skills.rest
HITsz-TMG
HITsz-TMGOFFICIAL
Explore all skills by this author →
WebsiteGitHub

video-claw

Generates complete videos from ideas through a six-stage AI production workflow.

1.8k|270|Updated Aug 29, 2024
Search Tags:#video generation#storyboard#text-to-video#aigc#ai video#digital human#action transfer
One-click install
npx skills add https://github.com/HITsz-TMG/VideoClaw --skill video-claw-hitsz-tmg
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: video-claw
Source: https://github.com/HITsz-TMG/VideoClaw/tree/main/video-claw
Command: npx skills add https://github.com/HITsz-TMG/VideoClaw --skill video-claw-hitsz-tmg

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Turning a raw idea into a finished video normally requires coordinating scriptwriting, character design, storyboarding, image generation, video generation, and editing across many disconnected tools. This Skill orchestrates that entire pipeline through a locally deployed FastAPI backend and Next.js frontend, pausing at each stage for user confirmation. ## Core Features & Use Cases - Six-Stage Video Workflow: Runs script generation, character/scene design, storyboard planning, reference image generation, video clip generation, and post-production editing with mandatory user confirmation stop points between stages. - One-Shot Pipelines: Executes background pipelines for artistic short videos, action transfer, and digital human narration without manual intervention. - Sandbox Workbench: Provides single-shot tools for text-to-image, image-to-image style transfer, and short video generation via local API calls. - Use Case: A creator says "make a 4-episode short drama about a cat detective" — the agent starts the local services, creates the project, generates the script, characters, storyboards, reference images, and video clips, confirms each stage with the user, and delivers the final edited episodes. ## Quick Start Ask the agent to generate a video from your story idea, and it will start the local backend and frontend, confirm your plot and model settings, then walk you through each production stage.

Frequently Asked Questions about video-claw

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I generate a video from a text idea with VideoClaw?▼

Start the local backend on port 8000 and frontend on port 3000, then describe your story idea. The agent creates a project and runs six stages — script, characters, storyboard, reference images, video clips, and editing — pausing for your confirmation at each stage.

What AI models does this video generation workflow support?▼

It supports LLM providers like DeepSeek, OpenAI, Gemini, and DashScope, image models like Seedream and Wan, and video models including Wan, Kling, and Jimeng. Models are selected in backend/config.yaml or the frontend settings page.

Can I generate short videos without the full six-stage workflow?▼

Yes. Three one-shot pipelines run in the background without manual stops: artistic short video, action transfer, and digital human narration. A sandbox workbench also handles single text-to-image, image-to-image, and short video generation calls.

Why does the backend fail to start or return connection errors?▼

Connection errors usually mean the FastAPI backend is not running or Python dependencies are missing. Activate the virtual environment, run python api_server.py, and verify with curl http://localhost:8000/api/health.

What API keys are required before generating videos?▼

At minimum you need one LLM key and one image or video provider key configured in backend/config.yaml. Supported providers include DashScope, Volcengine Ark, DeepSeek, OpenAI, Gemini, and Kling, each with its own api_key field.

Related Skills

Explore More Design & Creative Skills→
AceDataCloudAceDataCloud
13

wan-video

Generate AI videos from text, images, or reference videos via the /wan/videos API.

Official
Intermediate
luzhenqianluzhenqian
149

cover-thumbnail

Generate episode cover thumbnails and per-platform metadata from a single preset.

Community
Intermediate
indiearkindieark
1

video-gen-storyboard

Produces high-resolution grid, script-composed, and four-column layouts for video storyboards.

Official
Advanced
EurecaMomentEurecaMoment

sam3-local

Run local SAM3 image segmentation and video prediction APIs.

Official
Intermediate