Skills-SI市场
skills-si.com
提交技能
/
界面已本地化；技能介绍与说明文档按所选语言由 AI 自动翻译，代码、命令、路径和专业术语保留原文，以原作者文档为准。
探索技能
效率与协作 · SKILL.md

shorts-creation

shorts-creation

【效率与协作】Transforms long YouTube/Google Drive videos into multiple captioned vertical shorts through a multi-step workflow

独立 GitHub 快照 · 自动分类 · 以原作者文档为准

Bbuda-ai / bunny-agentApache-2.0

技能说明

GitHub

name: shorts-creation description: Transforms long YouTube/Google Drive videos into multiple captioned vertical shorts through a multi-step workflow

Shorts Creation (Multi-Task Workflow)

When to Use This Skill

  • Have a long-form YouTube or Google Drive video to repurpose
  • Need vertical short clips with subtitles for social platforms
  • Want to extract viral-worthy segments from existing content
  • Need 9:16 format videos for YouTube Shorts, Instagram Reels, TikTok
  • Want to maximize content reach by creating multiple shorts from one video

What This Skill Does

This skill executes a multi-task sequential workflow:

  1. Download Video: Retrieves video from YouTube or Google Drive
  2. Transcribe Audio: Converts speech to text using OpenAI Whisper
  3. AI Segment Recommendation: Uses Gemini to identify viral-worthy segments
  4. Subtitle Configuration: Confirms font, size, color, position with user
  5. Generate Shorts: Creates vertical 9:16 videos with burned-in subtitles
  6. Export & Upload: Saves as MP4 and uploads to Google Drive

Multi-Task Workflow Overview

┌─────────────────────┐
│  1. Download Video  │ ← Input: YouTube URL or Drive URL
└──────────┬──────────┘
           ↓
┌─────────────────────┐
│ 2. Transcribe Audio │ ← Tool: OpenAI Whisper
└──────────┬──────────┘
           ↓
┌─────────────────────┐
│ 3. Recommend Clips  │ ← Tool: Gemini AI
└──────────┬──────────┘
           ↓
┌─────────────────────┐
│ 4. Configure Subs   │ ← User confirms settings
└──────────┬──────────┘
           ↓
┌─────────────────────┐
│ 5. Generate Shorts  │ ← FFmpeg + subtitle burn-in
└──────────┬──────────┘
           ↓
┌─────────────────────┐
│ 6. Upload to Drive  │ ← Tool: Google Drive MCP
└─────────────────────┘

How to Use

From YouTube URL:

Create shorts from this YouTube video: https://youtube.com/watch?v=abc123
Target: 3-5 shorts, 30-45 seconds each

From Google Drive:

Make shorts from this Drive video: https://drive.google.com/file/d/FILE_ID
Focus on tutorial segments
Add yellow subtitles

With Specific Preferences:

Create shorts from: https://youtube.com/watch?v=xyz789
Number of shorts: 5
Duration: 15-30 seconds
Subtitle style: Bold white text with black outline
Position: Bottom center

Instructions

Task 1: Download Video

Objective: Retrieve source video for processing

From YouTube:

  1. Validate YouTube URL format
  2. Use yt-dlp to download video: 
    yt-dlp -f "bestvideo[height<=1080]+bestaudio/best[height<=1080]" \
  -o "source_video.%(ext)s" \
  "https://youtube.com/watch?v=VIDEO_ID"
  3. Extract video metadata:
    • Title
    • Duration
    • Resolution
    • Channel name
  4. Verify download completed successfully
  5. Check file integrity

From Google Drive:

  1. Validate Google Drive share URL
  2. Use Google Drive MCP to download: 
    Tool: google-drive
Method: download_file
Parameters: {
  "file_id": "extracted_from_url",
  "destination": "source_video.mp4"
}
  3. Extract video metadata (same as YouTube)

Output: Downloaded source video file

Task 2: Transcribe Audio to Text

Objective: Generate accurate transcript with timestamps

Using OpenAI Whisper:

  1. Extract audio from video:

    ffmpeg -i source_video.mp4 -vn -acodec pcm_s16le -ar 16000 \
  -ac 1 audio.wav

  2. Transcribe with Whisper (if available via MCP or local):

    # Example using OpenAI Whisper API
import openai

with open("audio.wav", "rb") as audio_file:
    transcript = openai.Audio.transcribe(
        model="whisper-1",
        file=audio_file,
        response_format="verbose_json",
        timestamp_granularity="word"
    )

