Murmur

A sophisticated multi-agent system that orchestrates different specialized AI agents using local LLM models to process and respond to user queries. The system implements a pipeline of Interpreter, Reasoner, Generator, and Critic agents to provide well-thought-out and refined responses.

🌟 Features

• Multi-Agent Architecture: Four specialized agents working in concert:

  • Interpreter: Analyzes user intent and context
  • Reasoner: Develops logical approach to the problem
  • Generator: Creates initial responses
  • Critic: Reviews and refines generated content

• Local LLM Integration: Works with locally hosted language models through a REST API

• Asynchronous Processing: Built with asyncio for efficient concurrent operations

• Robust Error Handling: Comprehensive error management with retries and graceful fallbacks

• Conversation Memory: Maintains context through conversation history

• Confidence Scoring: Evaluates response quality with multiple metrics

🔧 Prerequisites

• Python 3.7+
• Local LLM server (compatible with Ollama API)
• Required Python packages:

  aiohttp
  asyncio
  

🚀 Installation

1. Clone the repository:

git clone https://github.com/KazKozDev/murmur.git
cd murmur

2. Install dependencies:

pip install -r requirements.txt

3. Ensure your local LLM server is running (default: http://localhost:11434)

💻 Usage

1. Navigate to the project directory and run:

cd murmur
python src/main.py

Or navigate directly to the source directory:

cd murmur/src
python main.py

2. Enter your queries when prompted. Type 'quit' to exit.

Example interaction:

Enter your message: What is the capital of France?

Response: The capital of France is Paris.
Confidence: 0.95

🏗️ Architecture

The system follows a pipeline architecture:

1. User Input → Interpreter Agent

   • Analyzes core intent and context
   • Identifies implicit requirements

2. Interpreted Message → Reasoner Agent

   • Breaks down the problem
   • Develops logical approach

3. Reasoning → Generator Agent

   • Creates initial response
   • Structures content clearly

4. Generated Content → Critic Agent

   • Reviews for accuracy and completeness
   • Suggests improvements
   • Produces final version

⚙️ Configuration

The system uses the following default models:

• Interpreter: mistral-nemo:latest
• Reasoner: llama3.2-vision:11b
• Generator: gemma2:9b
• Critic: llama3.2-vision:11b

Models can be configured by modifying the AgentOrchestrator initialization.

🔐 Error Handling

The system implements multiple layers of error handling:

• Connection retries (max 3 attempts)
• Timeout management
• Graceful degradation
• Comprehensive error logging

🤝 Contributing

1. Fork the repository (https://github.com/KazKozDev/murmur/fork)
2. Create your feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add some amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

📝 License

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

🪲 Known Issues

• High CPU usage with multiple concurrent requests
• Memory consumption may increase with long conversations
• Some LLM models may require significant local resources

🔜 Future Improvements

• Add support for streaming responses
• Implement agent personality customization
• Add websocket support for real-time communication
• Enhance conversation memory management
• Add support for more LLM providers
• Implement response caching

📞 Support

For support, please open an issue in the GitHub repository or contact the maintainers.

🙏 Acknowledgments

• Thanks to the Ollama team for their local LLM server
• Inspired by multi-agent architectures in AI systems