Who it is forAn editor who wants to see each candidate's motion and source before final clips are cut.
The problemApproving clips from a text list is guesswork, and delivering footage without recording its usage terms is risky.
The outcomeDecisions imported from the storyboard, usage conditions recorded and verified clips delivered with their sources.
What you need first
- Python 3.11+
- FFmpeg and ffprobe built with libfreetype (drawtext)
- Node 22+ with npm and npx
- curl and Git
- A project folder where the B-roll files will be kept
Steps
- Build the storyboard with
python3 scripts/gb.py review --project /caminho/meu-video - Serve it with
python3 scripts/gb.py serve --background --project /caminho/meu-videoand open the link - Approve or reject each candidate after checking the GIF previews and frame sequences
- Import decisions with
python3 scripts/gb.py import-review --by "Seu nome" --project /caminho/meu-video - Run
python3 scripts/gb.py verify --project /caminho/meu-videoand stop the server withpython3 scripts/gb.py serve --stop --project /caminho/meu-video
Prompt for your agent
Using engenheirodevideo/get-brolls, build and serve the storyboard for <project folder>, wait for my decisions, import them, verify every clip and list any clip whose usage conditions are still missing.
Caveat. Record each source's real usage conditions with permit before fetching final clips; previews are working media only.
Use this in your AI agent
Ask your agent
Using the Deduped MCP, get the setup steps for engenheirodevideo/get-brolls and "Approve or reject B-roll picks in a local storyboard"
Add Deduped to Claude Code once
claude mcp add --transport http deduped https://dedupedai.com/mcp
Every Monday: the 5 hottest repos and exactly how to use them. Free, one email a week, unsubscribe anytime.
Get them freeShare: X LinkedIn Hacker News Email Copy link
Third-party repos we don't own. AI-written use cases can be wrong. Disclaimer