Video Clipper Skill

Create video clips from a source video using ffmpeg. Supports two modes:

Single snippet: Extract one clip from a time range (e.g., "55s to 2m35s")
Multi-chapter: Split video into multiple clips based on chapter timestamps

Quick Start

Single Snippet Mode

For extracting one clip from a time range:

Say: "Create a clip from 55s to 2m35s" (or any time range)
Provide video file path when prompted
Choose encoding mode (copy or encode)
Done! Your clip is created

Example time formats:

55s to 2m35s (natural language)
1:30 to 5:45 (MM:SS)
00:01:30 to 00:05:45 (HH:MM:SS)

Multi-Chapter Mode

For splitting into multiple clips:

Have your video file ready
- Any common format: MP4, MOV, AVI, MKV, etc.
Prepare chapter timestamps
- Format: HH:MM:SS - Chapter Name or MM:SS - Chapter Name
- Example:
```
00:00:00 - Introduction
00:05:30 - Main Content
00:15:45 - Q&A Session
```
Invoke the skill
- Say: "Split this video by chapters"
- Or: "Use video-clipper skill"
Provide information when prompted
- Video file path
- Chapter timestamps
- Output directory (optional)
- Encoding mode: "copy" (fast) or "encode" (precise)

When to Use

Use this skill when you need to:

Create a quick snippet from a time range
Extract specific segments from a video
Split a long video into topic-based clips
Create individual clips from chapter markers
Process videos after running Video Transcript Analyzer

Requirements

ffmpeg must be installed:

macOS: brew install ffmpeg
Ubuntu/Debian: sudo apt-get install ffmpeg
RHEL/CentOS: sudo yum install ffmpeg
Windows: Download from ffmpeg.org

Encoding Modes

Copy Mode (Fast)

Speed: Very fast (10-100x real-time)
Quality: No quality loss
Precision: May not cut at exact frame (keyframe limitations)
Best for: Large files, quick previews, when timing not critical

Encode Mode (Precise)

Speed: Slower (0.5-2x real-time)
Quality: Slight quality loss (adjustable)
Precision: Frame-accurate cuts
Best for: Professional work, precise timing, format conversion

Example Usage

Example 1: Basic Usage

User: "I need to split my customer interview video by topics"

Claude: "I'll help you split the video. First, is ffmpeg installed?"
→ Checks ffmpeg installation

Claude: "What's the path to your video file?"
User: "/Users/alice/videos/interview.mp4"

Claude: "Please provide chapter timestamps in this format:
        00:00:00 - Chapter Name
        00:05:30 - Next Chapter"

User: "00:00:00 - Introduction
       00:03:15 - Product Feedback
       00:12:30 - Feature Requests
       00:18:45 - Closing"

Claude: "Where would you like clips saved? (default: same folder as video)"
User: [Enter]

Claude: "Encoding mode? (copy = fast, encode = precise)"
User: "copy"

Claude: "Creating 4 clips from interview.mp4..."
→ Generates and executes ffmpeg commands
→ Creates clips:
  01-Introduction.mp4
  02-Product_Feedback.mp4
  03-Feature_Requests.mp4
  04-Closing.mp4

Claude: "✓ Successfully created 4 clips (total: 245 MB)
         Location: /Users/alice/videos/interview_clips/"

Example 2: After Video Transcript Analyzer

User: "I just analyzed a transcript and got chapter markers. Now create clips from those."

Claude: "Great! I can use those chapter markers. What's the video file path?"
User: "/Users/bob/recording.mp4"

Claude: "I see these chapters from your analysis:
        00:02:57 - Memberships & Registration
        00:09:41 - Lists vs Views
        00:15:30 - API Limitations

        Your first chapter starts at 00:02:57. Include 00:00:00 - 00:02:57 as opening clip?"
User: "Yes"

Claude: "Encoding mode?"
User: "encode"

Claude: "Creating 4 clips with frame-accurate cuts..."
→ Creates clips including opening segment

Claude: "✓ Complete! 4 clips created with precise timing."

