What is MCP (Model Context Protocol)?

What is MCP (Model Context Protocol)?

MCP (Model Context Protocol) is an open standard developed by Anthropic that defines how AI models connect to external tools, data sources, and services. It's the "USB standard for AI" — a universal interface for LLMs to interact with the world.

Core Problem MCP Solves

textCopy
Before MCP:
  Claude → Custom code → Tool A
  Claude → Custom code → Tool B
  GPT-4  → Custom code → Tool A  (different custom code!)

With MCP:
  Any AI client → MCP Protocol → [Tool A, Tool B, Tool N...]

Architecture

textCopy
┌─────────────┐    MCP (JSON-RPC)    ┌─────────────────┐
│  MCP Client │ ←──────────────────→ │   MCP Server    │
│  (Claude,   │                      │  (Tool provider) │
│   Cursor,   │                      │  - File system  │
│   IDE...)   │                      │  - Database     │
└─────────────┘                      │  - APIs         │
                                     └─────────────────┘

What MCP Servers Expose

search_web

run_sql

write_file

code-review-checklist

Simple MCP Server

pythonCopy
from mcp.server import Server
from mcp.server.stdio import stdio_server
import mcp.types as types

app = Server("weather-server")

@app.list_tools()
async def list_tools():
    return [types.Tool(
        name="get_weather",
        description="Get weather for a city",
        inputSchema={
            "type": "object",
            "properties": {"city": {"type": "string"}},
            "required": ["city"]
        }
    )]

@app.call_tool()
async def call_tool(name: str, arguments: dict):
    if name == "get_weather":
        return [types.TextContent(type="text",
            text=f"Weather in {arguments['city']}: 22°C, sunny")]

async def main():
    async with stdio_server() as (read, write):
        await app.run(read, write, app.create_initialization_options())

Configure Claude to Use MCP Servers

jsonCopy
{
  "mcpServers": {
    "filesystem": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-filesystem", "/Users/me/projects"]
    },
    "github": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-github"],
      "env": {"GITHUB_TOKEN": "ghp_..."}
    }
  }
}

Popular MCP Servers

Key Benefits

• Security — AI only accesses what MCP servers explicitly expose
• Standardization — one protocol works across Claude, Cursor, and other clients
• Composability — combine multiple MCP servers for rich tool access
• Open source — anyone can publish MCP servers