infinite-creation
SwotAtmk/infinite-creation
infinite-creation is an agent skill project by SwotAtmk on GitHub: Submit a novel or script and it produces video continuously with local ComfyUI + MiniMax H3 + an agent, with no length limit. It covers adaptation, prompts and production, with prompts for ComfyUI (local) and MiniMax H3. 76 stars, last updated 2026-10-16, MIT license.
Its 28 skills also include ad and promo-video ones
Editor's notes
It is a local Next.js web app that needs Node 22+, pnpm, Bun and ffmpeg, plus a local ComfyUI (or a RunningHub cloud key) and an OpenAI-compatible LLM key. Generation is split into four stages, text writing, asset images, voice design and shot videos, each of which can be run or resumed on its own; videos are rendered shot by shot in sequence, and each shot can be previewed and regenerated individually. The author runs it on an RTX 3080 20G, and VRAM requirements have not been tested in detail. Pasting a new ComfyUI workflow JSON makes the system parse its nodes and draft a matching SKILL.md.
- Type
- All-in-one: script → storyboard → video
- Stages
- Adaptation · Prompts · Production
- Video models
- ComfyUI (local) · MiniMax H3
- Format
- Agent skill
- Language
- Chinese
- GitHub stars
- ★ 76
- Last updated
- 2026-10-16
- License
- MIT
- SKILL.md files
- 30
- Checked by hand
- 2026-10-13
In the author's words
infinite-creation takes a submitted novel or screenplay and uses local ComfyUI-MiniMax H3 + Agent + Skill to produce video fully automatically and endlessly, with no length limit.
Translated from the Chinese original.
Skills in this repo (30)
Names and descriptions come from each SKILL.md frontmatter; Chinese descriptions are shown in our English translation.
- infinite-creation
A local end-to-end "novel/screenplay → manju/short video" creation agent solution. Three ways to use it: ① launch this project and run it in one click; ② without launching the project, have the agent follow this spec plus the skills/ library to reproduce the whole pipeline (script → character bible/style → asset images → storyboard → per-shot video → merged final cut); ③ plug in the user's own text-to-image/image-to-image/video/voice ComfyUI workflows. Use when the user wants to clone/install/launch this project, generate manju/short videos with this project's approach, or connect custom workflows.
- 0715-scriptwriter
Develops a story outline, plot synopsis or project proposal into a complete script of about 70 to 90 episodes, 1 to 2 minutes each, for a Chinese AI manju, vertical short drama or micro-drama, covering audience decisions, transformation of reference works, character and long-arc design, payoff beats and twists, AI-visualizable description, and the 0715 rich-text script format. Use when the user asks to expand an outline, write long scripts episode by episode, write scripts for short-video platforms or manju, design characters and episodes, design hooks and twists, or explicitly asks for the 0715 script format; not for plain novel writing, storyboard image generation or text-to-video prompts.
- 3d-animation-short-generator
Create complete stylized 3D animated shorts from a story idea through an ordered production workflow covering project brief, story outline, character and environment cards, standardized shot planning, text or optional pencil storyboards, video-model selection, single-shot generation, assembly, BGM…
- ai-manga-planner
AI manju pre-production planning assistant offering a complete 9-stage manju planning workflow. Activates when the user says "help me plan a manju", "I want to make an AI manju", "start manju planning", "manju planning workflow", or mentions "AI manju" together with "planning/script/outline". Supports full-pipeline output from requirements gathering, market analysis, worldbuilding, character design, story structure and scriptwriting to storyboard scripts (with image prompts), AI video prompts (with camera movement) and editing and music design, adapted to mainstream AI generation platforms such as Seedance, Kling, Vidu, Runway and Midjourney.
- ai-music-studio
Uses AI to generate scores, vocals and sound effects from a description or melody: style setting, arrangement, mixing and vocal processing.
- brand-promo-video-generator
For marketers and creators producing promotional content for brands, products, websites, apps, shops, or personal projects. Users provide logos, product images, interface screenshots, official links, or other verifiable assets and confirm duration, aspect ratio, audience, and campaign focus. The Sk…
- cinematic-ai-comic-director
A professional film-grade AI manju director, deeply adapted to WorkBuddy knowledge-base calls, strictly following 6-step script creation + 7-step manju production + automatic matching rules for five cinematic shot libraries. Given a genre, story synopsis and episode length, it automatically outputs a full structured set of manju materials: script, storyboard script (with complete references to 80+ shot prompts), AI image keywords, voice-over and subtitle plan, and final-cut editing standards, ready for one-click copying into Excel. This skill should be used when the user needs to create AI comic drama, generate cinematic storyboard scripts with lens matching, produce AI image p…
- co-op-game-intro-generator
For users creating a two-player co-op game menu or opening animation. Users provide two player names, a game title, a target visual style, and optional character reference images. The Skill locks identity cues, generates an approval image from a fixed menu framework with coordinated color, buttons,…
- comic-drama-generator
High-quality premium 3D manju generator. Use this skill when the user needs to make short-video manju, generate storyboard scripts, or create character/scene image prompts. Supports Seedance 2.0 model adaptation, outputting complete storyboard scripts, character prompts and scene prompts that can be used directly for AI video generation. Trigger words: "manju", "storyboard script", "video script", "character prompt", "scene prompt", "3D manju", "short drama generation".
