MCP — Model Context Protocol

The Model Context Protocol (MCP) is an open standard created by Anthropic for connecting AI models to external tools, data sources, and services. It provides a universal way for AI agents to call functions, query databases, access APIs, and interact with the world beyond their training data.

What problem does it solve?

AI models are powerful reasoners but can’t natively access files, databases, APIs, or external services. Before MCP, every integration was custom — each tool vendor built their own protocol for each AI platform. MCP standardizes this: build one MCP server and it works with every compatible client.

How it works

An MCP server is a lightweight process that exposes tools (functions the AI can call), resources (data the AI can read), and prompts (templates for common tasks). The AI client discovers available capabilities at startup and invokes them as needed.

# Add an MCP server to Claude Code

claude mcp add my-server -- npx -y my-mcp-server

# Or configure in .cursor/mcp.json

{

"mcpServers": {

"my-server": {

"command": "npx",

"args": ["-y", "my-mcp-server"]

}

MCP capabilities

Capability	Description
Tools	Functions the AI can call — query a database, create a file, send a message
Resources	Data the AI can read — files, database records, API responses
Prompts	Reusable prompt templates that users can invoke

MCP vs SKILL.md

MCP servers and SKILL.md files complement each other. A skill teaches an agent how to approach a task — it’s instructions in markdown. An MCP server gives the agent new capabilities — it’s executable code that extends what the agent can do. For example, a PDF skill teaches the agent best practices for working with PDFs, while a PDF MCP server gives the agent actual tools to read, split, and merge PDF files.

Who supports it

MCP is supported by Claude Desktop, Claude Code, Cursor, VS Code (via GitHub Copilot), Windsurf, Continue, OpenCode, Cline, Roo Code, and a growing number of AI development tools. The ecosystem includes thousands of community-built MCP servers for databases, APIs, cloud services, and more.

Related specs

SKILL.md — for teaching agents how to approach tasks
AGENTS.md — for project-level context

Official Specification

The MCP specification and documentation are maintained by Anthropic.

Read the MCP documentation