (different scripts or additional install steps), re-review for hidden network endpoints, telemetry, or credential use. - Actionable steps: (1) run the scripts in a controlled environment (sandbox) on sample URLs; (2) inspect/grep any distributed version for calls to faster-whisper or other subprocess/network calls; (3) if you expect transcription/GPU support, request/verify an implementation that actually invokes faster-whisper and documents model downloads and resource needs. If the publisher cannot explain the discrepancy between docs and code, treat the package as untrusted.

Type: OpenClaw Skill Name: video-subtitle-downloader Version: 1.0.0 The skill bundle provides a legitimate utility for downloading video subtitles from various platforms using the well-known 'yt-dlp' and 'faster-whisper' libraries. The Python scripts (scripts/download_subtitle.py and scripts/batch_download.py) implement standard command-line interfaces and use secure subprocess calls (passing arguments as lists) to interact with external tools, minimizing shell injection risks. No evidence of data exfiltration, malicious persistence, or harmful prompt injection was found.