What Are MCP Servers?

Module: MCP Servers | Lesson: 1 of 4 | Time: ~10 minutes

What You Will Learn

What the Model Context Protocol (MCP) is
How MCP servers extend Claude's capabilities
How MCP tools differ from built-in tools

Prerequisites

Completed Module 07: Skills

What Is MCP?

MCP stands for Model Context Protocol. It is an open standard created by Anthropic that lets AI tools like Claude Code connect to external services and data sources.

Think of it this way: without MCP, Claude Code can already do a lot. It can read and edit files on your computer, run terminal commands, and search through your code. But what if you want Claude to:

Query a database?
Create a GitHub issue?
Search the web for documentation?
Control a web browser?
Access files outside your current project?

That is where MCP comes in. Each of those abilities comes from an MCP server that you connect to Claude Code.

Think of MCP Servers as Phone Apps

Your smartphone by itself can make calls and send texts. But when you install apps, it can do so much more — navigate with maps, order food, stream music. MCP servers work the same way for Claude Code. Each server you add is like installing a new app that gives Claude a new capability.

How Claude Code Works Without MCP

Out of the box, Claude Code has a set of built-in tools:

Built-in Tool	What It Does
Read	Reads files in your project
Edit	Modifies files in your project
Write	Creates new files
Bash	Runs terminal commands
Grep	Searches file contents
Glob	Finds files by name pattern

These tools are powerful, but they are limited to your local machine and the files in your current project directory.

How MCP Extends Claude

When you connect an MCP server, Claude gains new tools that it can use just like the built-in ones. For example:

Connect the GitHub MCP server and Claude gets tools like create_issue, create_pull_request, and list_repos
Connect the Brave Search MCP server and Claude gets a web_search tool
Connect a PostgreSQL MCP server and Claude gets tools to query your database

Claude sees these new tools in its toolbox and can decide when to use them, just like it decides when to read a file or run a command.

How It Works Under the Hood

Here is the simplified flow:

You configure an MCP server (we will cover how in the next lesson)
When Claude Code starts, it launches the MCP server as a background process
The server tells Claude what tools it provides
When Claude decides to use one of those tools, it sends a request to the MCP server
The server does the work (queries the database, calls the API, etc.) and returns the result
Claude uses that result to continue helping you

You  -->  Claude Code  -->  MCP Server  -->  External Service
                                              (GitHub, DB, etc.)

Local vs Remote

Most MCP servers run locally on your machine as a small process. Claude communicates with them through standard input/output (called "stdio"). Some newer MCP servers can run remotely over HTTP, but local stdio servers are by far the most common.

Built-in Tools vs MCP Tools

Feature	Built-in Tools	MCP Tools
Available by default	Yes	No — must be configured
Where they run	Inside Claude Code	In a separate MCP server process
What they access	Local files and terminal	Anything the server connects to
Examples	Read, Edit, Bash	GitHub, Brave Search, PostgreSQL
Permission needed	Standard tool permissions	Same permission flow as built-in tools

Key Point

From Claude's perspective, MCP tools work the same as built-in tools. Claude does not treat them differently. It simply has more tools available in its toolbox.

Why MCP Matters for You

As you get more comfortable with Claude Code, MCP servers let you:

Do more without leaving Claude — no need to switch to your browser for GitHub or a database client for queries
Automate complex workflows — Claude can chain together local file edits with API calls and database queries
Customize Claude for your workflow — add exactly the servers that match how you work

Try It Yourself

You do not need to install anything yet. For now, just check what MCP servers (if any) are already connected:

Open Claude Code in any project
Type /mcp and press Enter
You should see a list of connected MCP servers (it may be empty — that is fine!)

This command is your go-to tool for checking MCP status.

What You Learned

MCP (Model Context Protocol) is an open standard for connecting Claude to external tools and services
MCP servers are like apps — each one adds a new capability to Claude Code
Without MCP, Claude can only work with local files and terminal commands
With MCP, Claude can query databases, manage GitHub repos, search the web, and much more
MCP servers run as background processes and communicate with Claude through a standard protocol

Help Us Improve

How was this lesson? Take 2 minutes to share your feedback — it helps us make the tutorials better for everyone.

Give Feedback →

Next Up

Next: Configuring MCP Servers

What You Will Learn​

Prerequisites​

What Is MCP?​

How Claude Code Works Without MCP​

How MCP Extends Claude​

How It Works Under the Hood​

Built-in Tools vs MCP Tools​

Why MCP Matters for You​

Try It Yourself​

What You Learned​

Next Up​