Gemini Function Calling + Model Context Protocol(MCP) Flight Search

This project demonstrates how to use Google's Gemini 2.5 Pro with function calling capabilities to interact with the mcp-flight-search tool via Model Context Protocol (MCP). This client implementation shows how to:

1. Connect to a local MCP server process (mcp-flight-search) using stdio communication
2. Use natural language prompts with Gemini 2.5 Pro to search for flights (e.g., "Find flights from Atlanta to Las Vegas on 2025-05-05")
3. Let Gemini automatically determine the correct function parameters from the natural language input
4. Execute the flight search using the MCP tool
5. Display formatted results from the search

Features

• Natural language flight search using Gemini 2.5 Pro
• Automatic parameter extraction via function calling
• Integration with mcp-flight-search tool via stdio
• Formatted JSON output of flight results
• Environment-based configuration for API keys

Prerequisites

Before running this client, you'll need:

1. Python 3.7+
2. A Google AI Studio API key for Gemini
3. A SerpAPI key (used by the flight search tool)
4. The mcp-flight-search package installed

Dependencies

This project relies on several Python packages:

• google-generativeai: Google's official Python library for accessing Gemini 2.5 Pro and other Google AI models.

  • Provides the client interface for Gemini 2.5 Pro
  • Handles function calling capabilities
  • Manages API authentication and requests

• mcp-sdk-python: Model Context Protocol (MCP) SDK for Python.

  • Provides ClientSession for managing MCP communication
  • Includes StdioServerParameters for configuring server processes
  • Handles tool registration and invocation

• mcp-flight-search: A flight search service built with MCP.

  • Implements flight search functionality using SerpAPI
  • Provides MCP-compliant tools for flight searches
  • Handles both stdio and HTTP communication modes

• asyncio: Python's built-in library for writing asynchronous code.

  • Manages asynchronous operations and coroutines
  • Handles concurrent I/O operations
  • Required for MCP client-server communication

• json: Python's built-in JSON encoder and decoder.

  • Parses flight search results
  • Formats output for display
  • Handles data serialization/deserialization

Setup

1. Clone the Repository:

   git clone https://github.com/arjunprabhulal/mcp-gemini-search.git
   cd mcp-gemini-search
   

2. Install Dependencies:

   # Install required Python libraries
   pip install -r requirements.txt
   # Install the MCP flight search tool
   pip install mcp-flight-search
   

3. Set Environment Variables:

   export GEMINI_API_KEY="YOUR_GEMINI_API_KEY"
   export SERP_API_KEY="YOUR_SERPAPI_API_KEY"
   

   Replace the placeholder values with your actual API keys:

   • Get your Gemini API key from Google AI Studio
   • Get your SerpAPI key from SerpApi

Architecture

This project integrates multiple components to enable natural language flight search. Here's how the system works:

Component Interactions

1. User to Client

   • User provides natural language query (e.g., "Find flights from Atlanta to Las Vegas tomorrow")
   • Client script (client.py) processes the input

2. Client to MCP Server

   • Client starts the MCP server process (mcp-flight-search)
   • Establishes stdio communication channel
   • Retrieves available tools and their descriptions

3. Client to Gemini 2.5 Pro

   • Sends the user's query
   • Provides tool descriptions for function calling
   • Receives structured function call with extracted parameters

4. Client to MCP Tool

   • Takes function call parameters from Gemini
   • Calls appropriate MCP tool with parameters
   • Handles response processing

5. MCP Server to SerpAPI

   • MCP server makes requests to SerpAPI
   • Queries Google Flights data
   • Processes and formats flight information

Data Flow

1. Input Processing

   User Query → Natural Language Text → Gemini 2.5 Pro → Structured Parameters
   

2. Flight Search

   Parameters → MCP Tool → SerpAPI → Flight Data → JSON Response
   

3. Result Handling

   JSON Response → Parse → Format → Display to User
   

Communication Protocols

1. Client ↔ MCP Server

   • Uses stdio communication
   • Follows MCP protocol for tool registration and calls
   • Handles asynchronous operations

2. MCP Server ↔ SerpAPI

   • HTTPS requests
   • JSON data exchange
   • API key authentication

3. Client ↔ Gemini 2.5 Pro

   • HTTPS requests
   • Function calling protocol
   • API key authentication

Error Handling

The integration includes error handling at multiple levels:

• Input validation
• API communication errors
• Tool execution failures
• Response parsing issues
• Data formatting problems

Usage

Run the client:

python client.py

The script will:

1. Start the MCP flight search server process
2. Send your flight search query to 2.5 Pro
3. Use Gemini's function calling to extract search parameters
4. Execute the search via the MCP tool
5. Display the formatted results

Related Projects

This client uses the mcp-flight-search tool, which is available at:

• GitHub: arjunprabhulal/mcp-flight-search
• PyPI: mcp-flight-search

Author

For more articles on AI/ML and Generative AI, follow me on Medium: @arjun-prabhulal

License

This project is licensed under the MIT License.