mcp-server-tutorial

Step-by-step tutorial: create an MCP (Model Context Protocol) server that connects AI agents (Claude, GPT, etc.) to real-time data sources.

MCP (Model Context Protocol) lets AI agents call external tools and data sources. Instead of the AI guessing, it can query live APIs, databases, and services.

Example: Ask Claude "What's trending on Hacker News?" → Claude calls your MCP server → server queries HN API → returns real data.

pip install mcp
python server.py

Build an MCP server that researches any company — domain info, tech stack, WHOIS data.

from mcp.server import Server
from mcp.types import Tool, TextContent
import requests

server = Server("company-researcher")

@server.tool()
async def research_company(domain: str) -> list[TextContent]:
    """Research a company by its domain name."""
    results = {}

    # WHOIS data
    whois_url = f"https://rdap.org/domain/{domain}"
    try:
        resp = requests.get(whois_url, timeout=10)
        if resp.ok:
            data = resp.json()
            results["registrar"] = data.get("entities", [{}])[0].get("vcardArray", [None, []])[1][1][3] if data.get("entities") else "unknown"
            results["created"] = data.get("events", [{}])[0].get("eventDate", "unknown")
    except Exception:
        results["whois"] = "unavailable"

    # DNS records
    dns_url = f"https://dns.google/resolve?name={domain}&type=A"
    try:
        resp = requests.get(dns_url, timeout=5)
        if resp.ok:
            answers = resp.json().get("Answer", [])
            results["ip_addresses"] = [a["data"] for a in answers]
    except Exception:
        results["dns"] = "unavailable"

    # Tech detection via headers
    try:
        resp = requests.head(f"https://{domain}", timeout=5, allow_redirects=True)
        headers = dict(resp.headers)
        results["ser