OmniNoval

OmniNoval is an advanced multimodal multi-agent automation framework, built on LangGraph, supporting complex agent collaboration workflows.

🌟 Features

• Multi-agent Collaboration: Supports multiple specialized agents working together, each with specific roles and expertise
• Multimodal Support: Integrated vision models, supporting image understanding and processing
• Flexible Workflow Engine: Extensible workflow system based on LangGraph
• Web Automation: Integrated Playwright, supporting web browsing and automation operations
• RESTful API: Provides complete API interfaces, supporting integration into existing systems
• Docker Support: Provides complete containerized deployment solutions
• Flexible Configuration: Supports multiple LLM providers and configuration options

📋 Directory Structure

OmniNoval/
├── assets/                 # Static resource files
├── src/                    # Source code directory
│   ├── agents/            # Agent definitions
│   ├── api/               # API interfaces
│   ├── config/            # Configuration management
│   ├── engine/            # Workflow engine
│   ├── llms/              # LLM abstraction layer
│   ├── prompts/           # Prompt templates
│   ├── service/           # Service layer
│   ├── tools/             # Toolset
│   ├── utils/             # Utility functions
│   └── workflow.py        # Workflow entry
├── conf.yaml.example      # Configuration file example
├── docker-compose.yml     # Docker Compose configuration
├── Dockerfile            # Docker image configuration
├── main.py               # CLI entry
├── server.py             # API server entry
├── pyproject.toml        # Project configuration
└── README.md             # Project documentation

🚀 Quick Start

Environment Requirements

• Python 3.12+
• uv package manager (recommended)

Local Installation

1. Clone the Project

   git clone <repository-url>
   cd OmniNoval

2. Install Dependencies

   # Install dependencies using uv
   uv sync
   
   # Install Playwright browser
   uv run playwright install chromium --with-deps

3. Configure Environment

   # Copy configuration file
   cp conf.yaml.example conf.yaml
   
   # Edit configuration file, set your API keys
   vim conf.yaml

4. Run the Project

   # CLI mode
   uv run main.py "Your query"
   
   # API server mode
   uv run server.py --reload

Using Makefile

The project provides convenient Makefile commands:

# Initial setup
make setup

# Run CLI
make run

# Start development server
make dev

# Start production server
make api

# Run tests
make test

# Clean environment
make clean

🐳 Docker Deployment

Using Docker Compose (Recommended)

# Start production environment
docker-compose up api

# Start development environment (supports hot reload)
docker-compose up dev

Manual Docker Image Build

# Build image
docker build -t omninova .

# Run container
docker run -p 8000:8000 -v $(pwd)/.env:/app/.env -v $(pwd)/conf.yaml:/app/conf.yaml omninova

⚙️ Configuration Instructions

Configuration File Structure

The project supports two configuration methods:

• conf.yaml configuration file (recommended)
• .env environment variable file (compatibility mode)

LLM Configuration

Configure different LLM models in conf.yaml:

# Reasoning Model
REASONING_MODEL:
  model: "volcengine/ep-xxxx"
  api_key: $REASONING_API_KEY
  api_base: $REASONING_BASE_URL

# Basic Model
BASIC_MODEL:
  model: "azure/gpt-4o-2024-08-06"
  api_base: $AZURE_API_BASE
  api_version: $AZURE_API_VERSION
  api_key: $AZURE_API_KEY

# Vision Model
VISION_MODEL:
  model: "azure/gpt-4o-2024-08-06"
  api_base: $AZURE_API_BASE
  api_version: $AZURE_API_VERSION
  api_key: $AZURE_API_KEY

Environment Variables

Create .env file to set sensitive information:

# API Keys
REASONING_API_KEY=your_reasoning_api_key
AZURE_API_KEY=your_azure_api_key

# API Endpoints
REASONING_BASE_URL=https://api.example.com
AZURE_API_BASE=https://your-resource.openai.azure.com
AZURE_API_VERSION=2024-02-15-preview

📖 Usage

CLI Mode

# Direct run
python main.py "Help me analyze the content of this webpage"

# Interactive mode
python main.py
# Then input your query

API Mode

Start API server:

python server.py --host 0.0.0.0 --port 8000 --reload

API Endpoints:

• GET /health - Health check
• POST /workflow - Execute workflow
• GET /docs - API documentation (Swagger UI)

Workflow Example

from src.workflow import run_agent_workflow

# Execute workflow
result = run_agent_workflow(
    user_input="Help me analyze the content of this webpage",
    debug=True
)

# View result
print(result["messages"])

🔧 Development Guide

Project Structure Explanation

• agents/: Defines various agent roles and behaviors
• engine/: Workflow engine, built on LangGraph
• llms/: LLM abstraction layer, supports multiple model providers
• tools/: Toolset, including web browsing, search, etc.
• api/: RESTful API interfaces

Adding New Agents

1. Create new agent file in src/agents/
2. Register agent configuration in src/config/
3. Integrate new agent in workflow

Adding New Tools

1. Create tool file in src/tools/
2. Implement tool interface
3. Register and use tool in agents

Code Standards

The project uses the following tools for code quality control:

# Code formatting
uv run black .

# Code checking
uv run ruff check .

# Run tests
uv run pytest

🤝 Contribution Guide

1. Fork the project
2. Create feature branch (git checkout -b feature/AmazingFeature)
3. Commit changes (git commit -m 'Add some AmazingFeature')
4. Push to branch (git push origin feature/AmazingFeature)
5. Open Pull Request

📝 License

This project is licensed under the MIT License - see the LICENSE file for details.

🆘 FAQ

Q: How to switch different LLM providers?

A: Modify model configuration in conf.yaml, supports all LiteLLM compatible providers.

Q: How to add custom tools?

A: Create new tool class in src/tools/ directory, and register use in agents.

Q: How to debug workflow?

A: Set debug=True parameter, or check log output.

Q: What browser automation features are supported?

A: Based on Playwright, supports all features of Chromium browser.

📞 Support

If you encounter problems or have suggestions, please:

1. Check Issues page
2. Create new Issue
3. Contact development team

---

OmniNoval - Making multi-agent collaboration simple and efficient 🚀