AI Agent with MCPs using Gemini and PydanticAI

A Python-based AI agent that uses Google Gemini as the LLM and PydanticAI as the agent framework. It integrates with MCP (Model Context Protocol) servers to extend its capabilities, including real-time time lookups and Airbnb listing searches.

Apart from the jupyiter notebook a direct python file is also provided for you if you want easy testing for the server on your preferred IDE, just install the requirements and run the files and do not forget to add the api key

also a note that the version of node in notebook and python files are different

Features

• Basic Q&A powered by Gemini
• Custom date/time tool using Python's datetime
• MCP Time Server integration for timezone-aware time queries
• Airbnb search and listing details via the OpenBnB MCP server

Prerequisites

• Python 3.10 or higher
• Node.js and npx (required for the Airbnb MCP server)
• A Google API key with access to the Gemini API

Setup

1. Clone or navigate to the project directory.

2. Create and activate a virtual environment:

python -m venv venv
source venv/bin/activate

3. Install dependencies:

pip install -r requirements.txt

4. Create a .env file in the project root with your API key:

GOOGLE_API_KEY=your_api_key_here

Usage

Run the test script:

python test_agent.py

You will see a menu with the following options:

| Option | Description | |--------|------------------------------------------| | 1 | Basic agent test (general Q&A) | | 2 | Custom date tool test | | 3 | MCP Time Server test | | 4 | Airbnb agent test (needs Node.js/npx) | | 5 | Run all tests | | 0 | Exit |

Enter one or more choices separated by commas (e.g. 1,3).

Google Colab

The build_airbnb_agent_mcp.ipynb notebook contains a Colab-ready version of the Airbnb agent. Open it in Google Colab, add your API key via Colab Secrets, and run the cells.

Project Structure

.
├── .env                          # Environment variables (API key)
├── build_airbnb_agent_mcp.ipynb  # Colab notebook for the Airbnb agent
├── requirements.txt              # Python dependencies
├── test_agent.py                 # Local test script with interactive menu
└── venv/                         # Python virtual environment