  3. Parse transcript into timestamped segments:

    {
  "segments": [
    {
      "start": 0.0,
      "end": 3.5,
      "text": "Welcome to this tutorial on React hooks"
    },
    {
      "start": 3.5,
      "end": 8.2,
      "text": "Today we're going to learn about useState"
    }
  ]
}

  4. Generate SRT subtitle file:

    1
00:00:00,000 --> 00:00:03,500
Welcome to this tutorial on React hooks

2
00:00:03,500 --> 00:00:08,200
Today we're going to learn about useState

Fallback (if Whisper unavailable):

  • Use YouTube auto-generated captions (if YouTube source)
  • Provide manual transcription instructions
  • Suggest alternative transcription services (Rev, Descript)

Output: Full transcript with timestamps (SRT format)

Task 3: AI Segment Recommendation (Gemini)

Objective: Identify most viral-worthy segments using AI analysis

Using Gemini AI:

  1. Prepare analysis prompt for Gemini:

    You are a viral content expert analyzing video transcripts to identify
the most engaging segments for social media shorts.

Video Title: [title]
Duration: [duration]
Platform: YouTube Shorts, Instagram Reels, TikTok

Full Transcript:
[Insert timestamped transcript]

Task: Identify the 5-7 best segments for short-form content. For each segment:
1. Start and end timestamps
2. Duration (15-60 seconds)
3. Hook quality (why it's engaging)
4. Viral potential (high/medium)
5. Recommended title
6. Target platform focus

Prioritize segments with:
- Strong opening hooks
- Self-contained complete thoughts
- Actionable tips or surprising facts
- Emotional or funny moments
- Clear value proposition

  2. Send to Gemini via Google AI MCP:

    Tool: google-ai (Gemini)
Method: generate_content
Parameters: {
  "model": "gemini-pro",
  "prompt": [analysis_prompt],
  "temperature": 0.7
}

  3. Parse Gemini's recommendations:

    {
  "recommendations": [
    {
      "segment_id": 1,
      "start": "00:00:45",
      "end": "00:01:15",
      "duration": 30,
      "hook": "Starts with common mistake everyone makes",
      "viral_potential": "high",
      "title": "Stop Making This React Mistake",
      "platform": "TikTok (problem-solution format)",
      "transcript": "Here's a mistake I see beginners make..."
    },
    // ... more segments
  ]
}

  4. Present recommendations to user:

    **AI-Recommended Segments:**

🔥 Segment 1: "Stop Making This React Mistake" (30s)
   Time: 0:45 - 1:15
   Hook: Opens with common mistake
   Viral Potential: HIGH
   Best for: TikTok, YouTube Shorts

⭐ Segment 2: "useState in 30 Seconds" (28s)
   Time: 3:10 - 3:38
   Hook: Quick, actionable tutorial
   Viral Potential: HIGH
   Best for: All platforms

[... more segments ...]

**Which segments would you like to create?**
Select: [All] [1,2,4] [Custom]

  5. Allow user to:

    • Accept all recommendations
    • Select specific segments
    • Provide custom timestamp ranges
    • Adjust segment boundaries

Output: Approved list of segments with timestamps and metadata

Task 4: Confirm Subtitle Settings

Objective: Configure subtitle appearance with user preferences

Present Configuration Options:

**Subtitle Configuration**

Please confirm or customize subtitle settings:

1. Font Family:
   [ ] Arial (Default)
   [ ] Helvetica
   [ ] Impact (Bold, attention-grabbing)
   [ ] Montserrat (Modern, clean)
   [Custom font name]

2. Font Size:
   [ ] Small (32px) - Subtle
   [ ] Medium (48px) - Default
   [ ] Large (64px) - High visibility
   [ ] Extra Large (80px) - Maximum impact

3. Font Color:
   [ ] White (Default)
   [ ] Yellow (High visibility)
   [ ] Cyan/Light Blue
   [Custom hex color: #FFFFFF]

