What is MCP (Model Context Protocol)?

Definition

MCP stands for Model Context Protocol — an open standard created by Anthropic that defines how AI applications connect to external data sources and tools.

Think of MCP as “USB-C for AI.” Before USB-C, every device had a different connector. Before MCP, every AI tool had to build custom integrations for every data source. MCP provides one standard protocol so any AI application can connect to any compatible data source.

How It Works

MCP uses a client-server architecture:

┌─────────────┐     MCP Protocol      ┌─────────────────┐
│  AI App      │ ←──────────────────→  │  MCP Server     │
│  (Client)    │   JSON-RPC over       │  (Data Source)   │
│              │   stdio / SSE         │                  │
│  e.g. Claude │                       │  e.g. GitHub,    │
│  Code, IDE   │                       │  Slack, Database │
└─────────────┘                        └─────────────────┘

MCP servers expose three types of capabilities:

Tools — Functions the AI can call (e.g., create_pull_request, send_message)
Resources — Data the AI can read (e.g., file contents, database records)
Prompts — Reusable prompt templates

Why It Matters

Interoperability — Build one MCP server, and it works with Claude, VS Code, JetBrains, and any MCP-compatible client
Ecosystem — Growing library of pre-built servers for GitHub, Slack, Google Drive, databases, and more
Security — Standardized permission model instead of ad-hoc API key sharing
Developer experience — Build integrations once instead of per-application

Example

# Building a simple MCP server (Python)
from mcp.server import Server

app = Server("weather-server")

@app.tool()
async def get_weather(city: str) -> str:
    """Get current weather for a city."""
    return f"72°F and sunny in {city}"

@app.resource("weather://{city}/forecast")
async def get_forecast(city: str) -> str:
    """Get 5-day forecast for a city."""
    return f"5-day forecast for {city}: ..."

// Claude Code MCP server configuration (~/.claude.json)
{
  "mcpServers": {
    "weather": {
      "command": "python",
      "args": ["weather_server.py"]
    }
  }
}

MCP vs Direct API Integration

	MCP	Direct API
Reusability	One server, many clients	Custom per app
Discovery	Auto-discovers tools/resources	Hardcoded
Standard	Open protocol	Proprietary
Best for	AI tool integrations	General software

Key Takeaways

MCP is an open standard for connecting AI apps to external tools and data
It eliminates the N×M integration problem (N apps × M data sources → N+M)
Supported by Claude Code, VS Code, JetBrains IDEs, and growing
You can build custom MCP servers to expose any data or functionality to AI tools
The protocol uses JSON-RPC — simple, well-understood, easy to implement

Part of the DeepRaft Glossary — AI and ML terms explained for developers.

MCP