gitmyhub

CutClaw

Python ★ 937 updated 3mo ago

Agentic Hours-Long Video Editing via Music Synchronization

An AI tool that automatically edits raw video footage and music into a finished montage from a plain text instruction.

PythonLiteLLMStreamlitsetup: moderatecomplexity 3/5

CutClaw is an AI-powered video editing tool that takes hours of raw footage and a music track and automatically assembles them into a polished montage. Instead of manually scrubbing through footage and cutting clips to match the beat, you give CutClaw your video files and an instruction in plain text — something like "fast-paced character montage" or "slow emotional narrative" — and it handles the rest.

The system works through a multi-agent pipeline, which means several AI modules work together in stages: first, it analyzes and "deconstructs" all the raw video and audio into labeled, searchable clips; next, a planning agent decides which shots to use and in what order; then another agent picks the exact timestamps for each cut and validates the result before rendering the final video. Music synchronization is central to how cuts are timed — the tool extracts the beats and energy patterns from the audio track and uses them to decide where cuts should land. There is also a smart auto-cropping feature that detects the main subject in each clip and adjusts the frame to fit different aspect ratios (useful for social media formats).

You run it either through a web interface (a Streamlit app in your browser) or via command line. It is built in Python and powered by LiteLLM (a layer that lets you connect different AI language models). The full README is longer than what was provided.

Where it fits