- Home
- Skills
- Agents & Orchestration
- AI-driven infinite creation in animation - one-click generat
Works with the AI Agents you already use
AI-driven infinite creation in animation - one-click generat
Managing the leap from a raw novel or script to a fully realized short video is a fragmented, manual process that breaks character consistency.
Free
AI-driven infinite creation in animation - one-click generat
Example session with this skill installed
I have a 3-chapter wuxia novel. I need you to generate a character bible for the protagonist, create a storyboard for the first chapter, and prepare the ComfyUI prompts for a 16:9 cinematic video.
- Read your context and instructions
- Compiled the ai-driven infinite creation
I have analyzed the novel and generated the production assets. The protagonist, Li Chen, has been defined with a consistent visual anchor (white robes, jade pendant) and a calm, deep vocal profile. I have split the first chapter into 12 shots using a cinematic style guide.
Production Assets
| Asset Type | Description | Reference ID |
|---|---|---|
| Character Bible | Male, age 24, white silk robes, silver hair crown, jade pendant. | bible_protagonist_01 |
| Style Guide | Wuxia cinematic, 4k, misty mountains, high contrast, teal and orange. | style_cinematic_wuxia |
| Storyboard | 12 shots total. Shot 1: Wide shot of mountain peak. Shot 2: Close-up Li Chen eyes. | storyboard_ch1_v1 |
ComfyUI Prompt Preview (Shot 01)
{
"positive_prompt": "cinematic wide shot, misty mountain peak, ancient chinese architecture, Li Chen standing on the edge, white robes fluttering, 4k, highly detailed, wuxia style",
"negative_prompt": "modern, car, electricity, blurry, low resolution, distorted face",
"resolution": "1920x1080",
"workflow_kind": "t2i"
}
Next steps
- Run
pnpm run generate:imagesto create the initial character and scene assets. - Review the
bible.mdandstyle.mdfiles in the project directory. - Execute the
videophase to begin generating the MiniMax H3 video shots.
Connects securely to your tools. The creator never sees your data.
About this skill
Managing the leap from a raw novel or script to a fully realized short video is a fragmented, manual process that breaks character consistency. This skill provides a local, full-stack Agent solution for automated "Text-to-Video" production, ensuring character and style persistence across every shot.
What it does
- Script Analysis translates raw novels into structured screenplays, storyboards, and character bibles.
- Consistency Management creates a "Bible" and "Style Guide" to anchor visual and vocal traits across the entire production.
- Asset Generation automates the creation of character views, scene backgrounds, props, and vocal profiles.
- Video Orchestration manages the MiniMax H3 Ref2VA pipeline, combining reference images and audio into shot-specific video clips.
- Workflow Integration maps custom ComfyUI workflows (T2I, I2I, R2V, TTS) into executable Agent skills.
How it works
- Input Stage: The user provides a script or story idea via the CLI or web interface.
- Analysis Stage: The agent generates a project manifest including character assets and a storyboard JSON.
- Synthesis Stage: The system calls ComfyUI nodes to generate images, audio samples, and individual video shots based on the storyboard.
- Final Assembly: The agent uses ffmpeg to merge shots, audio, and subtitles into a final video file.
Frameworks & tools
Built with Node.js (Next.js 15), SQLite, and ffmpeg. It interacts with ComfyUI via REST/WebSocket, specifically optimized for MiniMax H3, Qwen-Image 2.1, and Qwen3-TTS.
Why this beats prompting it yourself
Writing prompts for a single image is easy, but maintaining a character's face, clothing, and voice across 50 individual shots is nearly impossible for a manual prompter. This skill implements a systematic "Reference-Anchor" methodology that automates asset reuse and shot-by-shot consistency.
Use cases
- Manga Dramas: Converting web novels into animated shorts for social media platforms.
- Concept Previz: Rapidly prototyping film scenes with consistent characters and lighting.
- Workflow Automation: Turning complex ComfyUI JSON templates into callable Agent skills.
Known limitations
Requires a local ComfyUI instance and specific models for optimal results. High VRAM is necessary for video generation, and Bun is required for the initial Agent kernel build.
Requires local ComfyUI instance. Bun needed for kernel build. Performance depends on local VRAM availability. MiniMax H3 and Qwen models are the primary tested targets.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
1 use
Times used to date
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 1 month ago
- Free with an Agensi account
Requires