bsisduck/video-analyzer-skill

6 stars · Last commit 2026-05-04

Video analysis skill for Claude Code — frame extraction, scene detection, transcription, parallel subagent analysis

README preview

# Video Analyzer Skill for Claude Code

A video analysis skill for [Claude Code](https://claude.ai/claude-code) that extracts frames, detects scene changes, transcribes audio, and dispatches parallel subagents for visual and audio analysis.

Follows the [Agent Skills](https://agentskills.io) open standard.

## Features

- **Automatic tier selection** — adapts frame rate based on video duration
- **Custom frame rate & time range** — override defaults with "at 5fps" or "from 0:10 to 0:30"
- **Smart stream detection** — skips attached thumbnails common in Instagram/TikTok downloads
- **Portrait-aware grids** — 3x5 layout for portrait, 4x4 for landscape
- **Scene change detection** — auto-detects visual transitions, extracts high-res key frames
- **Parallel subagent analysis** — dispatches grid + key frame + audio agents simultaneously
- **Three transcription modes** — automatic (Whisper), user-provided, or visual-only
- **Structured output** — produces a Markdown analysis document with timeline and scene breakdown

## Analysis Tiers

| Tier | Duration | Frame Rate | Focus |

View full repository on GitHub →