MCP servers are lightweight bridge services that connect Claude Code to external data sources, APIs, and tools through the standardized Model Context Protocol. MCP servers enable Claude to directly query databases, interact with cloud platforms, and access third-party services without manual data copying, making them essential infrastructure for any serious Claude Code workflow.
MCP1,253 verified servers
30 Best Claude Code MCP Servers — Connect to Everything
The Model Context Protocol (MCP) is a standardized interface that lets Claude Code communicate with external systems in real time. Instead of copying data into prompts or manually describing API responses, MCP servers give Claude direct, structured access to databases, cloud platforms, version control systems, monitoring dashboards, and dozens of other tools developers rely on daily.
Think of MCP servers as adapters. Each server wraps a specific service or data source and exposes it through a consistent protocol that Claude understands natively. When you connect the Postgres MCP server, Claude can query your production database, inspect schemas, and write migrations without you ever pasting a SQL dump into chat. Connect the GitHub MCP server, and Claude gains the ability to browse repositories, read pull request diffs, search issues, and even create commits on your behalf.
This directory contains 1,200+ verified MCP servers, making it the most comprehensive collection available. Each server has been tested for protocol compliance, documentation quality, and real-world reliability. We cover servers across every major category: database connectors (Postgres, MySQL, SQLite, MongoDB), cloud platform integrations (AWS, GCP, Azure, Supabase), developer tool bridges (GitHub, Linear, Sentry, Datadog), and file system and search servers that let Claude navigate local and remote codebases.
Getting started is straightforward. Most MCP servers require just a few lines of JSON configuration in your project's settings file. You specify the server command (often an npx or Docker invocation), pass any required API keys as environment variables, and restart Claude Code. The server will appear in Claude's available tools list immediately. For teams, MCP server configurations can be committed to your repository so every developer on the project shares the same integrations.
Whether you are building a full-stack application and need Claude to query your database, managing infrastructure and want Claude to read your cloud logs, or automating workflows across multiple SaaS tools, MCP servers turn Claude Code from a capable assistant into a deeply integrated development partner. Browse the collection below, sorted by popularity, and find the servers that fit your stack.
Quick Comparison: Top 5 MCP Servers
Name
Best For
Protocol
Difficulty
Postgres MCP
Database queries, schema inspection, migrations
stdio
Beginner
GitHub MCP
Repository browsing, PR reviews, issue management
stdio
Beginner
Filesystem MCP
Local file access, directory traversal, file search
stdio
Beginner
Puppeteer MCP
Web scraping, browser automation, screenshot capture
stdio
Intermediate
AWS MCP
Cloud resource management, S3, Lambda, CloudWatch
stdio
Advanced
How to Install an MCP Server
Most MCP servers can be added to Claude Code with a single CLI command. The claude mcp add command registers the server in your project configuration and makes it immediately available in your next session.
# Add the Postgres MCP server
claude mcp add postgres-server -- npx @anthropic-ai/mcp-server-postgres \
claude mcp add github-server -- npx @anthropic-ai/mcp-server-github \
--token "$GITHUB_TOKEN"
# List all active MCP servers
claude mcp list
After adding a server, restart your Claude Code session. The server's tools will appear automatically in Claude's available toolkit. You can verify the connection by asking Claude to list its available MCP tools.
All MCP Servers
Top 60 of 1,253
Vultr API MCP Server
Advanced
Wraps the Vultr API as MCP tools so Claude can call every endpoint with validated inputs.
An MCP (Model Context Protocol) server is a lightweight service that exposes external data sources, APIs, or tools to Claude Code through a standardized protocol. It acts as a bridge between Claude and systems like databases, file storage, cloud platforms, or third-party APIs, allowing Claude to read, query, and interact with resources it otherwise cannot access directly.
How do I install an MCP server?
Most MCP servers can be installed by adding a configuration block to your Claude Code settings file (typically .claude/settings.json or the project-level CLAUDE.md). You specify the server command, any required arguments, and environment variables such as API keys. Many servers are also available as npm packages or Docker containers for easier setup.
Which MCP servers are official?
Anthropic maintains several official MCP servers including the filesystem server, GitHub server, and the Postgres database server. These are part of the @anthropic-ai/mcp-servers package and receive regular updates. Community-built servers follow the same protocol specification and are often just as capable for their specific use cases.
Can I build my own MCP server?
Yes. The Model Context Protocol is an open specification, and Anthropic provides SDKs in TypeScript and Python to help you build custom servers. You define the resources and tools your server exposes, implement the handler functions, and Claude Code will automatically discover and use them. The official MCP SDK documentation includes starter templates.
Are MCP servers free?
The MCP protocol itself is free and open source. Most community-built MCP servers are also free and open source. However, the underlying services they connect to (such as cloud databases, SaaS APIs, or premium data sources) may have their own pricing. Always check the requirements and costs of the external service an MCP server connects to.
Know an MCP server we're missing?
Help us build the most complete directory of Claude Code MCP servers.
