video-understand

Analyze local videos with ffmpeg frame extraction and Whisper transcription.

602|12|Updated Apr 22, 2026
One-click install
npx skills add https://github.com/video-production-buddy/video-production-buddy --skill video-understand-video-production-buddy
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: video-understand
Source: https://github.com/video-production-buddy/video-production-buddy/tree/main/.agents/local/skills/video-understand
Command: npx skills add https://github.com/video-production-buddy/video-production-buddy --skill video-understand-video-production-buddy

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires ffmpeg, ffprobe, openai-whisper, and includes scripts (resource) and references (resource) components.

What problem does it solve?

Analyzing video content traditionally requires paid cloud APIs or manual watching to extract visual frames and audio transcripts. This Skill eliminates that cost and effort by enabling full offline video analysis with no API keys required.

Core Features & Use Cases

  • Local Frame Extraction: Extract key frames via scene change detection, I-frame keyframe extraction, or regular interval sampling using ffmpeg, with automatic fallback to interval mode if no scene changes are detected.
  • Offline Audio Transcription: Transcribe video audio locally using Whisper (optional, installable via pip) without sending any data to cloud services.
  • Structured Output: Returns a JSON object with frame paths, timestamps, transcript segments, and full video metadata for easy downstream analysis. For example, a content creator can use this Skill to quickly extract key moments and a full transcript from a 30-minute interview video to create social media clips without watching the entire recording.

Quick Start

Use the video-understand skill to analyze the video file 'client-interview.mp4' and get a full transcript plus key frames of all scene changes.

Frequently Asked Questions about video-understand

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I extract key frames and transcribe audio from a video locally without API keys?

Local video analysis uses ffmpeg for frame extraction and openai-whisper for audio transcription, outputting structured JSON with frame paths, timestamps, and transcript segments without requiring cloud API keys.

What is the best way to detect scene changes in a video for frame extraction?

Scene change detection uses ffmpeg to extract key frames during transitions, automatically falling back to regular interval sampling mode if no scene changes are detected in the video.

Can I transcribe video audio offline using Whisper without sending data to cloud services?

Yes, offline audio transcription uses openai-whisper locally to process video audio without sending any data to external cloud services or requiring paid API access.

Do I need ffmpeg and openai-whisper installed to use local video analysis?

ffmpeg is required for frame extraction and metadata processing, while openai-whisper is an optional dependency needed only if you want to generate local audio transcripts from the video.

What structured output format do I get from local video content analysis?

Local video content analysis outputs a structured JSON object containing extracted frame paths, timestamps, transcript segments, and full video metadata for downstream processing.