- era-consistency-optimizer
Era consistency optimizer: optimizes AIGC manju prompts so that every prop, tech product, costume, building, vehicle and other element in a scene matches the specified era exactly, avoiding anachronistic objects and improving the realism and professionalism of the generated content. Triggers: an explicit or implied period setting (e.g. the Tang dynasty, 1990s China, the Victorian era, near-future 2045).
- h3-prompt-writing
Write MiniMax H3 video generation prompts for T2VA, I2VA, FL2VA, L2VA, and Ref2VA. Use when rewriting multimodal requests into H3 prompt structures, composing integrated_multimodal_description, overall_soundscape, and non_diegetic_music, aligning keyframes, or defining reference labels for images,…
- handdrawn-live-video-generator
For creators making surreal short videos that blend rough glowing hand-drawn animation with live-action spaces. Users provide a scene idea, contact object or hand, desired mood, and optional language or style constraints. The Skill clarifies the physical contact, designs continuous morphing, escape…
- hf-drama-storyboard-script
Generates text storyboard scripts for short dramas from episode scripts and the director's visual suggestions. Suited to shot breakdown, shot-size design, camera blocking, VFX notes, editing rhythm, selling-point tag integration and shootable storyboard JSON output for short dramas, manju, in-feed ad dramas and vertical dramas.
- image-prompt-writing
Writes Chinese and English prompts for Krea2 hyper-realistic text-to-image (t2i) and Qwen-Image-Edit-2511 image-to-image (i2i). When to use: generating asset images for manju projects, such as character turnaround sheets, scene environment images and props on white backgrounds, or making consistent variants from a reference image. Outputs prompt and negative_prompt.
- manga-drama-generator
One-click full-pipeline AI manju generation skill. Converts novel content into the full set of materials needed for AI anime short-drama production: a professional script, character settings (information extraction + bio + visual keywords + turnaround-sheet prompts), scene settings (extraction + scene prompts), storyboard scripts (breakdown + first-frame/last-frame/video prompts), and standardized Excel table output. Use when the user needs to turn a novel into a manju/anime short drama, generate storyboard scripts or AI image prompts, or mentions manju generation, novel-to-anime or AI short drama.
- manga-full-stack
AI manju full-stack skill pack: a full-pipeline AI manju production skill integrated from 19 professional sources, covering storyboard prompts (14-element formula), character consistency control (five-dimension framework + reference-image locking + parameter pinning), script structure design (three-act structure + universal story formula), multiple style systems (3D anime / period xuanhuan fantasy / cyberpunk / dark suspense) and 50+ camera-movement prompt templates.
- manju-director-agent
AI manju director agent (equivalent to the Coze 3 workflow solution). Novel → script → element extraction → design prompts → storyboard, all in one go. Triggers: "manju director", "AI manju director", "manju agent", "make a manju", "novel to manju", "help me make a manju episode".
- micro-drama-creator
Comprehensive short drama and manju creation assistant: a full-pipeline AI-assisted creation tool from topic planning, scriptwriting, character design and storyboarding to production advice, supporting vertical short dramas, motion comics, anime-style comics and other formats.
- minimalist-product-ad-generator
Turn product images and ad requirements into minimalist product ad shorts for e-commerce promotion and product launches. The Skill confirms format and product variants, extracts selling points, writes concise English ad copy, builds product anchors, plans beat-synced typography/storyboards, and gen…
- music-caption-rewriter
Turn a brief music description and optional tagged lyrics into a professional MiniMax Music 3 structured caption with Global Metadata, Vocal Details, and a section-aware Arrangement. Use when users ask to enhance a music-generation prompt, preserve lyric-section directives, retrieve a similar style…
- music-video-subtitle-generator
For musicians, video creators, and social-media editors producing AI music videos or emotional short films with lyric typography. Users provide music, lyrics, references, characters, typography direction, mood, or target platform. The Skill analyzes beat and vocal timing, separates character, scene…
- novel-to-skitscreenplay
Converts long novels into script format that meets manju video standards. Manju = a film's opening (into the story within 3 seconds) + a TV series' density (3 conflicts per minute) + the rhythm of crosstalk comedy (constant punchlines/twists) + an emotional drug (extreme satisfaction or angst). Trigger words: "novel to script", "novel to screenplay", "short drama adaptation", "manju adaptation", "script format conversion", "turn a novel into a script", "manju video script".
