TubeScribe 🎬

Turn any YouTube video into a polished document + audio summary.

Drop a YouTube link → get a beautiful transcript with speaker labels, key quotes, timestamps that link back to the video, and an audio summary you can listen to on the go.

💸 Free & No Paid APIs

• No subscriptions or API keys — works out of the box
• Local processing — transcription, speaker detection, and TTS run on your machine
• Network access — fetching from YouTube (captions, metadata, comments) requires internet
• No data uploaded — nothing is sent to external services; all processing stays on your machine
• Safe sub-agent — spawned sub-agent has strict instructions: no software installation, no network calls beyond YouTube

✨ Features

• 📄 Transcript with summary and key quotes — Export as DOCX, HTML, or Markdown
• 🎯 Smart Speaker Detection — Automatically identifies participants
• 🔊 Audio Summaries — Listen to key points (MP3/WAV)
• 📝 Clickable Timestamps — Every quote links directly to that moment in the video
• 💬 YouTube Comments — Viewer sentiment analysis and best comments
• 📋 Queue Support — Send multiple links, they get processed in order
• 🚀 Non-Blocking Workflow — Conversation continues while video processes in background

🎬 Works With Any Video

• Interviews & podcasts (multi-speaker detection)
• Lectures & tutorials (single speaker)
• Music videos (lyrics extraction)
• News & documentaries
• Any YouTube content with captions

Quick Start

When user sends a YouTube URL: 1. Spawn sub-agent with the full pipeline task immediately 2. Reply: "🎬 TubeScribe is processing — I'll let you know when it's ready!" 3. Continue conversation (don't wait!) 4. Sub-agent notification will announce completion with title and details

DO NOT BLOCK — spawn and move on instantly.

First-Time Setup

Run setup to check dependencies and configure defaults:

python skills/tubescribe/scripts/setup.py

This checks: summarize CLI, pandoc, ffmpeg, Kokoro TTS

Full Workflow (Single Sub-Agent)

Spawn ONE sub-agent that does the entire pipeline:

sessions_spawn(
    task=f"""
## TubeScribe: Process {youtube_url}

⚠️ CRITICAL: Do NOT install any software.
No pip, brew, curl, venv, or binary downloads.
If a tool is missing, STOP and report what's needed.

Run the COMPLETE pipeline — do not stop until all steps are done.

