MCP stands for Model Context Protocol. It is a standard way to plug extra tools and data sources into Claude Code. Instead of building every integration into the CLI, Claude talks to small external programs called MCP servers, and each server exposes a set of tools Claude can call.
A server can wrap almost anything: your filesystem, a database, a ticketing system, a browser, an internal API. Once a server is connected, its tools show up alongside Claude's built-in ones and Claude can use them during a session.
You register a server with claude mcp add. The part after -- is the
command Claude runs to start the server:
claude mcp add files -- npx -y @modelcontextprotocol/server-filesystem /root
Here files is the name you give the server, and everything after -- is
how it launches. Many servers ship as npm packages, so npx pulls them on
demand the first time they are needed. You do not install anything by hand.
See what is configured with:
claude mcp list
Remove one by name when you no longer need it:
claude mcp remove files
When you add a server, Claude writes it to .mcp.json in the current
project (or to your user config, depending on scope). That file is just
JSON, so you can read it, commit the project-scoped one to share with your
team, or edit it by hand. The next time you start Claude in that project,
the servers are available again.
Your lab setup
This VM comes with Claude Code preinstalled and signed in - just run claude to start.