Output

The skill creates:

Individual video clips
- Numbered sequentially: 01-Chapter_Name.mp4
- Spaces replaced with underscores
- Special characters removed
- Saved in output directory
Summary report
- List of created clips with durations and sizes
- Total size of all clips
- Commands used (for reference)
- Any warnings or errors

Tips

Getting Good Results

Start at 00:00:00
- Always include the video beginning as first chapter
- Avoids confusion about opening content
Use copy mode first
- Much faster for testing
- Can re-run with encode mode if cuts aren't precise
Verify timestamps
- Watch video and note exact transition points
- Use video player's timestamp feature
Descriptive chapter names
- Make it clear what's in each clip
- Avoid generic names like "Part 1"

Workflow Integration

Combined with Video Transcript Analyzer:

Run Video Transcript Analyzer on your transcript
Get chapter markers with timestamps
Use Video Clipper to create clips from those markers
Result: Organized video library with transcripts

Choosing Encoding Mode

Use Copy Mode when:

Video is large (>500 MB)
You need quick results
Exact frame precision not critical
Keeping original quality important

Use Encode Mode when:

You need frame-accurate cuts
Converting formats
Normalizing quality across clips
Professional/final output

Troubleshooting

Clips don't start at exact timestamp

Cause: Copy mode cuts at keyframes, not exact frames Solution: Use encode mode for frame-accurate cuts

"Non-monotonous DTS" error

Cause: Timestamp issue in copy mode Solution: Skill automatically adds -avoid_negative_ts 1 flag

Audio/video out of sync

Cause: Variable frame rate source video Solution: Use encode mode with fixed frame rate

Processing very slow

Cause: Using encode mode on large video Solutions:

Use copy mode (much faster)
Use faster preset: -preset fast
Enable hardware acceleration (if available)

ffmpeg not found

Solution: Install ffmpeg (see Requirements section)

File Organization

After processing, your files will look like:

videos/
├── my-video.mp4                           (original)
└── my-video_clips/                        (created)
    ├── 01-Introduction.mp4
    ├── 02-Main_Content.mp4
    ├── 03-Demo.mp4
    └── 04-Conclusion.mp4

Performance

Typical processing times:

Video Size	Copy Mode	Encode Mode
100 MB	~10 sec	~5 min
500 MB	~30 sec	~20 min
1 GB	~1 min	~40 min
5 GB	~3 min	~3 hours

Times vary by system specs, resolution, and codec

Supported Formats

Input Formats

MP4, MOV, AVI, MKV, WebM
FLV, WMV, MPEG, TS, M4V
Most common video formats

Output Format

MP4 (H.264 video, AAC audio)
Most widely compatible format
Works on all platforms and devices

Advanced Usage

Custom Output Directory

User: "Save clips to /Users/alice/Desktop/clips"

Custom Filename Prefix

User: "Use 'segment' as filename prefix instead of numbers"
→ Creates: segment-Introduction.mp4, segment-Main_Content.mp4

Quality Settings (Encode Mode)

User: "Use high quality encoding (CRF 18)"
→ Skill adjusts ffmpeg command: -crf 18 (higher quality, larger files)

User: "Use smaller file sizes (CRF 28)"
→ Skill adjusts ffmpeg command: -crf 28 (lower quality, smaller files)

Examples Directory

See resources/EXAMPLES.md for complete examples with:

Full input/output for various scenarios
Error cases and solutions
Complex filename handling
Large video processing

Technical Reference

See resources/REFERENCE.md for:

Detailed ffmpeg command reference
Timestamp parsing logic
Filename sanitization rules
Troubleshooting guide
Performance optimization tips
Codec and format information

Support

Common Questions

Q: Can I split a video without chapter markers? A: Yes! Just provide timestamps with generic names:

00:00:00 - Part 1
00:10:00 - Part 2
00:20:00 - Part 3

Q: What if I only have a few timestamps? A: That works! You can split a video at any number of points (minimum 2).

Q: Can I extract just one segment? A: Yes! Provide just two timestamps (start and end of desired segment).

Q: Does this modify my original video? A: No, the original video is never modified. Clips are saved separately.

Q: Can I process multiple videos at once? A: Currently processes one video at a time, but you can run multiple times with different videos.

Q: What if my timestamps are slightly off? A: You can adjust timestamps and re-run. Copy mode is fast for iteration.

Privacy & Security

All processing done locally on your machine
No video content sent to external services
Original video never modified or deleted
Output clips saved only to specified directory
No logging of video content or filenames

Version History

1.0.0 (2025-11-11): Initial release
- Support for MP4, MOV, AVI, MKV formats
- Copy and encode modes
- Automatic filename sanitization
- Error handling and validation
- Integration with Video Transcript Analyzer

License

This skill uses ffmpeg, which is licensed under LGPL/GPL depending on build configuration.

Ready to split your videos?

In Claude Code, say: "Use video-clipper skill" or "Split my video by chapters"