Agent Skills: Download from Video Platforms

Download from Video Platforms

Download videos, audio, and captions from YouTube and 1000+ video platforms using yt-dlp.

Confirmation Required

IMPORTANT: Before executing any download, you MUST confirm with the user using AskUserQuestion:

1. Show the URL to download
2. Show the quality setting (audio/video)
3. Show the output directory
4. Ask for confirmation

Example confirmation:

Ready to download:
- URL: https://youtube.com/watch?v=xxx
- Type: Audio only / Video (1080p)
- Save to: Current directory

Confirm download?

Only proceed with the download command after user confirms.

When to Use

• Download YouTube videos/audio for offline use
• Extract captions from video platforms
• Get audio for local transcription or editing

When NOT to Use

• Just need transcription (use /omnicaptions:transcribe - Gemini handles URLs directly)
• Converting existing caption formats (use /omnicaptions:convert)

Setup

pip install omni-captions-skills --extra-index-url https://lattifai.github.io/pypi/simple/

CLI Usage

Note: By default, files are saved to the current working directory. Do not specify -o unless the user explicitly requests a different location.

# Download audio only (default, saves to current directory)
omnicaptions download "https://www.youtube.com/watch?v=VIDEO_ID"

# Supports bare YouTube video ID (auto-validates via yt-dlp)
omnicaptions download e882eXLtwkI

# Download video (1080p recommended)
omnicaptions download "https://youtube.com/watch?v=VIDEO_ID" -q 1080p

# Only use -o when user explicitly requests a different location
omnicaptions download "https://youtube.com/watch?v=VIDEO_ID" -o ./downloads/

| Option | Description | |--------|-------------| | -o, --output | Output directory (default: current) | | -q, --quality | Quality: audio (default), best, 1080p, 720p, 480p, 360p | | -v, --verbose | Verbose output |

Quality Presets

| Preset | Description | |--------|-------------| | audio | Audio only (m4a/mp3), smallest size | | 1080p | 1080p video + audio (recommended for video) | | 720p | 720p video + audio | | 480p | 480p video + audio | | 360p | 360p video + audio | | best | Best available quality (may be 4K+, very large) |

Output

Downloads produce:

• Audio/Video file: .m4a, .mp4, etc.
• Captions (if available): .vtt or .srt
• Metadata: .meta.json (video resolution, title, etc. for ASS font scaling)

Video: ./VIDEO_ID.mp4
Audio: ./VIDEO_ID.m4a
Caption: ./VIDEO_ID.en.vtt
Metadata: ./VIDEO_ID.meta.json  # Used by convert for auto font size
Title: Video Title Here

The .meta.json file stores video resolution, which omnicaptions convert uses to auto-calculate font size for ASS karaoke output.

Supported Platforms

YouTube, Bilibili, Vimeo, Twitter/X, and 1000+ sites.

Related Skills

| Skill | Use When | |-------|----------| | /omnicaptions:transcribe | Transcribe downloaded audio/video | | /omnicaptions:translate | Translate captions with Gemini | | /omnicaptions:translate | Translate captions with Claude (no API) | | /omnicaptions:convert | Convert caption format |

Workflow Examples

Important: Generate bilingual captions AFTER LaiCut alignment. Preserve language tag in filename.

# Has caption: download → LaiCut align (JSON) → convert → translate
omnicaptions download "https://youtube.com/watch?v=xxx"
# → xxx.en.vtt
omnicaptions LaiCut xxx.mp4 xxx.en.vtt
# → xxx.en_LaiCut.json
omnicaptions convert xxx.en_LaiCut.json -o xxx.en_LaiCut.srt
# → xxx.en_LaiCut_Claude_zh.srt (after translate)

# No caption: download → transcribe → LaiCut align (JSON) → convert → translate
omnicaptions download "https://youtube.com/watch?v=xxx"
omnicaptions transcribe xxx.mp4
# → xxx_GeminiUnd.md
omnicaptions LaiCut xxx.mp4 xxx_GeminiUnd.md
# → xxx_GeminiUnd_LaiCut.json
omnicaptions convert xxx_GeminiUnd_LaiCut.json -o xxx_GeminiUnd_LaiCut.srt
# → xxx_GeminiUnd_LaiCut_Claude_zh.srt (after translate)