MCP Integration
Giving Claude Code Superpowers with External Tools
Video Lesson Coming Soon
A video walkthrough for this module is in production. For now, dive into the written content below.
What You'll Learn
- ✓ What MCP (Model Context Protocol) is
- ✓ Why MCP matters for Claude Code
- ✓ Configuring an MCP server
- ✓ Popular MCP servers
- ✓ Project vs user configuration
- ✓ Setting up your first MCP server
In this module 7 sections
What Is MCP?
MCP is an open standard that lets AI tools connect to external services. Each MCP server exposes a set of tools — read a database, search Slack messages, manage GitHub issues, control a browser — that Claude Code can call directly during your conversation.
Think of MCP servers as plugins for Claude Code. Without them, Claude Code can only read files and run shell commands. With them, it can interact with virtually any service or API.
Why MCP Matters
Without MCP, Claude Code lives inside your project directory. It can read code, write code, and run terminal commands. That is powerful, but limited.
With MCP, Claude Code reaches beyond your project:
- Query your database and understand your data model - Read your Slack conversations for context on decisions - Access your Notion docs for project requirements - Manage GitHub issues and pull requests - Control a browser for testing web applications - Connect to Figma designs for reference - Access Sentry for error tracking and debugging
MCP transforms Claude Code from a code editor into a tool that can interact with your entire development ecosystem.
Configuring an MCP Server
MCP servers are configured in a JSON file. You can set them up at the project level (shared with your team) or user level (personal tools).
{
"mcpServers": {
"github": {
"command": "npx",
"args": ["-y", "@modelcontextprotocol/server-github"],
"env": {
"GITHUB_TOKEN": "ghp_your_token_here"
}
},
"filesystem": {
"command": "npx",
"args": ["-y", "@modelcontextprotocol/server-filesystem", "/path/to/allowed/dir"]
}
}
}Popular MCP Servers
| Server | What It Does | Example Use Case |
|---|---|---|
| GitHub | Manage issues, PRs, repos, and code search | "Create a PR for the changes I just made" |
| PostgreSQL | Query databases and understand schemas | "Show me all users who signed up this week" |
| Slack | Read and send messages across channels | "Find the decision about the API redesign in #engineering" |
| Notion | Read and write Notion pages and databases | "Update the project status page with today's progress" |
| Puppeteer | Control a headless browser for testing | "Open localhost:3000 and check if the login page renders" |
| Sentry | Access error tracking and debugging data | "What are the most common errors this week?" |
| Google Drive | Access documents and spreadsheets | "Read the requirements doc in the shared drive" |
Project vs User Configuration
MCP servers can be configured at two levels:
Project level (.claude/mcp.json in your project root) — these tools are available to everyone who works on this project. Good for shared services like the project's database or the team's Slack workspace.
User level (~/.claude/mcp.json in your home directory) — these are your personal tool connections that work across all projects. Good for your personal GitHub token, your own Notion workspace, or tools you always want available.
Setting Up Your First MCP Server
GitHub is the best first MCP server. It is simple to set up, immediately useful, and does not require infrastructure.
Create .claude/mcp.json in your project root (or ~/.claude/mcp.json for user-level). Add the server configuration as shown in the example above.
Most MCP servers need an API token. For GitHub, create a Personal Access Token at github.com/settings/tokens and add it to the env section of the config.
MCP servers are loaded when Claude Code starts. After adding the config, exit and restart Claude Code for the new server to be available.
Ask Claude Code to use the new tool: "List my open GitHub issues" or "Search for PRs I created this week." If it works, you are connected.
The MCP Ecosystem
The MCP ecosystem is growing rapidly. Anthropic maintains official servers for common services, and the community builds new ones constantly. You can also build your own MCP server for internal tools or custom APIs. Check the MCP registry and the Model Context Protocol GitHub organization for the latest available servers.
Key Takeaways
MCP (Model Context Protocol) lets Claude Code connect to external services like databases, Slack, GitHub, and browsers
Configuration goes in .claude/mcp.json (project-level) or ~/.claude/mcp.json (user-level)
Popular servers include GitHub, PostgreSQL, Slack, Notion, Puppeteer, Sentry, and Google Drive
Start with the GitHub MCP server — it is simple to set up and immediately useful for managing issues and PRs