LangChain MCP Weather Application

A complete example demonstrating how to build and consume a Model Context Protocol (MCP) server using:

• Python
• FastMCP
• LangChain
• OpenAI
• Open-Meteo APIs

The project exposes weather information through an MCP server and allows a LangChain agent to discover and invoke weather tools dynamically.

---

Features

The weather tool returns:

• Latitude
• Longitude
• Timezone
• Humidity
• Wind Speed
• Weather Conditions
• Sunrise Time
• Sunset Time
• 7-Day Forecast

The application demonstrates:

• Creating an MCP server with FastMCP
• Registering MCP tools
• Connecting to MCP servers with LangChain
• Discovering tools dynamically
• Building AI agents that invoke MCP tools
• Integrating external REST APIs

---

Architecture

+-------------------+
|   LangChain Agent |
+---------+---------+
          |
          v
+-------------------+
| MCP Client        |
| MultiServerMCP    |
+---------+---------+
          |
          | stdio
          |
          v
+-------------------+
| MCP Server        |
| FastMCP           |
+---------+---------+
          |
          v
+-------------------+
| Weather Service   |
+---------+---------+
          |
          v
+-------------------+
| Open-Meteo APIs   |
+-------------------+

---

Project Structure

project/
│
├── server.py
├── client.py
├── agent.py
├── weather_service.py
├── requirements.txt
├── .env
└── README.md

---

Components

1. weather_service.py

Contains the business logic responsible for:

Geocoding

Converts a city name into:

• Latitude
• Longitude
• Timezone

Uses:

https://geocoding-api.open-meteo.com

Weather Retrieval

Fetches:

• Current humidity
• Wind speed
• Weather conditions
• Sunrise
• Sunset
• 7-day forecast

Uses:

https://api.open-meteo.com

Example Output

{
  "city": "Matadi",
  "latitude": -5.799,
  "longitude": 13.440,
  "timezone": "Africa/Kinshasa",
  "humidity": 82,
  "wind_speed": 12.4,
  "weather_conditions": 1,
  "sunrise": "2026-06-01T06:03",
  "sunset": "2026-06-01T17:58",
  "forecast": [
    {
      "date": "2026-06-01",
      "min_temp": 20.2,
      "max_temp": 29.1
    }
  ]
}

---

2. server.py

Creates the MCP server.

MCP Server Initialization

mcp = FastMCP("WeatherServer")

Tool Registration

@mcp.tool()
def weather(city: str):
    return get_weather(city)

Start Server

mcp.run()

The server exposes the weather tool to any MCP-compatible client.

---

3. client.py

Demonstrates connecting to the MCP server and discovering tools.

Create MCP Client

client = MultiServerMCPClient(
    {
        "weather_server": {
            "transport": "stdio",
            "command": "python",
            "args": ["server.py"]
        }
    }
)

Discover Tools

tools = await client.get_tools()

Example Output

TOOLS
[StructuredTool(name='weather', ...)]

---

4. agent.py

Builds an AI agent capable of invoking MCP tools.

Load Environment Variables

load_dotenv()

Create OpenAI Model

llm = ChatOpenAI(
    model="gpt-5-nano"
)

Retrieve MCP Tools

tools = await client.get_tools()

Create Agent

agent = create_agent(
    model=llm,
    tools=tools,
    system_prompt="You are a weather assistant."
)

Invoke Agent

result = await agent.ainvoke(
    {
        "messages": [
            {
                "role": "user",
                "content": "What is the weather in Matadi?"
            }
        ]
    }
)

---

Installation

Clone Repository

git clone https://github.com/your-org/weather-mcp.git

cd weather-mcp

---

Create Virtual Environment

Using venv

python -m venv .venv

Activate:

Linux / macOS

source .venv/bin/activate

Windows

.venv\Scripts\activate

---

Install Dependencies

pip install -r requirements.txt

---

requirements.txt

mcp
langchain
langchain-openai
langchain-mcp-adapters
python-dotenv
requests
openai

---

Environment Variables

Create a .env file:

OPENAI_API_KEY=your_openai_api_key

---

Running the Application

Option 1: Run MCP Server

python server.py

The server starts and waits for MCP client requests.

---

Option 2: Test MCP Client

python client.py

Example:

TOOLS
[StructuredTool(name='weather', ...)]

---

Option 3: Run LangChain Agent

python agent.py

Example:

{
  "city": "Matadi",
  "latitude": -5.799,
  "longitude": 13.440,
  "timezone": "Africa/Kinshasa",
  "humidity": 82,
  "wind_speed": 12.4,
  "weather_conditions": 1,
  "sunrise": "2026-06-01T06:03",
  "sunset": "2026-06-01T17:58",
  "forecast": [
    {
      "date": "2026-06-01",
      "min_temp": 20.2,
      "max_temp": 29.1
    }
  ]
}

---

MCP Workflow

User Question
      |
      v
LangChain Agent
      |
      v
MCP Tool Discovery
      |
      v
Weather Tool
      |
      v
Open-Meteo APIs
      |
      v
Weather Data
      |
      v
Agent Response

---

Example Queries

What is the weather in Paris?

What is the humidity in New York?

Give me the 7-day forecast for Tokyo.

When is sunrise in London?

What is the wind speed in Matadi?

---

Future Enhancements

• Multiple MCP servers
• Weather alerts
• Historical weather data
• Air quality information
• LangGraph integration
• Redis caching
• Azure deployment
• Streaming responses
• RAG integration
• Multi-agent orchestration

---

Technologies Used

• Python
• FastMCP
• LangChain
• OpenAI
• Open-Meteo API
• AsyncIO
• MCP (Model Context Protocol)

---

Learning Objectives

This project teaches:

1. MCP Server Development
2. MCP Tool Registration
3. MCP Client Integration
4. LangChain Tool Discovery
5. AI Agent Tool Calling
6. REST API Integration
7. Async Python Programming
8. OpenAI + LangChain Integration

---

License

MIT License

Copyright (c) 2026

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files to deal in the Software without restriction.