VideoClaw

HITsz-TMG/VideoClaw

VideoClaw is an open-source app project by HITsz-TMG on GitHub: VideoClaw from Harbin Institute of Technology (Shenzhen): an AI video coworker that turns a chatted idea into a film, covering the full short-drama workflow. It covers script, storyboard and production. 1.9k stars, last updated 2026-08-26, MIT license.

Editor's notes

It runs a local FastAPI backend and Next.js frontend (Python 3.9+, Node.js 18+); the skill is written for OpenClaw and can be installed through ClawHub. All models are cloud APIs, so you need keys for whichever platforms your chosen models use, such as DashScope, Volcengine Ark or Kling. The workflow has 7 confirmation stops where the script, character and scene images, storyboard and video clips go to the user for approval, and the post-production stage stitches the final cut automatically. Video defaults to first-frame-to-video, with first-and-last-frame or reference-image modes as alternatives; it also bundles non-drama flows such as motion transfer and talking-head avatars.

Type
Related open-source platforms (full apps, not skills)
Stages
Script · Storyboard · Production
Format
Full application
Language
Multilingual
GitHub stars
★ 1.9k
Last updated
2026-08-26
License
MIT
Checked by hand
2026-10-13

In the author's words

🚀 A fully automated AI video-generation employee | Your First AIGC Coworker. Chat an Idea. Get a Film. 🦞

Translated from the Chinese original.

Similar projects