- novel-to-storyboard
Automatically converts novel/script text into storyboard scripts and extracts character feature cards and scene feature cards, to support film, comic and short-video production.
- novel-to-video
Fully automatically adapts a novel or story into a finished short video/manju with voice. When to use: the user provides novel text or a story idea and wants fully automatic video output with consistent character looks, continuous scenes, and a storyboard that can be reviewed and regenerated segment by segment. Covers: plot adaptation, script and storyboard, character/scene/prop asset generation, MiniMax H3 Ref2VA video generation, and final-cut stitching. Not for: single images, simple retouching, or single shots without a plot.
- paper-collage-explainer-generator
For creators, educators, and social-video editors who need a tactile paper-collage language for narration, knowledge points, opinions, or abstract topics. Users provide source copy, story beats, or a core concept and may specify aspect ratio, duration, palette, and audio needs. The Skill extracts m…
- papercraft-stop-motion-explainer
For creators explaining science, education, or general knowledge through tactile handmade papercraft visuals. Users provide a topic, core knowledge points, or source material and may specify audience, duration, aspect ratio, and deliverable type. The Skill extracts the learning goal and visual meta…
- script-master
[v2.1] Master orchestrator for script creation (the user's only interaction entry point). Five-layer knowledge base: pattern-library (macro patterns) / writing-craft-library (mid-level scene craft, including AI-tone detection) / lines-library (micro-level dialogue) / martial-arts-library (fight scenes) / sound-design-library (sound). Call it when the user has an idea, outline, first draft, consultation or dialogue-diagnosis need. This skill is the only entry point; internally it reads the downstream expert documents under references/ by stage and follows their instructions, manages a nine-stage pipeline, and outputs a complete executable script package. When the user wants to deliver the script or show it to the client, it switches to delivery mode and outputs [Project Introduction] and [Project Script…
- script-to-manga
Professional script-to-AI-manju generation skill. Triggers when the user provides a script, scene passage, character profile or plot excerpt and wants standardized prompts for AI manju/AI video. Supports splitting full scripts, parsing single scenes and adapting character settings, outputting film-grade AI video prompts (with comic style, cinematic tone, camera plan, and composition and lighting parameters) for mainstream AI video models such as Kling, Sora, Runway and Seedance. Trigger words: "script to manju", "generate manju prompts", "script to manga", "AI manju", "comic-style prompts", "script to video", "scene prompts".
- story-pipeline-cn
Chinese novel/script creation pipeline: novel to script (mining material → three-act structure → finished scene script), script to storyboard (storyboard table + character profile cards + AI image/video prompts), and novel straight to storyboard (one-click breakdown of characters/scenes/shots, outputting JSON compatible with downstream generation tools), forming a complete "novel → script → storyboard" chain. Use when the user needs to adapt a novel into a script, write a storyboard script, break a novel into shots, or generate character profile cards and AI image prompts.
- tts-voice-design
Writes character voice parameters for the Qwen3-TTS voice design workflow. When to use: designing a dedicated voice (voice sample voice_ref) for a manju character, which requires a representative line, text, and a voice description, voice_description. Outputs text and voice_description.
Install
npx skills add SwotAtmk/infinite-creation
# install a single skill
npx skills add SwotAtmk/infinite-creation --skill infinite-creation
# pick agents (claude-code / codex / cursor / opencode / gemini-cli / openclaw)
npx skills add SwotAtmk/infinite-creation -a claude-code -a codexDefault commands use the skills CLI and were not verified one by one; follow the repo README. Read SKILL.md before installing. Install guide →
Similar projects
Short-drama production controller: novel outline/characters/script, director's storyboard, prompts, sound assets, rough cut and QC (16:9 by default)
Dramake: text to AI short drama, chaining script, storyboard, voiceover, editing and QC, for Douyin/Hongguo in landscape or vertical, with built-in evals
End-to-end short-drama automation: script, character design, storyboard, image and video generation, and submission
- OnlyShot★ 322
Shenxianyu short-drama pipeline: one line → script, storyboard, storyboard images, video and editing in five steps, with 17 documented failure modes
- workrally★ 172
Open-sourced by Tencent: end-to-end toolkit for AI agents to create AIGC manju videos