Building Your First MCP Server

This lesson teaches you to build a production-ready MCP server in Python. You'll create a "Crypto Price Tracker" MCP that fetches live cryptocurrency prices, allowing Claude to analyze crypto markets in real-time.

Setup: MCP SDK Installation

bash

pip install mcp

This installs Anthropic's MCP SDK (server and client libraries).

Building Crypto Price Tracker MCP

File Structure:

code

crypto-mcp/
├── server.py          (MCP server)
├── requirements.txt
└── CLAUDE.md

Step 1: Write server.py

python

import asyncio
import json
from mcp.server import Server
from mcp.types import Tool, Resource, TextContent
import httpx

app = Server("crypto-tracker")

# Resource: Live crypto prices
@app.resource("crypto://prices")
async def get_crypto_prices():
    async with httpx.AsyncClient() as client:
        response = await client.get(
            "https://api.coingecko.com/api/v3/simple/price",
            params={"ids": "bitcoin,ethereum,cardano", "vs_currencies": "usd,pkr"}
        )
    prices = response.json()
    return TextContent(
        mime_type="application/json",
        text=json.dumps(prices, indent=2)
    )

# Tool: Analyze crypto trends
@app.tool("analyze_crypto_trend")
async def analyze_trend(crypto: str, days: int = 7):
    async with httpx.AsyncClient() as client:
        response = await client.get(
            f"https://api.coingecko.com/api/v3/coins/{crypto}/market_chart",
            params={"vs_currency": "usd", "days": days}
        )
    data = response.json()
    prices = data["prices"]
    current = prices[-1][1]
    previous = prices[0][1]
    change = ((current - previous) / previous) * 100
    return f"{crypto.capitalize()} changed {change:.2f}% in last {days} days"

if __name__ == "__main__":
    asyncio.run(app.run())

Step 2: requirements.txt

code

mcp>=0.1.0
httpx>=0.24.0

Step 3: Test Server

bash

python server.py

Server runs on localhost:3000. Claude Code can now call it.

Using Your MCP Server

From Claude Code:

bash

claude run "What's the current price of Bitcoin in PKR?"

Claude (via your MCP server):

Calls crypto://prices resource
Gets live data from CoinGecko API
Returns: "Bitcoin is currently PKR 5,400,000"

Analyze crypto trends:

bash

claude run "Ethereum's trend over 30 days?"

Claude calls analyze_crypto_trend(ethereum, 30) tool, returns analysis.

Pakistan-Specific MCP: JazzCash Integration

Pakistani developer builds "JazzCash MCP" (payments):

python

from mcp.server import Server
import requests

app = Server("jazzcash-integration")

@app.tool("process_payment")
async def process_payment(phone: str, amount: int, description: str):
    # Call JazzCash API
    payload = {
        "pp_merchant_id": "YOUR_MERCHANT_ID",
        "pp_password": "YOUR_PASSWORD",
        "pp_amount": amount,
        "pp_bill_ref": description,
        "pp_cell": phone
    }
    response = requests.post(
        "https://sandbox.jazzcash.com.pk/ApplicationAPI/API/Payment/DoTransaction",
        json=payload
    )
    return response.json()

@app.tool("check_payment_status")
async def check_status(transaction_id: str):
    # Check payment status
    response = requests.post(
        "https://sandbox.jazzcash.com.pk/ApplicationAPI/API/Payment/QueryTransaction",
        json={"transaction_id": transaction_id}
    )
    return response.json()

if __name__ == "__main__":
    asyncio.run(app.run())

Now Claude can process payments in real-time:

bash

claude run "Process PKR 1,000 payment from 03001234567 for invoice INV-001"

Claude calls MCP tool → JazzCash API → Payment processed. All seamlessly.

Practice Lab

Task 1: Build Crypto MCP — Create the Crypto Price Tracker MCP. Test that it works: (1) Run server, (2) Call it from Claude Code, (3) Verify live prices are fetched.

Task 2: Extend with Pakistan — Add a new tool to Crypto MCP: convert_to_pkr(amount_usd) that converts USD prices to PKR using live exchange rates (XE.com API).

Pakistan Example

Bilal builds "Stock Market MCP" for Pakistani stocks (PSX):

python

@app.resource("psx://stocks")
async def get_psx_stocks():
    # Fetch PSX data (would integrate with PSX API if available)
    stocks = {
        "HBL": 210.50,  # HBL Bank
        "LUCK": 545.80,  # Lucky Cement
        "ENGRO": 128.75  # Engro Chemical
    }
    return TextContent(text=json.dumps(stocks))

@app.tool("analyze_psx_stock")
async def analyze_stock(symbol: str):
    # Returns analysis
    return f"{symbol} is a strong performer in PSX"

Pakistani investors use this MCP through Claude: "Show me top 5 PSX stocks to buy today". Claude queries MCP, analyzes market data, returns recommendations.

2.2 — Building Your First MCP Server — File System & Database Access

Building Your First MCP Server

Setup: MCP SDK Installation

Building Crypto Price Tracker MCP

Using Your MCP Server

Pakistan-Specific MCP: JazzCash Integration

Practice Lab

Pakistan Example

Lesson Summary

Building Your First MCP Server Quiz