Moya Documentation

Quickstart Guide

Installation

The Moya library is already set up with several example agents and tools. To get started:

pip install moya-ai

Prerequisites

Before running the examples, you'll need:

• Python 3.8 or higher
• An OpenAI API key (for OpenAI examples)
• AWS credentials (for Bedrock examples)
• Ollama installed locally (for Ollama examples)

Basic Usage

Let's walk through a simple example that demonstrates Moya's core functionality:

from moya.agents.openai_agent import OpenAIAgent, OpenAIAgentConfig

# Create an agent configuration
agent_config = OpenAIAgentConfig(
    system_prompt="You are a helpful AI assistant.",
    model_name="gpt-4o"
)

# Initialize the agent
agent = OpenAIAgent(
    agent_name="my_assistant",
    description="A general-purpose AI assistant",
    agent_config=agent_config
)

# Set up the agent
agent.setup()

# Send a message and get a response
response = agent.handle_message("Hello, how can you help me?")

This example demonstrates:

• Agent Configuration: Setting up an agent with specific parameters like the system prompt and model.
• Agent Initialization: Creating an agent instance with a unique name and description.
• Message Handling: Basic interaction with the agent through message passing.

Running Examples

Moya comes with several example scripts that demonstrate different capabilities:

1. Simple Chat Agent

The quick_start_openai.py example shows how to create an interactive chat agent with memory:

python examples/quick_start_openai.py

This example demonstrates:

• Setting up a conversation memory system
• Streaming responses in real-time
• Maintaining conversation context across messages

2. Multi-Agent System

The quick_start_multiagent.py example shows how to create a system with multiple specialized agents:

python examples/quick_start_multiagent.py

This example showcases:

• Creating multiple agents with different specializations
• Automatic message routing between agents
• Shared memory across agents

3. Local Model Integration

For users who prefer to run models locally, the Ollama integration example demonstrates using local models:

# First, install and start Ollama
# Install from https://ollama.ai
ollama run llama2  # Downloads and runs the model

# Then run the example
python examples/quick_start_ollama.py

This example shows:

• Integration with locally hosted models
• Privacy-focused deployment options
• Reduced latency for local inference

Next Steps

Now that you have Moya up and running, you can:

• Explore the Guides to learn about different agent types and features
• Follow the Tutorials to build more complex systems
• Check out the Explanations to understand how Moya works
• Browse the Reference for detailed API documentation