Social Media Clip Guru

Your agent for creating social media clips from longer content — extracting highlights, formatting for platform requirements, adding captions, and optimizing for engagement.

When to Use This Agent

Choose Social Media Clip Guru when:

• Extracting highlights from long-form content (podcasts, webinars, interviews)
• Formatting video/audio clips for platform-specific requirements (TikTok, Reels, Shorts)
• Adding captions, subtitles, and text overlays to clips
• Creating audiograms (waveform videos) from podcast highlights
• Batch-producing social media content from a single source

Consider alternatives when:

• You need full video editing — use a video processing agent
• You need social media strategy — use a social media content agent
• You need audio editing without video — use an Audio Mixer agent

Quick Start

Copy
# .claude/agents/social-media-clip.yml
name: Social Media Clip Guru
model: claude-sonnet
tools:
  - Read
  - Write
  - Edit
  - Bash
  - Glob
  - Grep
description: Social media clip creation agent for extracting highlights, adding captions, and formatting for platforms

Example invocation:

claude "Extract 5 highlight clips (30-60 seconds each) from our 1-hour podcast, add burned-in captions, format for Instagram Reels (9:16), and create an audiogram version for Twitter"

Core Concepts

Platform Specifications

Platform	Aspect Ratio	Max Duration	Format
TikTok	9:16	10 min	MP4 H.264
Instagram Reels	9:16	90 sec	MP4 H.264
YouTube Shorts	9:16	60 sec	MP4 H.264
Twitter/X	16:9 or 1:1	2:20 min	MP4 H.264
LinkedIn	1:1 or 16:9	10 min	MP4 H.264

Clip Creation Pipeline

Copy
# Extract clip from source video
ffmpeg -i source.mp4 -ss 00:15:30 -to 00:16:15 \
  -c copy clip_raw.mp4

# Resize for vertical (9:16) with padding
ffmpeg -i clip_raw.mp4 \
  -vf "scale=1080:1920:force_original_aspect_ratio=decrease,
       pad=1080:1920:(ow-iw)/2:(oh-ih)/2" \
  clip_vertical.mp4

# Add burned-in subtitles
ffmpeg -i clip_vertical.mp4 \
  -vf "subtitles=captions.srt:force_style='FontSize=24,PrimaryColour=&HFFFFFF'" \
  clip_final.mp4

Configuration

Parameter	Description	Default
platform	Target platform (tiktok, reels, shorts, twitter, linkedin)	reels
aspect_ratio	Output aspect ratio	9:16
max_duration	Maximum clip length in seconds	60
captions	Add burned-in captions	true
audiogram	Generate waveform audiogram	false

Best Practices

1. Start clips with a hook in the first 3 seconds. Social media users decide to keep watching within 3 seconds. Begin clips with the most compelling statement, question, or moment — not with context or introductions.

2. Add burned-in captions for all clips. 85% of social media videos are watched without sound. Burned-in captions (hardcoded into the video) ensure readability across all platforms, unlike platform-generated captions that may be inaccurate.

3. Format for vertical (9:16) as the primary format. TikTok, Instagram Reels, and YouTube Shorts all use 9:16. Create the vertical version first, then crop or letterbox for landscape platforms (LinkedIn, Twitter).

4. Keep clips under 60 seconds for maximum engagement. Shorter clips have higher completion rates, which boost algorithmic distribution. If a highlight runs 90 seconds, edit it down to the essential 45-60 seconds.

5. Create multiple clips per source for A/B testing. Extract 5-10 potential highlights and publish the best 3. Test different hooks, lengths, and formats to learn what your audience engages with.

Common Issues

Caption timing is off when clips are extracted from longer sources. When cutting a clip from a longer video, subtitle timestamps need to be adjusted to start from zero. Use FFmpeg's -ss flag with subtitle filters, or re-time the SRT file to match the clip's new timeline.

Video quality degrades after re-encoding for multiple platforms. Each re-encode loses quality. Encode from the original source at the highest quality, then create platform-specific versions from that master. Use -crf 18 for near-lossless quality.

Vertical crop cuts off important visual content. Automatic center-cropping from 16:9 to 9:16 may miss speakers or important visuals. Use content-aware cropping or manually set crop points. For talking-head content, face detection can automate the crop position.