Claude Code MCP Servers: How to Connect Any Tool in Under 5 Minutes

MCP (Model Context Protocol) is the most underused feature in Claude Code. Most developers run Claude Code in isolation — but with MCP servers, you can give Claude direct access to your database, browser, file system, APIs, and more.

Here's how to set it up in under 5 minutes.

What MCP Actually Does

Without MCP, Claude Code can only interact with your project via file reads, bash commands, and the tools it ships with.

With MCP, Claude Code gets new tools that you define — and it can call them natively during a conversation.

Real examples:

• Query your PostgreSQL database directly
• Browse URLs and scrape pages
• Search your codebase semantically
• Read from Google Sheets
• Call internal APIs

Step 1: Add an MCP Server to Your Config

MCP servers are configured in ~/.claude/settings.json (global) or .claude/settings.json (project-level).

{
  "mcpServers": {
    "postgres": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-postgres", "postgresql://localhost/mydb"]
    }
  }
}

That's it. Restart Claude Code and it will now have a query_database tool available.

Step 2: Test That It's Connected

In Claude Code, just ask:

> What tables are in my database?

Claude will call the MCP tool and return real results from your DB.

Built-in MCP Servers Worth Adding

These are official servers from the MCP ecosystem — all npx-installable:

Filesystem (extended access)

"filesystem": {
  "command": "npx",
  "args": ["-y", "@modelcontextprotocol/server-filesystem", "/path/to/allowed"]
}

Brave Search (web search from Claude)

"brave-search": {
  "command": "npx",
  "args": ["-y", "@modelcontextprotocol/server-brave-search"],
  "env": {
    "BRAVE_API_KEY": "your-key-here"
  }
}

GitHub (PR/issue management)

"github": {
  "command": "npx",
  "args": ["-y", "@modelcontextprotocol/server-github"],
  "env": {
    "GITHUB_PERSONAL_ACCESS_TOKEN": "your-token"
  }
}

SQLite (local database)

"sqlite": {
  "command": "npx",
  "args": ["-y", "@modelcontextprotocol/server-sqlite", "/path/to/db.sqlite"]
}

Playwright (browser control)

"playwright": {
  "command": "npx",
  "args": ["-y", "@executeautomation/playwright-mcp-server"]
}

Full settings.json with Multiple MCP Servers

Here's a production-ready config:

{
  "model": "claude-opus-4-5",
  "mcpServers": {
    "postgres": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-postgres",
        "postgresql://localhost/myapp_development"]
    },
    "github": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-github"],
      "env": {
        "GITHUB_PERSONAL_ACCESS_TOKEN": "ghp_xxxxxxxxxxxx"
      }
    },
    "brave-search": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-brave-search"],
      "env": {
        "BRAVE_API_KEY": "BSA_xxxxxxxxxxxx"
      }
    },
    "playwright": {
      "command": "npx",
      "args": ["-y", "@executeautomation/playwright-mcp-server"]
    }
  },
  "permissions": {
    "allow": ["Bash", "Read", "Write", "Edit"]
  }
}

Writing Your Own MCP Server

If the built-in servers don't cover your use case, you can write a custom one. Here's a minimal Node.js MCP server:

// my-mcp-server.js
import { Server } from '@modelcontextprotocol/sdk/server/index.js';
import { StdioServerTransport } from '@modelcontextprotocol/sdk/server/stdio.js';

const server = new Server(
  { name: 'my-api', version: '1.0.0' },
  { capabilities: { tools: {} } }
);

// Define a tool
server.setRequestHandler('tools/list', async () => ({
  tools: [{
    name: 'get_user',
    description: 'Get user from internal API',
    inputSchema: {
      type: 'object',
      properties: {
        userId: { type: 'string', description: 'User ID' }
      },
      required: ['userId']
    }
  }]
}));

// Handle tool calls
server.setRequestHandler('tools/call', async (request) => {
  if (request.params.name === 'get_user') {
    const { userId } = request.params.arguments;
    const user = await fetchFromInternalAPI(`/users/${userId}`);
    return {
      content: [{ type: 'text', text: JSON.stringify(user) }]
    };
  }
});

// Start server
const transport = new StdioServerTransport();
await server.connect(transport);

Then add it to your config:

"my-api": {
  "command": "node",
  "args": ["/path/to/my-mcp-server.js"]
}

The Rate Limit Problem with MCP

Here's the catch: MCP tools make Claude Code significantly more powerful, which means it uses more tokens per conversation. Combined with Claude's usage limits, you'll hit rate limit interruptions faster.

The fix most teams use is setting ANTHROPIC_BASE_URL to a rate-limit-free proxy:

export ANTHROPIC_BASE_URL="https://simplylouie.com/api/proxy"

SimplyLouie is ✌️2/month for unlimited Claude access — built specifically for developers who run heavy Claude Code sessions with MCP tools.

Quick Reference: MCP Server List

Server	Package	What It Does
PostgreSQL	@modelcontextprotocol/server-postgres	SQL queries
SQLite	@modelcontextprotocol/server-sqlite	Local DB
Filesystem	@modelcontextprotocol/server-filesystem	File access
GitHub	@modelcontextprotocol/server-github	PRs, issues, code
Brave Search	@modelcontextprotocol/server-brave-search	Web search
Playwright	@executeautomation/playwright-mcp-server	Browser control
Puppeteer	@modelcontextprotocol/server-puppeteer	Browser automation
Slack	@modelcontextprotocol/server-slack	Messages, channels
Memory	@modelcontextprotocol/server-memory	Persistent context

Debugging MCP Connections

If your MCP server isn't connecting, check:

# Test the server manually
npx -y @modelcontextprotocol/server-postgres postgresql://localhost/mydb

# Check Claude Code logs
cat ~/.claude/logs/mcp.log

# Verify the server is listed
claude /mcp

---

MCP is what turns Claude Code from a smart autocomplete into an actual agent that can interact with your real infrastructure. The 5-minute setup is worth it.

What MCP servers are you running? Drop them in the comments.