Distributed RAG Architecture

Overview

This project implements a distributed Retrieval-Augmented Generation (RAG) architecture designed to integrate multiple data sources, specialized agents, and large language models (LLMs) for enhanced information retrieval and decision-making processes.

Key Features

• Multi-Database Integration: Supports Vector DB, RDBMS, and NoSQL databases.
• Specialized Agents: Includes Code, Cyber Security, Project Management, and Cyber Research agents.
• Enhanced Consensus Mechanism: Implements a sophisticated vote-based decision-making algorithm.
• LLM Integration: Incorporates multiple LLMs for various reasoning tasks.
• Robust Error Handling: Comprehensive error management for database operations, API calls, and general exceptions.

Installation

1. Clone the repository:

   git clone https://github.com/your-username/distributed-rag-architecture.git
   cd distributed-rag-architecture
   

2. Install the required dependencies:

   pip install -r requirements.txt
   

3. Set up your config.json file with the necessary configurations for databases, APIs, and LLMs.

Usage

1. Ensure your config.json file is properly set up with your database connections, API endpoints, and LLM configurations.

2. Run the main script:

   python main.py
   

3. The script will demonstrate the workflow of the distributed RAG architecture, including:

   • Task assignment by the Project Manager agent
   • Code generation by the Code agent
   • Security evaluation by the Cyber Security agent
   • Threat research by the Cyber Research agent
   • LLM-based reasoning
   • Consensus-based decision making
   • Database queries and API calls

Configuration

The config.json file should include the following sections:

• databases: Configuration for Vector DB, RDBMS, and NoSQL databases.
• apis: Endpoints for external API calls.
• llm: Settings for different LLMs used by various agents.

Example structure:

{
  "databases": {
    "vector_db": { "host": "vector_db_host", "port": 5000 },
    "rdbms": { "host": "rdbms_host", "port": 3306 },
    "nosql_db": { "host": "nosql_host", "port": 27017 }
  },
  "apis": {
    "security_updates": "https://api.example.com/security-updates"
  },
  "llm": {
    "security": { "name": "SecurityLLM", "endpoint": "https://llm.example.com/security" },
    "code": { "name": "CodeLLM", "endpoint": "https://llm.example.com/code" }
  }
}

Extending the Project

To extend this project:

1. Add new agent types by creating new classes that inherit from the Agent base class.
2. Implement new database connectors by extending the Database base class.
3. Enhance the consensus mechanism by modifying the EnhancedConsensusMechanism class.
4. Add new LLM integrations in the LLMReasoning class.

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

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

Acknowledgments

• This project was inspired by the need for more sophisticated RAG architectures in AI applications.
• Special thanks to the open-source community for providing valuable tools and libraries that made this project possible.