eternityspring/reelbench-skills

What it solves

This project provides specialized tools for AI-assisted video analysis and processing. It helps users break down films into shot-by-shot data, synchronize video footage with descriptive metadata, and scrub sensitive or unnecessary technical metadata from video files while maintaining visual integrity.

How it works

The project functions as a collection of modular "skills" designed to be integrated into AI coding assistants like Claude Code or Codex. These tools leverage ffmpeg and ffprobe for low-level video processing and data extraction. The skills are self-contained, requiring no external API keys or complex dependencies, and they use automated quality gates to verify the accuracy of the AI-generated analysis.

Who it’s for

It is designed for video editors, AI content creators, and researchers who need to perform technical analysis on video files or automate the documentation of film structures using AI-powered workflows.

Highlights

• Shot Breakdown: Automatically segments video into shots, capturing duration, camera movement, and shot size with 15 quality verification gates.
• Interactive Reporting: Generates a single, offline-ready HTML report that allows users to click through shots, filter data, and view frame-by-frame analysis.
• Metadata Scrubbing: Deep-cleans video files by removing hidden telemetry and encoder metadata that standard tools often miss, verified by byte-level checks.
• Seamless Integration: Installs via simple symlinks to work directly within existing AI coding environments like Claude Code or Codex.