Prototype Reconstruction System: A Multi-Agent Workflow for Prototype Reconstruction in AIGC Short Film Creation

Core Idea: A multi-agent workflow system for AIGC short film creators, focusing on "method复刻而非内容复制" (reproduce film methodology instead of copying content). Key Features:

• Coordinated by Master Agent + 12 sub-agents (A0-A11) to break down excellent creative short films into transferable film methodologies.
• Supports Image2 and Seedance 2.0 backend production. Source:
• Author/Maintainer: wp746
• Platform: GitHub
• Repo: prototype-reconstruction-system
• Release Date: 2026-06-01

AIGC video tools like Image2 and Seedance are advancing, but creators face key issues:

1. Creative Understanding: Hard to grasp why a reference is good (e.g., lens language, rhythm) and transfer it to own work.
2. Production Field Gap: Traditional video analysis only describes frames, lacking actionable fields for Image2/Seedance (e.g., prompt words for composition/rhythm).
3. Copyright Risk: Copying references may infringe on IP, but replacing elements while retaining core methodology is challenging. 12 Key Pain Points: Includes creative comprehension, copyright avoidance, missing production fields, misjudged camera movement/rhythm, style transfer difficulties, etc.

Architecture: Master Agent + 12 sub-agents (A0: task control, A1: copyright check, A2: video analysis, ..., A11: QA). Three Branches:

• A Line: Asset Direct Output (for users with existing assets; uses asset responsibility isolation model).
• B Line: Storyboard Control (for prioritizing lens order/composition; uses clean black-white storyboards).
• C Line: Text Direct Output (for quick validation without assets/storyboards; uses compact causal locks). Core Workflow:

1. Input processing & copyright audit → 2. Video analysis & DNA breakdown →3. Variable/invariant separation →4. New film mapping & compliance →5. Backend handover (generate Image2/Seedance prompts).

Solutions: Addresses 12 key pain points in AIGC video creation. Technical Highlights:

1. DNA_SIGNOFF: Must pass DNA forensics before production to ensure correct methodology extraction.
2. Asset Responsibility Isolation: A line uses strict asset separation (identity/environment/prop/execution/payoff) to avoid confusion.
3. Storyboard Standards: Uses WHITE_STORYBOARD_SHEET_TEMPLATE with 10 fixed sections for Image2 prompts.
4. Clear Reference Mechanism: A/B lines use explicit references (e.g., "Reference A01 / M03 FACE CLOSE-UP") for accurate asset usage.

Significance:

• Evolves AIGC video creation from "prompt engineering" to "methodology engineering".
• Enables creators to learn film language without copying content; improves team production efficiency and quality stability. Usage:

1. Install as Skill: git clone https://github.com/wp746/prototype-reconstruction-system.git
2. Core SOP: Link
3. Recommended Call: Invoke the skill with reference material (video/link/screenshot/asset package) and request DNA breakdown + A/B/C branch output.