DEDUPEDWho it is forA creator iterating on prompts on a laptop before running them on a GPU box.
The problemBad prompts waste long GPU renders.
The outcomeA validated plan for the prompt with no GPU spent.
What you need first
- Linux with an NVIDIA GPU (WSL2 on Windows)
- About 54 GB of disk for H3 weights
- 12 GiB or more VRAM for the supported INT8 tiers
Steps
- Clone the repo and install with
pip install -e . - Run
open-video "sunset waves" --dry-run - Adjust the prompt until it validates
- Run the final prompt on the GPU machine
Prompt for your agent
With agent-next/video-agent, dry-run <prompt> using `--dry-run` and report any validation errors. Confirm no GPU job was started.
Caveat. On macOS the clone supports setup and dry-run, but H3 generation there is not the default path.
Use this in your AI agent
Ask your agent
Using the Deduped MCP, get the setup steps for agent-next/video-agent and "Plan and validate a video prompt without spending GPU time"
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