OpenAutoGrowth is an AI-driven, multi-agent growth engine that automates the entire marketing lifecycle through a closed-loop system.
- Intelligent Orchestration: High-level goal interpretation and dynamic task scheduling.
- Multi-Agent Collaboration: Specialized agents for Copywriting, Multimodal Assets (Image/Video), Strategy, and Execution.
- Closed-Loop Optimization: Autonomous performance analysis and strategy correction.
- Modern UI: Premium Glassmorphism dashboard for real-time monitoring.
The system follows a Layered + Multi-Agent + Dynamic Orchestration design.
All design specifications live in the /docs directory:
| Document | Description |
|---|---|
| System Overview | Architecture layers, ADR decisions |
| Agent Design | 8 Agents' functions, I/O specs |
| System Flow | Core pipeline, sequence diagrams |
| State Machine | Campaign, Task, Asset states |
| Domain Model | DDD bounded contexts & aggregates |
| Entity Relations | ER diagram & field definitions |
- User Goal → Interpreted by Orchestrator.
- Planner creates a dynamic Task DAG.
- Execution Agents generate content and deploy to channels.
- Data Analysis Agent pulls results.
- Optimizer Agent refines the strategy and triggers a new cycle.
- Clone the repo:
git clone https://github.com/CadanHu/OpenAutoGrowth.git cd OpenAutoGrowth - Install dependencies:
npm install
- Run local dashboard:
npm run dev
- Framework: Vite + Vanilla JS (Agent Logic)
- Styling: Vanilla CSS (Rich Aesthetics)
- Intelligence: Multi-Agent Orchestration (inspired by LangGraph/AutoGen)
Developed by Antigravity AI for CadanHu.