rea-video
Breaks a reference video into shots, overlays and sound for an agent to study
What it is
A Python command-line tool with an agent skill that takes a video apart: it cuts it into shots, labels how each shot was probably made, measures captions and other overlays, and splits the audio into voice, music and effects. The result is a layered Markdown pack plus an HTML report that an agent can read cheaply. For people who study how a video is built before planning their own.
Good for
- Get a shot-by-shot timeline of a reference video
- Measure caption timing and overlay style
- Separate voice, music and sound effects
- Draft a production plan from a reference
What you need
Python 3.10+, ffmpeg; downloads model files on first run; optional vision-model API key
Sources
- https://github.com/JuneYaooo/rea-video (project README, read 13 Nov 2026)
- https://api.github.com/repos/JuneYaooo/rea-video (GitHub API, read 13 Nov 2026)
The summary is ours, written from the README. Spotted an error? See how we work.