### Step 1: Extract
```bash
python3 skills/tubescribe/scripts/tubescribe.py "{youtube_url}"

Note the Source and Output paths printed by the script. Use those exact paths in subsequent steps.

Step 2: Read source JSON

Read the Source path from Step 1 output and note: - metadata.title (for filename) - metadata.video_id - metadata.channel, upload_date, duration_string

Step 3: Create formatted markdown

Write to the Output path from Step 1:

1. # **<title>**

2. Video info block — Channel, Date, Duration, URL (clickable). Empty line between each field.

1. ## **Participants** — table with bold headers: | **Name** | **Role** | **Description** | |----------|----------|-----------------| ---

4. ## **Summary** — 3-5 paragraphs of prose

1. ## **Key Quotes** — 5 best with clickable YouTube timestamps. Format each as: ``` "Quote text here." - 12:34

"Another quote." - 25:10 ```

Use regular dash -, NOT em dash —. Do NOT use blockquotes >. Plain paragraphs only.

6. ## **Viewer Sentiment** (if comments exist)

1. ## **Best Comments** (if comments exist) — Top 5, NO lines between them: ``` Comment text here.

- ▲ 123 @AuthorName

Next comment text here.

- ▲ 45 @AnotherAuthor `` Attribution line: dash + italic. Just blank line between comments, NO---` separators.

1. ## **Full Transcript** — merge segments, speaker labels, clickable timestamps

Step 4: Create DOCX

Clean the title for filename (remove special chars), then: bash pandoc <output_path> -o ~/Documents/TubeScribe/<safe_title>.docx

Step 5: Generate audio

Write the summary text to a temp file, then use TubeScribe's built-in audio generation: ```bash

Write summary to temp file (use python3 to write, avoids shell escaping issues)

python3 -c " text = '''YOUR SUMMARY TEXT HERE''' with open('/tubescribe__summary.txt', 'w') as f: f.write(text) "

Generate audio (auto-detects engine, voice, format from config)

python3 skills/tubescribe/scripts/tubescribe.py \ --generate-audio /tubescribe__summary.txt \ --audio-output ~/Documents/TubeScribe/_summary `` This reads~/.tubescribe/config.json` and uses the configured TTS engine (mlx/kokoro/builtin), voice blend, and speed automatically. Output format (mp3/wav) comes from config.

Step 6: Cleanup

python3 skills/tubescribe/scripts/tubescribe.py --cleanup <video_id>

Step 7: Open folder

open ~/Documents/TubeScribe/

Report

Tell what was created: DOCX name, MP3 name + duration, video stats. """, label="tubescribe", runTimeoutSeconds=900, cleanup="delete" ) ```

After spawning, reply immediately:

🎬 TubeScribe is processing - I'll let you know when it's ready! Then continue the conversation. The sub-agent notification announces completion.

Configuration

Config file: ~/.tubescribe/config.json

{
  "output": {
    "folder": "~/Documents/TubeScribe",
    "open_folder_after": true,
    "open_document_after": false,
    "open_audio_after": false
  },
  "document": {
    "format": "docx",
    "engine": "pandoc"
  },
  "audio": {
    "enabled": true,
    "format": "mp3",
    "tts_engine": "mlx"
  },
  "mlx_audio": {
    "path": "~/.openclaw/tools/mlx-audio",
    "model": "mlx-community/Kokoro-82M-bf16",
    "voice": "af_heart",
    "lang_code": "a",
    "speed": 1.05
  },
  "kokoro": {
    "path": "~/.openclaw/tools/kokoro",
    "voice_blend": { "af_heart": 0.6, "af_sky": 0.4 },
    "speed": 1.05
  },
  "processing": {
    "subagent_timeout": 600,
    "cleanup_temp_files": true
  }
}

Output Options

Document Options

Audio Options

MLX-Audio Options (preferred on Apple Silicon)

Kokoro PyTorch Options (fallback)

Processing Options

Comment Options

Queue Options

Output Structure

~/Documents/TubeScribe/
├── {Video Title}.html         # Formatted document (or .docx / .md)
└── {Video Title}_summary.mp3  # Audio summary (or .wav)

After generation, opens the folder (not individual files) so you can access everything.

Dependencies

Required: - summarize CLI — brew install steipete/tap/summarize - Python 3.8+

Optional (better quality): - pandoc — DOCX output: brew install pandoc - ffmpeg — MP3 audio: brew install ffmpeg - yt-dlp — YouTube comments: brew install yt-dlp - mlx-audio — Fastest TTS on Apple Silicon: pip install mlx-audio (uses MLX backend for Kokoro) - Kokoro TTS — PyTorch fallback: see https://github.com/hexgrad/kokoro

yt-dlp Search Paths

TubeScribe checks these locations (in order):

If not found, setup downloads a standalone binary to the tools directory. The tools directory version doesn't conflict with system installations.

Queue Handling

When user sends multiple YouTube URLs while one is processing:

Check Before Starting

python skills/tubescribe/scripts/tubescribe.py --queue-status

If Already Processing

# Add to queue instead of starting parallel processing
python skills/tubescribe/scripts/tubescribe.py --queue-add "NEW_URL"
# → Replies: "📋 Added to queue (position 2)"

After Completion

# Check if more in queue
python skills/tubescribe/scripts/tubescribe.py --queue-next
# → Automatically pops and processes next URL

Queue Commands

Batch Processing (multiple URLs at once)

python skills/tubescribe/scripts/tubescribe.py url1 url2 url3

Processes all URLs sequentially with a summary at the end.

Error Handling

The script detects and reports these errors with clear messages:

When an error occurs, report it to the user and don't proceed with that video.

Tips

• For long videos (>30 min), increase sub-agent timeout to 900s
• Speaker detection works best with clear interview/podcast formats
• Single-speaker videos (tutorials, lectures) skip speaker labels automatically
• Timestamps link directly to YouTube at that moment
• Use batch mode for multiple videos: tubescribe url1 url2 url3