4. Text Outline/Stroke:
   [ ] Black outline (3px) - Default
   [ ] No outline
   [ ] Custom color/thickness

5. Background:
   [ ] None (transparent) - Default
   [ ] Semi-transparent black box
   [ ] Subtle shadow

6. Position:
   [ ] Bottom Center (Default)
   [ ] Top Center
   [ ] Middle Center
   [ ] Custom (y-offset)

7. Animation:
   [ ] None - Static
   [ ] Fade in per word (TikTok style)
   [ ] Karaoke style (word highlight)

Default Settings:

Font: Arial Bold
Size: 48px
Color: White (#FFFFFF)
Outline: Black 3px
Background: None
Position: Bottom center (y=100px from bottom)
Animation: None (static)

Save Configuration:

{
  "subtitle_config": {
    "font": "Arial-Bold",
    "size": 48,
    "color": "#FFFFFF",
    "outline_color": "#000000",
    "outline_width": 3,
    "background": "none",
    "position": "bottom_center",
    "y_offset": 100,
    "animation": "none"
  }
}

Output: Finalized subtitle configuration for video generation

Task 5: Generate Vertical Shorts with Subtitles

Objective: Create 9:16 vertical videos with burned-in subtitles

For Each Approved Segment:

  1. Extract segment from source video:

    ffmpeg -ss 00:00:45 -i source_video.mp4 -t 30 -c copy segment_raw.mp4

  2. Convert to vertical 9:16 format:

    # Smart crop to vertical (center-focused)
ffmpeg -i segment_raw.mp4 \
  -vf "scale=1080:1920:force_original_aspect_ratio=increase,crop=1080:1920" \
  -c:a copy segment_vertical.mp4

  3. Extract segment subtitle from full transcript:

    1
00:00:00,000 --> 00:00:03,500
Here's a mistake I see beginners make

2
00:00:03,500 --> 00:00:07,200
They forget to include the dependency array

  4. Burn in subtitles with configured style:

    ffmpeg -i segment_vertical.mp4 -vf "subtitles=segment.srt:force_style='\
  FontName=Arial-Bold,\
  FontSize=48,\
  PrimaryColour=&HFFFFFF&,\
  OutlineColour=&H000000&,\
  Outline=3,\
  Alignment=2,\
  MarginV=100'" \
  -c:a copy segment_with_subs.mp4

    Alternative (drawtext for word-by-word):

    # For animated karaoke-style subtitles
ffmpeg -i segment_vertical.mp4 -vf "drawtext=textfile=words.txt:\
  fontfile=/path/to/Arial-Bold.ttf:fontsize=48:fontcolor=white:\
  borderw=3:bordercolor=black:x=(w-text_w)/2:y=h-100:\
  enable='between(t,START_TIME,END_TIME)'" \
  segment_with_subs.mp4

  5. Add fade in/out:

    ffmpeg -i segment_with_subs.mp4 \
  -vf "fade=t=in:st=0:d=0.5,fade=t=out:st=29.5:d=0.5" \
  -c:a copy segment_final.mp4

  6. Optimize for mobile:

    ffmpeg -i segment_final.mp4 \
  -c:v libx264 -preset medium -crf 23 \
  -vf "scale=1080:1920" \
  -c:a aac -b:a 128k \
  -movflags +faststart \
  short_01_final.mp4

  7. Apply naming convention:

    [source_title]_short_[01-99]_9x16.mp4

Example: react_tutorial_short_01_9x16.mp4

Repeat for all approved segments

Output: Multiple vertical short videos with burned-in subtitles

Task 6: Export and Upload to Google Drive

Objective: Save files locally and upload to cloud storage

Local Export:

  1. Create output directory structure:

    output/
├── shorts/
│   ├── react_tutorial_short_01_9x16.mp4
│   ├── react_tutorial_short_02_9x16.mp4
│   └── ...
├── metadata/
│   └── shorts_metadata.json
└── subtitles/
    ├── short_01.srt
    └── ...

