Video

Skill naveedharri/benai-skills/plugins/all-skills/skills/video

Install

npx -y skills add naveedharri/benai-skills --skill video

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

What its author says it does

Copied from the file, not written here

Edit, process, and render videos using FFmpeg and Remotion. Handles stitching, transitions, captions, teasers, and transcription. USE THIS SKILL WHEN user mentions video, clip, footage, mp4, mov, avi, mkv, or any video format, wants to combine, stitch, merge, or join video files, asks about transitions, fades, crossfades, or effects between clips, needs captions, subtitles, or TikTok-style word highlighting, wants a teaser, trailer, or highlight reel from longer video, mentions FFmpeg, ffprobe, Remotion, or related errors, asks about video transcription or whisper speech-to-text, asks about video duration, resolution, fps, or trimming, has .mp4, .mov, .captions.json files or video folders, or asks about thumbnails, overlays, title cards, or graphics for video.

SKILL.md

6.4 KB, as published. Nobody here has run it

Video Editing Skill

Edits, processes, and renders videos using FFmpeg and Remotion.

Intelligent Editing Workflow

This skill enables Claude to act as a smart video editor. Before executing commands:

Analyze - Examine videos (duration, resolution, audio) using ffprobe
Transcribe - Get speech content for smart cuts (if needed)
Ask - Clarify user intent and preferences
Plan - Propose edit approach for approval
QA Test - Run automated tests before user preview (see references/qa-testing.md)
Preview - Only show to user after QA passes
Iterate - Refine based on feedback

See references/intelligent-editing.md for detailed workflow, questions to ask, and quality checks.

Tool Selection

Task	FFmpeg	Remotion
Stitching	Same codec, no effects	Transitions, overlays, programmatic
Transitions	Simple crossfades	Multiple types, custom timing
Captions	SRT burn-in	TikTok-style word highlighting
Teasers	Quick clips	Text overlays, branded elements

General guidance:

FFmpeg: Fast CLI operations, batch processing, format conversion
Remotion: Styled content, animations, preview before render, React components

Reference Files

Read these files as needed using the Read tool:

Intelligence Layer

File	Contents
references/intelligent-editing.md	Smart editing workflow, questions, quality checks
references/video-analysis.md	Analyze videos with ffprobe
references/transcription.md	Whisper transcription for speech analysis
references/qa-testing.md	QA tests before user preview

Core Editing

File	Contents
references/stitching.md	Combine multiple clips into one video
references/transitions.md	Add fade, slide, wipe between clips
references/captions.md	Add subtitles and TikTok-style captions
references/teasers.md	Create 30-second teasers/trailers
references/title-cards.md	Add chapter headers, bumpers, section titles
references/graphics-generation.md	Generate thumbnails, overlays, social graphics

Technical Reference

File	Contents
references/ffmpeg-basics.md	Common FFmpeg patterns and troubleshooting
references/remotion-setup.md	Remotion project setup and workflow
references/remotion-tips.md	Animations, timing, springs, sequences, and captions
references/reference-implementation.md	Complete working code to recreate the project

Quick Start

Transcription (Fast)

# 1. Setup whisper.cpp (one-time)
npx ts-node scripts/setup-whisper.ts

# 2. Transcribe all videos (GPU accelerated)
npx ts-node scripts/transcribe-fast.ts

FFmpeg (CLI)

# Check installation
which ffmpeg || brew install ffmpeg

# Basic concat (same codec)
ffmpeg -f concat -safe 0 -i list.txt -c copy output.mp4

# Burn in subtitles
ffmpeg -i video.mp4 -vf "subtitles=captions.srt" output.mp4

Remotion (React)

# Start preview (always do this first)
npm run dev

# Render (only when user requests)
npx remotion render CompositionName out/video.mp4

Key Lessons Learned

Transcription (IMPORTANT)

Use @remotion/install-whisper-cpp - NOT Python whisper (10x faster with GPU)
Use base.en model for speed, medium.en for final quality
Required parameter: whisperCppVersion: '1.5.5' in transcribe() call
Enable tokenLevelTimestamps: true for word-level captions
Whisper.cpp uses Metal GPU on Mac - 8.5min video in ~85 seconds

Teasers

Never start clips with: "But", "And", "So", "Now", "It", "This", "That", "They"
Each clip must have an explicit subject - not pronouns or references
Aim for 3-4 clips totaling 25-30 seconds
Playback speed: 1.3x for energy (1.5x is too fast)
Transitions: Smooth fade (0.4s) between clips
First clip: Must NOT overlap with intro start (avoid repetition)

Captions

Active word: Yellow #FFD700
Spoken words: White #FFFFFF
Upcoming words: Light gray #E0E0E0
Font: Roboto at 48px for 1080p

Black stroke outline (works on ANY background):

WebkitTextStroke: '3px black',
paintOrder: 'stroke fill',
textShadow: '0px 0px 4px rgba(0,0,0,1), 0px 0px 8px rgba(0,0,0,0.8)'

Remotion Video Trimming

Use OffthreadVideo component
Set BOTH startFrom AND endAt props
Convert ms to frames: Math.round((ms / 1000) * fps)

Edge Cases & Gotchas

File Naming

Avoid spaces in video filenames: use intro.mp4 not Intro Video.mp4
Remotion Folder names can only contain letters, numbers, and hyphens

Ports

Remotion Studio may start on port 3001 if 3000 is in use

Codec Mismatch

FFmpeg concat demuxer (-c copy) fails silently with mismatched codecs
Use ffprobe to check codecs before concatenating
Use filter_complex for different codecs/resolutions

Workflow

Always preview in Remotion Studio before rendering
Never render automatically - wait for user approval
Copy videos to public/ folder for Remotion projects