How to Set Up MCP Servers: Quick Start Guide for Claude Code & Cursor

The Model Context Protocol (MCP) is how AI coding tools connect to external capabilities. Think of MCP servers as plugins — each one gives your AI agent a new superpower: browsing the web, reading databases, managing GitHub, automating browsers, and more. Without MCP, your AI can only read and write files. With MCP, it can interact with your entire development stack. This guide shows you how to install and configure MCP servers in under 5 minutes.

Claude Code has built-in MCP support. Add a server with one command:

# Add a server for the current project
claude mcp add server-name npx -y @package/server

# Add a server globally (all projects)
claude mcp add --scope user server-name npx -y @package/server

# List installed servers
claude mcp list

# Remove a server
claude mcp remove server-name

That's it. After adding a server, Claude Code automatically discovers its tools and can use them in conversations.

Cursor reads MCP server configuration from a JSON file:

// .cursor/mcp.json (project-level) or ~/.cursor/mcp.json (global)
{
  "mcpServers": {
    "filesystem": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-filesystem", "/path/to/dir"]
    },
    "github": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-github"],
      "env": {
        "GITHUB_PERSONAL_ACCESS_TOKEN": "ghp_your_token_here"
      }
    }
  }
}

Restart Cursor after editing the config file. The servers will be available in Cursor's Agent mode.

These are the highest-impact servers that every developer should have: 1. Filesystem MCP — Read, write, search, and manage files. The foundation of any agent workflow.

claude mcp add filesystem npx -y @modelcontextprotocol/server-filesystem /path/to/project

2. GitHub MCP — Create PRs, manage issues, search repositories, review code.

claude mcp add github npx -y @modelcontextprotocol/server-github
# Requires GITHUB_PERSONAL_ACCESS_TOKEN env var

3. Playwright MCP — Browser automation, screenshots, web scraping, form filling.

claude mcp add playwright npx -y @playwright/mcp@latest

4. Context7 — Up-to-date documentation for any library. Replaces searching docs manually.

claude mcp add context7 npx -y @context7/mcp@latest

5. Memory MCP — Persistent key-value memory across sessions.

claude mcp add memory npx -y @modelcontextprotocol/server-memory

"Server failed to start" • Make sure Node.js 18+ is installed: node --version • Try running the npx command manually to see error output • Check if the port is already in use "Tools not showing up" • Restart your AI tool after adding the server • Verify the server is listed: claude mcp list • Check server logs: claude mcp logs server-name "Permission denied" • Some servers need environment variables (API keys, tokens) • Check the server's README for required configuration • Use --env flag or set vars in your shell profile "Timeout errors" • Increase the server timeout in settings • Check if the server requires network access (firewalls, VPN) • Try a different transport (stdio vs SSE)

You're set up. Here's where to go from here: • Browse all 14+ MCP servers in the directory • Build your own MCP server for custom tools • Deep dive into the MCP protocol • Chrome DevTools MCP — Debug and profile web apps with AI • n8n MCP — Connect your agent to 400+ workflow integrations