  2. Generate metadata file:

    {
  "source_video": "Complete React Tutorial",
  "source_url": "https://youtube.com/watch?v=abc123",
  "created_date": "2026-01-07",
  "shorts_generated": 5,
  "shorts": [
    {
      "filename": "react_tutorial_short_01_9x16.mp4",
      "duration": 30,
      "original_timestamp": "00:00:45 - 00:01:15",
      "title": "Stop Making This React Mistake",
      "description": "Common useState mistake and how to fix it",
      "hashtags": ["#react", "#coding", "#programming", "#webdev"],
      "platform_recommendations": ["TikTok", "YouTube Shorts"]
    }
  ]
}

Upload to Google Drive:

  1. Use upload-to-drive skill to batch upload:

    Tool: google-drive (via upload-to-drive skill)

Upload all shorts to:
Video Projects/Shorts/React Tutorial/

Files:
- All MP4 files → Finals/
- Metadata JSON → Finals/
- SRT files → Subtitles/

  2. Generate shareable links for each short

  3. Create summary document in Drive:

    **Shorts Creation Summary**

Source: Complete React Tutorial
Created: 2026-01-07
Total Shorts: 5

**Files:**
1. react_tutorial_short_01_9x16.mp4
   Link: https://drive.google.com/file/d/...
   Title: Stop Making This React Mistake
   Duration: 30s

[... continue for all shorts ...]

**Recommended Hashtags:**
#react #coding #webdev #programming #tutorial

**Platform-Specific Notes:**
- YouTube Shorts: Upload as-is, max 60s
- TikTok: Shorts 1, 4, 5 have strong hooks
- Instagram Reels: All shorts suitable

Output: All shorts saved locally and uploaded to Google Drive with shareable links

Complete Workflow Example

User Request:

Create shorts from https://youtube.com/watch?v=xyz789 Focus on beginner-friendly segments Use yellow subtitles with black background

Agent Execution:

✓ Task 1: Downloading Video

Source: "React Hooks Complete Guide" (22:15)
Resolution: 1920x1080
Downloaded: react_hooks_guide.mp4 (145 MB)

✓ Task 2: Transcribing Audio

Using: OpenAI Whisper
Processing audio... [Progress: 100%]
Generated: full_transcript.srt (485 lines)
Duration: 22:15

✓ Task 3: AI Segment Analysis (Gemini)

Analyzing transcript for viral segments...

Gemini Recommendations:

🔥 Segment 1: "useState Explained in 30 Seconds" (28s)
   Time: 02:15 - 02:43
   Hook: "Here's what most tutorials don't tell you about useState"
   Viral Score: 9/10
   Reason: Quick, actionable, addresses common confusion

⭐ Segment 2: "Fix This Common Hook Mistake" (32s)
   Time: 07:45 - 08:17
   Hook: "This mistake crashes your app"
   Viral Score: 10/10
   Reason: Problem-solution format, relatable error

⭐ Segment 3: "useEffect Dependency Array Trick" (25s)
   Time: 12:30 - 12:55
   Hook: "One simple trick saves hours of debugging"
   Viral Score: 8/10
   Reason: Specific tip, time-saving

✓ Segment 4: "Custom Hooks in 40 Seconds" (38s)
   Time: 16:20 - 16:58
   Hook: "Custom hooks are easier than you think"
   Viral Score: 7/10
   Reason: Beginner-friendly, empowering

✓ Segment 5: "React Hook Rules Explained" (35s)
   Time: 19:10 - 19:45
   Hook: "Break these rules and React breaks"
   Viral Score: 8/10
   Reason: Clear rules, consequences shown

Recommended: Create Segments 1, 2, 3, 5 (4 shorts total)

User confirms: "Create all 5"

✓ Task 4: Subtitle Configuration

Applying custom settings:
- Font: Arial Bold
- Size: 52px (larger for readability)
- Color: Yellow (#FFFF00)
- Background: Semi-transparent black box
- Outline: None (background provides contrast)
- Position: Bottom center, 120px from bottom

✓ Task 5: Generating Shorts

Creating 5 vertical shorts with subtitles...

[1/5] "useState Explained in 30 Seconds"
  ✓ Extracted segment (02:15 - 02:43)
  ✓ Converted to 9:16 (1080x1920)
  ✓ Burned in yellow subtitles
  ✓ Added fade transitions
  ✓ Optimized: react_hooks_short_01_9x16.mp4 (12.3 MB)

[2/5] "Fix This Common Hook Mistake"
  ✓ Extracted segment (07:45 - 08:17)
  ✓ Converted to 9:16 (1080x1920)
  ✓ Burned in yellow subtitles
  ✓ Added fade transitions
  ✓ Optimized: react_hooks_short_02_9x16.mp4 (13.8 MB)

[3/5] "useEffect Dependency Array Trick"
  ✓ Extracted segment (12:30 - 12:55)
  ✓ Converted to 9:16 (1080x1920)
  ✓ Burned in yellow subtitles
  ✓ Added fade transitions
  ✓ Optimized: react_hooks_short_03_9x16.mp4 (10.5 MB)

