Claude Academy
Loading

Introducing MCP

Lesson 472 min

Model Context Protocol (MCP) is a communication layer that provides Claude with context and tools without requiring you to write a bunch of tedious integration code. Think of it as a way to shift the burden of tool definitions and execution away from your server to specialized MCP servers.

MCP architecture: our server holds the MCP client, which connects to two MCP servers, each reaching its own outside service.

When you first encounter MCP, you'll see diagrams showing the basic architecture: an MCP Client (your server) connecting to MCP Servers that contain tools, prompts, and resources. Each MCP server acts as an interface to some outside service.

Understanding MCP Through a Real Example

Let's say you're building a chat interface where users can ask Claude about their GitHub data. A user might ask "What open pull requests are there across all my repositories?" To answer this, Claude needs tools to access GitHub's API.

Our server holds two tools, get_repos and get_pull_requests, each with a schema and a function that calls the GitHub API.

Without MCP, you'd need to create all the GitHub integration tools yourself. This means writing schemas and functions for every piece of GitHub functionality you want to support.

The Tool Function Problem

GitHub has massive functionality - repositories, pull requests, issues, projects, and much more. To build a complete GitHub chatbot, you'd need to author an incredible number of tools:

Our server holding eight GitHub tools, such as get_repos, search_issues and create_file, each with a schema and a function.

Each tool requires both a schema definition and a function implementation. This represents a lot of code that you have to write, test, and maintain as a developer.

How MCP Solves This

MCP shifts the burden of tool definitions and execution from your server to MCP servers. Instead of you writing all those GitHub tools, they're authored and executed inside a dedicated MCP server.

Our server holds a get_repos tool made of "schema" and "fn" blocks, with two-way arrows to an empty MCP server and Claude.

The MCP server acts as a wrapper around GitHub's functionality, providing pre-built tools that you can use without having to implement them yourself.

Our server connects both ways to Claude and an MCP server whose get_repos tool has a schema and a function calling GitHub.

MCP servers provide access to data or functionality implemented by outside services. They package up complex integrations into reusable components that any application can connect to.

Common Questions About MCP

Three common questions about MCP servers, on authoring, how they differ from calling an API, and tool use, answered below.

Who Authors MCP Servers?

Anyone can create an MCP server implementation. Often, service providers themselves will make their own official MCP implementations. For example, AWS might release an official MCP server with tools for their various services.

How is MCP Different from Direct API Calls?

MCP servers provide tool schemas and functions already defined for you. If you call an API directly, you're responsible for authoring those tool definitions yourself. MCP saves you that implementation work.

Isn't MCP Just Tool Use?

This is a common misconception. MCP servers and tool use are complementary but different concepts. MCP is about who does the work of creating and maintaining the tools. With MCP, someone else has already written the tool functions and schemas for you - they're packaged inside the MCP server.

The key insight is that MCP servers provide tool schemas and functions already defined for you, eliminating the need to build and maintain complex integrations yourself.