blog-writer-multi-agents

# Multi-AI Agent Blog Writer

A sophisticated multi-agent system that automatically researches, outlines, writes, and formats professional blog posts using AI agents.

## Overview

This project demonstrates the power of collaborative AI agents working together to create high-quality blog content. The system uses four specialized agents coordinated by a manager agent:

- **Researcher**: Gathers facts and data about topics using web search
- **Outliner**: Creates logical blog structure and outlines
- **Writer**: Generates the actual blog prose
- **Formatter**: Adds final styling and saves the output

## Features

- 🤖 **Multi-Agent Architecture**: Specialized agents for different tasks
- 🔍 **Web Research**: Automatic fact-finding using web search
- 📝 **Structured Writing**: Professional blog generation with proper formatting
- 💾 **File Output**: Automatic saving to markdown files
- 🔧 **Flexible Models**: Support for multiple AI providers (Gemini, Hugging Face, Local)

## Installation

1. Clone the repository:

```bash
git clone <repository-url>
cd multi_ai_agent
```

1. Create a virtual environment:

```bash
python -m venv .venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate
```

1. Install dependencies:

```bash
pip install -r requirements.txt
# or with uv
uv sync
```

## Configuration

Create a `.env` file in the project root:

```env
# For Gemini API (recommended)
GEMINI_API_KEY=your_gemini_api_key_here

# For Hugging Face API
ACCESS_TOKEN=your_huggingface_token_here
```

## Usage

### Basic Usage

```bash
python main.py "Your Blog Topic" [output_file.md]
```

Examples:

```bash
# Write about AI with default output file (blog.md)
python main.py "The Future of Artificial Intelligence"

# Specify custom output file
python main.py "Climate Change Solutions" "climate_blog.md"
```

### Direct Agent Usage

```bash
python agent.py "Your Blog Topic" [output_file.md]
```

## Architecture

### Agent Roles

1. **Researcher Agent**
   - Uses web search to gather relevant facts
   - Limited to 5 steps for efficiency
   - Returns research summaries

2. **Outliner Agent**
   - Creates logical blog structure
   - Generates markdown outlines
   - Ensures content flow

3. **Writer Agent**
   - Writes the actual blog content
   - Follows the provided outline
   - Creates engaging prose

4. **Formatter Agent**
   - Adds final styling and formatting
   - Saves content to specified file
   - Handles file operations

5. **Blog Manager Agent**
   - Coordinates all other agents
   - Manages the complete workflow
   - Ensures quality and consistency

### Supported AI Models

The system supports multiple AI providers:

- **Gemini AI** (default): `gemini/gemini-2.5-flash`
- **Hugging Face**: Various models available
- **Local AI**: Custom OpenAI-compatible servers

## Dependencies

- `smolagents`: Multi-agent framework
- `litellm`: AI model integration
- `python-dotenv`: Environment variable management
- `ddgs`: Web search capabilities

## Example Output

The system generates well-structured blog posts with:

- Clear introduction and conclusion
- Logical section organization
- Research-backed content
- Professional formatting

## Development

### Project Structure

```text
multi_ai_agent/
├── agent.py          # Main agent system
├── main.py           # CLI interface
├── tools/
│   └── file_write.py # File writing tool
├── pyproject.toml    # Project configuration
└── README.md         # This file
```

### Customization

You can modify the agents in `agent.py` to:

- Change model providers
- Adjust agent behavior
- Add new tools
- Modify prompts and workflows

## Contributing

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Test thoroughly
5. Submit a pull request

## License

This project is open source. Please check the license file for details.

## Troubleshooting

### Common Issues

1. **API Key Errors**: Ensure your `.env` file contains valid API keys
2. **Web Search Failures**: Check internet connectivity
3. **File Permission Errors**: Ensure write permissions in output directory

### Debug Mode

For debugging, you can modify the `max_steps` parameter in agent definitions or add logging statements to trace execution flow.

## Future Enhancements

- [ ] Support for additional AI providers
- [ ] Custom blog templates
- [ ] Image generation integration
- [ ] Multi-language support
- [ ] SEO optimization features