[4/5] "Custom Hooks in 40 Seconds"
  ✓ Extracted segment (16:20 - 16:58)
  ✓ Converted to 9:16 (1080x1920)
  ✓ Burned in yellow subtitles
  ✓ Added fade transitions
  ✓ Optimized: react_hooks_short_04_9x16.mp4 (16.2 MB)

[5/5] "React Hook Rules Explained"
  ✓ Extracted segment (19:10 - 19:45)
  ✓ Converted to 9:16 (1080x1920)
  ✓ Burned in yellow subtitles
  ✓ Added fade transitions
  ✓ Optimized: react_hooks_short_05_9x16.mp4 (14.7 MB)

All shorts generated successfully!
Total: 5 videos, 67.5 MB

✓ Task 6: Uploading to Google Drive

Uploading to: Video Projects/Shorts/React Hooks Guide/

[Upload Progress]
✓ react_hooks_short_01_9x16.mp4
✓ react_hooks_short_02_9x16.mp4
✓ react_hooks_short_03_9x16.mp4
✓ react_hooks_short_04_9x16.mp4
✓ react_hooks_short_05_9x16.mp4
✓ shorts_metadata.json

All files uploaded successfully!

🎉 Shorts Creation Complete!

📁 Google Drive Links:

  1. useState Explained (28s)
  2. Fix Common Mistake (32s)
  3. useEffect Trick (25s)
  4. Custom Hooks (38s)
  5. Hook Rules (35s)

📊 Summary:

  • Source: 22-minute tutorial → 5 shorts (2.7 minutes total)
  • Format: 1080x1920 (9:16 vertical)
  • Subtitles: Yellow text, black background
  • Total size: 67.5 MB
  • Upload location: Video Projects/Shorts/

🎯 Publishing Recommendations:

  • High Priority (Post first): Shorts #2, #1, #3
  • TikTok Focus: Shorts #2, #5 (strong problem hooks)
  • YouTube Shorts: All suitable
  • Instagram Reels: All suitable

#️⃣ Suggested Hashtags: #react #reacthooks #javascript #webdev #coding #programming #tutorial #learntocode

Advanced Features

  • Batch processing: Process multiple source videos
  • Custom subtitle animations: Word-by-word highlighting, karaoke style
  • Brand overlay: Add logo/watermark to all shorts
  • Intro/outro templates: Consistent branding across shorts
  • Multi-language subtitles: Generate shorts in multiple languages
  • Platform-specific exports: Different aspect ratios per platform

Tips for Success

  1. Source quality: Start with 1080p or higher source video
  2. Segment length: 15-45 seconds performs best (max 60s)
  3. Hook priority: First 2-3 seconds determine virality
  4. Subtitle readability: Test on mobile device before publishing
  5. Audio clarity: Ensure clean audio for accurate transcription
  6. Gemini prompting: Customize analysis prompt for your niche
  7. Batch processing: Process all shorts at once for efficiency

Troubleshooting

Download Issues

  • YouTube URL not working: Check video availability, try different format
  • Drive link fails: Verify sharing permissions set to "Anyone with link"

Transcription Errors

  • Poor accuracy: Improve source audio quality, use noise reduction
  • Missing timestamps: Verify Whisper word-level timestamps enabled
  • Wrong language: Specify language parameter in Whisper

Subtitle Rendering Issues

  • Text cut off: Adjust y-offset, reduce font size
  • Outline too thick: Reduce outline width parameter
  • Wrong timing: Verify SRT timestamps match video timing

Performance Issues

  • Slow processing: Process shorts sequentially instead of parallel
  • Large file sizes: Increase CRF value (lower quality, smaller size)
  • Memory issues: Process one short at a time, clear temp files

Integration with Other Skills

  • script-generation: Generate scripts for original shorts (not from existing video)
  • video-generation: Create original short-form content with Veo
  • heygen-avatar: Create avatar-narrated vertical shorts
  • upload-to-drive: Automatic upload integration (already included)

Related Skills

  • video-generation: For creating original short content
  • heygen-avatar: For avatar-based vertical videos
  • upload-to-drive: For organized cloud storage
  • script-generation: For planning original shorts

发现更多相关技能

查看全部技能
我的收藏