What is MCP (Model Context Protocol)?

MCP (Model Context Protocol) is an open standard introduced by Anthropic in November 2024 that standardizes how AI assistants like Claude connect to external tools, databases, and data sources. Think of it as USB for AI—a universal way to plug capabilities into any AI system.

Quick Answer

Before MCP, every AI application had to build custom integrations for each tool or data source. MCP solves this with a standard protocol:

• One standard: Write an MCP server once, use it with any MCP-compatible AI
• Two-way communication: AI can both read data and take actions
• Open source: Anyone can build MCP servers or clients

Major adopters include Claude Desktop, Cursor, Windsurf, and many developer tools.

How MCP Works

MCP has three components:

1. MCP Hosts (AI Applications)

Applications that want to connect to tools:

• Claude Desktop
• Cursor
• Custom AI applications

2. MCP Servers (Tool Providers)

Services that expose capabilities:

• File system access
• Database queries
• API integrations
• Browser automation

3. MCP Protocol

The standard that connects them:

• JSON-RPC based
• Supports resources, tools, and prompts
• Works over stdio, HTTP, or other transports

┌─────────────┐     MCP Protocol     ┌─────────────┐
│  Claude     │ ◄──────────────────► │  Postgres   │
│  Desktop    │                      │  MCP Server │
└─────────────┘                      └─────────────┘

What Can MCP Do?

Resources

Read-only data sources the AI can access:

• File contents
• Database records
• API responses
• Screenshots

Tools

Actions the AI can take:

• Run SQL queries
• Create files
• Send messages
• Control browsers

Prompts

Pre-built prompt templates for specific tasks.

Example: Using MCP with Claude Desktop

1. Install an MCP server (e.g., filesystem access):

// claude_desktop_config.json
{
  "mcpServers": {
    "filesystem": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-filesystem", "/Users/me/Documents"]
    }
  }
}

2. Claude can now:
   • List files in your Documents folder
   • Read file contents
   • Search across files

Popular MCP Servers

Server	Capability
filesystem	Read/write local files
postgres	Query PostgreSQL databases
github	Access repos, issues, PRs
slack	Read/send Slack messages
brave-search	Web search
puppeteer	Browser automation
sqlite	Query SQLite databases

Why MCP Matters

Before MCP

• Each AI app needed custom integrations
• Tool makers built for specific AI products
• Fragmented ecosystem

After MCP

• Build once, work everywhere
• AI apps get instant access to tool ecosystem
• Tool makers reach all MCP-compatible AIs

Building MCP Servers

Create your own MCP server in Python or TypeScript:

from mcp.server import Server
from mcp.types import Tool

server = Server("my-tool")

@server.tool()
async def get_weather(city: str) -> str:
    """Get current weather for a city."""
    # Your implementation
    return f"Weather in {city}: Sunny, 22°C"

server.run()

Related Questions

• How to create MCP tools?
• What is agentic AI?
• Best MCP tools for Claude?

---

Last verified: 2026-03-02