MOYA - Meta Orchestration framework for Your Agents

MOYA is a reference implementation of our research paper titled "Engineering LLM Powered Multi-agent Framework for Autonomous CloudOps". The framework provides a flexible and extensible architecture for creating, managing, and orchestrating multiple AI agents to handle various tasks autonomously.

Preprint of the paper can be accessed at arXiv.

Features

• Agent Management: Create, register, and manage multiple AI agents.
• Orchestration: Orchestrate conversations and tasks across multiple agents.
• Memory Tools: Integrate memory tools to maintain conversation context and history.
• Streaming Responses: Support for streaming responses from agents.
• Extensibility: Easily extend the framework with new agents, tools, and orchestrators.

Getting Started

pip install moya-ai  # Only core framework
pip install moya-ai[all] # All the supported services such as OpenAI, Bedrock, Ollama, Crewai
# or install specific ones - for example openai and ollama
pip install moya-ai[openai, ollama]

Contributing to MOYA

We accept contributions exclusively through forked repositories. Please follow these steps:

1. Fork this repository to your GitHub account
2. Create a new branch in your fork for your changes
3. Make your changes and commit them to your branch
4. Submit a pull request from your fork's branch to our main repository

Prerequisites

• Python 3.10+
• Install required dependencies:

  pip install .

Quick Start Examples

OpenAI Agent

Interactive chat example using OpenAI agent with conversation memory.

# filepath: ~/github/moya/examples/quick_start_openai.py

python -m examples.quick_start_openai

Bedrock Agent

Interactive chat example using BedrockAgent with conversation memory.

# filepath: ~/github/moya/examples/quick_start_bedrock.py

AWS_PROFILE=my_profile_name python -m examples.quick_start_bedrock

Multi-Agent Orchestration

Example demonstrating multi-agent orchestration with language and task classification.

# filepath: ~/github/moya/examples/quick_start_multiagent.py

python -m examples.quick_start_multiagent

Dynamic Agent Creation

Example demonstrating dynamic agent creation and registration during runtime.

# filepath: ~/github/moya/examples/dynamic_agents.py


python -m examples.dynamic_agents

Directory Structure

moya/
├── agents/                # Agent implementations (OpenAI, Bedrock, Ollama, Remote)
├── classifiers/           # Classifier implementations for agent selection
├── memory/                # Memory repository implementations
├── orchestrators/         # Orchestrator implementations for managing agent interactions
├── registry/              # Agent registry and repository implementations
├── tools/                 # Tool implementations (e.g., MemoryTool)
├── examples/              # Example scripts demonstrating various use cases
└── README.md              # This README file

Contributing

We welcome contributions to the MOYA framework. Please follow these steps to contribute:

1. Fork the repository.
2. Create a new branch for your feature or bugfix.
3. Make your changes and commit them with descriptive messages.
4. Push your changes to your fork.
5. Create a pull request to the main repository.

License

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

Contact

For any questions or inquiries, please contact the authors of the research paper or open an issue on the GitHub repository.