Youtube Transcriber

YouTube Transcriber Skill

Extract transcripts from YouTube videos, playlists, and channels with automatic intelligent processing.

Overview

One unified command that intelligently assesses input and handles everything—single videos, batch files, playlist expansion, channel extraction. No need to choose between commands; it figures out what to do.

Capabilities

• Auto-Detect Input: Single URL, file of URLs, playlist, channel
• Smart Expansion: Automatically expands playlists/channels to individual videos
• Batch Processing: Process 1-1000+ videos from mixed input file
• Timestamps: Automatic HH:MM:SS formatting on every line
• Multiple Formats: Markdown (.md) or plain text (.txt)
• Error Handling: Graceful failures, continues on errors

Quick Usage

# Single video
/transcribe https://www.youtube.com/watch?v=VIDEO_ID

# Playlist (auto-expands, downloads all)
/transcribe https://www.youtube.com/playlist?list=PL123

# Channel (auto-expands, downloads all uploads)
/transcribe https://www.youtube.com/@creator

# Batch file (auto-detects each URL type, expands if needed)
/transcribe urls.txt

# With options
/transcribe urls.txt --format md --output-dir research

One Command, Many Inputs

| Input | Behavior | |-------|----------| | Video URL | Download transcript immediately | | Playlist URL | Expand to all videos, download transcripts | | Channel URL | Expand to all uploads, download transcripts | | File (mixed) | Detect each line, expand playlists/channels, download all | | File (videos only) | Batch download all transcripts | | Video ID | Download transcript immediately |

How It Works

Input Assessment
  ↓
├─ Single URL?
│  ├─ Video → Download immediately
│  ├─ Playlist → Expand + Download
│  └─ Channel → Expand + Download
│
└─ File?
   ├─ Detect each line
   ├─ Expand playlists/channels
   └─ Download all transcripts
     ↓
  Output Directory (with all transcripts)

Options

/transcribe INPUT [OPTIONS]

Options:
  --format {md|txt|both}     Output format (default: both)
  --output-dir DIR           Output directory (default: transcripts)
  --expand/--no-expand       Auto-expand playlists/channels (default: on)
  --inspect                  Show what will be done (no download)

Examples

Single Video

/transcribe https://youtu.be/dQw4w9WgXcQ
# → Creates: transcripts/transcript_dQw4w9WgXcQ.md
#           transcripts/transcript_dQw4w9WgXcQ.txt

Playlist (Auto-Expansion)

/transcribe https://www.youtube.com/playlist?list=PLtqRgJ_TIq8Y6YG8G
# → Expands to all 50 videos
# → Downloads all 50 transcripts
# → Creates: transcripts/transcript_ID1.md, transcript_ID2.md, ...

Batch File (Mixed URLs)

# Create file:
cat > urls.txt  sources.txt
echo "https://www.youtube.com/@educator" >> sources.txt

# One command: expand, download, ready for analysis
/transcribe sources.txt --output-dir research

Content Curation

# Download transcripts from multiple creators
/transcribe creators.txt --format md
# → All uploads from all channels
# → Organized transcripts ready for processing

Quick Lookup

# Just need one video's transcript
/transcribe https://youtu.be/VIDEO_ID

Batch Research

# Process everything at once
/transcribe mixed_urls.txt --output-dir analysis
# → Auto-detects each line
# → Auto-expands playlists/channels
# → Downloads all transcripts
# → Organized in output-dir

Integration with Agents

In Workflow

Agent receives: List of YouTube URLs (mixed types)
  ↓
/transcribe urls.txt --inspect          (check what will process)
  ↓
/transcribe urls.txt --output-dir data  (download all transcripts)
  ↓
Agent reads transcript directory
  ↓
Agent processes/analyzes transcripts
  ↓
Agent generates report

Function Call (Python)

from skill import transcribe

result = transcribe(
    input="urls.txt",
    format="md",
    output_dir="transcripts",
    expand=True,
    inspect=False
)

# Returns:
# {
#   "success": True,
#   "processed": 150,
#   "output_dir": "transcripts",
#   "files": ["transcript_ID1.md", ...]
# }

Limitations

• YouTube must allow transcripts for the video
• Some videos may have restricted captions
• Large playlists (500+ videos) expand slowly
• Language support depends on available transcripts (default: English)

Requirements

• youtube-transcript-api (0.6.2)
• yt-dlp (≥2024.7.1)
• click (8.1.7)

Author

Aaron Storey | Research & Development (Nov 2025)

Version

1.1.0 - Unified single-command interface with automatic assessment

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: astoreyai
• Source: astoreyai/claude-skills
• License: MIT

Install and usage instructions live in the source repository linked above.