Ollama Pydantic Project

Ollama Pydantic Project

Welcome to the Ollama Pydantic Project! This project demonstrates how to use a local Ollama model with the Pydantic agent framework to create an intelligent agent. The agent is connected to an MCP server to utilize tools and provides a user-friendly interface using Streamlit.

This project is part of the blog post: Building a TypeScript MCP Server: A Guide for Integrating Existing Services. Visit the blog to learn more about the concepts and implementation details behind this project.

Overview

The main goal of this project is to showcase:

• Local Ollama Model Integration: Using a locally hosted Ollama model for generating responses.
• Pydantic Agent Framework: Creating an agent with Pydantic for data validation and interaction.
• MCP Server Connection: Enabling the agent to use tools via an MCP server.
• Streamlit UI: Providing a web-based chatbot interface for user interaction.

Prerequisites

Before setting up the project, ensure the following:

1. Python: Install Python 3.8 or higher. You can download it from python.org.
2. Ollama Model: Install and run the Ollama server locally:
   • Download the Ollama CLI from Ollama's official website.
   • Install the CLI by following the instructions provided on their website.
   • Start the Ollama server:

     ollama serve
     

   • Ensure the server is running on http://localhost:11434/v1.
3. MCP Server: Set up an MCP server to enable agent tools. For more details, refer to MCP Server Sample.

Setup Instructions

Follow these steps to set up the project:

1. Clone the Repository:

   git clone <repository-url>
   cd ollama-pydantic-project
   

2. Create a Virtual Environment:

   python3 -m venv venv
   

3. Activate the Virtual Environment:

   • On macOS/Linux:

     source venv/bin/activate
     

   • On Windows:

     venv\Scripts\activate
     

4. Install Dependencies:

   pip install -r requirements.txt
   

5. Ensure the Ollama Server is Running: Start the Ollama server as described in the prerequisites.

6. Run the Application: Start the Streamlit application:

   streamlit run src/streamlit_app.py
   

Usage

Once the application is running, open the provided URL in your browser (usually http://localhost:8501). You can interact with the chatbot by typing your queries in the input box. The agent will process your queries using the Ollama model and tools provided by the MCP server.

Example Interaction

Below is an example of how the chatbot interface looks when interacting with the agent:

Project Structure

The project is organized as follows:

ollama-pydantic-project/
├── src/
│   ├── streamlit_app.py        # Main Streamlit application
│   ├── agents/
│   │   ├── base_agent.py       # Abstract base class for agents
│   │   ├── ollama_agent.py     # Implementation of the Ollama agent
│   ├── utils/
│       ├── config.py           # Configuration settings
│       ├── logger.py           # Logger utility
├── requirements.txt            # Python dependencies
├── README.md                   # Project documentation
├── assets/
│   ├── ollama_agent_mcp_example.png  # Example interaction image
├── .gitignore                  # Git ignore file

Features

• Streamlit Chatbot: A user-friendly chatbot interface.
• Ollama Model Integration: Uses a local Ollama model for generating responses.
• MCP Server Tools: Connects to an MCP server to enhance agent capabilities.
• Pydantic Framework: Ensures data validation and type safety.

Related Projects

This project is closely related to another GitHub project that demonstrates how to set up an MCP server: Hello World MCP Server. The blog post associated with that project, Building a TypeScript MCP Server: A Guide for Integrating Existing Services, provides additional context and implementation details that complement this project.

Troubleshooting

• If you encounter issues with the Ollama server, ensure it is running on http://localhost:11434/v1.
• If dependencies fail to install, ensure you are using Python 3.8 or higher and that your virtual environment is activated.
• For MCP server-related issues, refer to the MCP Server Sample.

License

This project is licensed under the MIT License. You are free to use, modify, and distribute this project as per the terms of the license. See the LICENSE file for more details.

Contributing

Contributions are welcome! Feel free to open issues or submit pull requests.