ToolsLast updated: April 202610 min read

30 Best MCP Servers for Claude Code in 2026

MCP servers extend Claude Code with powerful integrations for databases, APIs, cloud platforms, and developer tools. Here are the 30 most useful MCP servers you should know about, with installation instructions for each.

What Are MCP Servers?

MCP (Model Context Protocol) servers are standardized plugins that give Claude Code access to external data sources, APIs, and services. They act as bridges between Claude Code and the tools you use every day -- your database, your project management platform, your cloud provider, your design tool.

Each MCP server exposes a set of tools that Claude Code can invoke during a conversation. For example, a PostgreSQL MCP server might expose tools like query, list_tables, and describe_schema. When you ask Claude Code to "check if there are any users who signed up in the last 24 hours," it uses these tools to query your database and return the answer.

The protocol is open and well-documented, meaning anyone can build an MCP server. For a deeper understanding of how Claude Code works with MCP, read our complete Claude Code guide.

Top 10 MCP Servers (Detailed)

1

GitHub MCP Server

Development Platform

Full GitHub integration including repository management, pull request creation, issue tracking, code review workflows, and Actions management. Essential for any developer using GitHub.

claude mcp add github -- npx -y @anthropic-ai/mcp-server-github
2

PostgreSQL MCP Server

Database

Query, inspect, and manage PostgreSQL databases directly from Claude Code. Supports schema exploration, query execution, migration generation, and performance analysis.

claude mcp add postgres -- npx -y @anthropic-ai/mcp-server-postgres $DATABASE_URL
3

Notion MCP Server

Productivity

Read and write Notion pages, databases, and blocks. Useful for syncing documentation, pulling product specs into your development workflow, or generating reports.

claude mcp add notion -- npx -y @anthropic-ai/mcp-server-notion
4

Redis MCP Server

Database

Interact with Redis instances for cache management, session debugging, pub/sub testing, and data inspection. Supports all major Redis data types and commands.

claude mcp add redis -- npx -y @anthropic-ai/mcp-server-redis
5

AWS MCP Server

Cloud

Manage AWS infrastructure including S3 buckets, Lambda functions, DynamoDB tables, CloudFormation stacks, and IAM policies. Supports multi-region and multi-account setups.

claude mcp add aws -- npx -y @anthropic-ai/mcp-server-aws
6

MongoDB MCP Server

Database

Query and manage MongoDB collections, run aggregation pipelines, inspect indexes, and generate Mongoose schemas. Supports both local and Atlas deployments.

claude mcp add mongodb -- npx -y @anthropic-ai/mcp-server-mongodb
7

Stripe MCP Server

Payments

Interact with the Stripe API for payment processing, subscription management, invoice generation, and webhook debugging. Essential for SaaS and e-commerce projects.

claude mcp add stripe -- npx -y @anthropic-ai/mcp-server-stripe
8

Figma MCP Server

Design

Read Figma designs, extract component properties, color tokens, and spacing values. Bridge the gap between design and code by letting Claude Code reference your designs directly.

claude mcp add figma -- npx -y @anthropic-ai/mcp-server-figma
9

Supabase MCP Server

Backend-as-a-Service

Full Supabase integration covering database queries, authentication management, storage operations, edge functions, and real-time subscriptions.

claude mcp add supabase -- npx -y @anthropic-ai/mcp-server-supabase
10

Firebase MCP Server

Backend-as-a-Service

Manage Firestore collections, Firebase Auth users, Cloud Functions, and Storage buckets. Includes support for security rules testing and emulator integration.

claude mcp add firebase -- npx -y @anthropic-ai/mcp-server-firebase

Servers 11-30: Quick Reference

#ServerCategoryDescription
11Slack MCP ServerCommunicationSend messages, read channels, manage workflows, and search message history.
12Linear MCP ServerProject ManagementCreate and manage issues, cycles, and projects in Linear.
13Jira MCP ServerProject ManagementCRUD operations on Jira issues, sprints, boards, and custom fields.
14Docker MCP ServerDevOpsBuild images, manage containers, inspect networks, and read logs.
15Kubernetes MCP ServerDevOpsManage pods, deployments, services, and config maps across clusters.
16Vercel MCP ServerHostingDeploy projects, manage environment variables, and check deployment logs.
17Sentry MCP ServerMonitoringQuery error reports, investigate stack traces, and manage issue assignments.
18Elasticsearch MCP ServerSearchRun queries, manage indices, and analyze log data.
19OpenAPI MCP ServerAPIParse OpenAPI specs and interact with any REST API automatically.
20GraphQL MCP ServerAPIIntrospect schemas, run queries, and test mutations against GraphQL endpoints.
21Cloudflare MCP ServerCloudManage Workers, KV namespaces, R2 buckets, and DNS records.
22PlanetScale MCP ServerDatabaseManage branches, run queries, and deploy schema changes.
23Twilio MCP ServerCommunicationSend SMS/MMS, manage phone numbers, and configure webhooks.
24SendGrid MCP ServerEmailSend transactional emails, manage templates, and check delivery stats.
25Airtable MCP ServerProductivityRead and write records, manage bases, and sync data with your codebase.
26Google Cloud MCP ServerCloudManage Cloud Run, BigQuery, Cloud Storage, and Pub/Sub resources.
27Azure MCP ServerCloudManage Azure Functions, Blob Storage, CosmosDB, and App Service deployments.
28Prisma MCP ServerORMGenerate and run migrations, inspect database schemas, and seed data.
29Puppeteer MCP ServerTestingAutomate browser interactions, take screenshots, and run E2E tests.
30SQLite MCP ServerDatabaseQuery, inspect, and manage local SQLite databases for prototyping and mobile development.

How to Install MCP Servers

The installation pattern for MCP servers is consistent across the ecosystem. There are two main approaches:

Global Installation (All Projects)

Use the claude mcp add command to register a server globally. It will be available in every project:

# Add a server globally
claude mcp add server-name -- npx -y @scope/mcp-server-name
# List installed servers
claude mcp list
# Remove a server
claude mcp remove server-name

Project-Level Installation

For project-specific servers, add them to your project's .mcp.json file so team members automatically get the same configuration:

{
"mcpServers": {
"postgres": {
"command": "npx",
"args": ["-y", "@anthropic-ai/mcp-server-postgres", "$DATABASE_URL"]
}
}
}

How to Choose the Right MCP Server

With hundreds of MCP servers available, choosing the right ones for your workflow matters. Here are the key factors to evaluate:

  • --Official vs Community: Start with servers published by Anthropic or the tool vendor. They tend to be better maintained and more secure. Community servers can be excellent but review the source code first.
  • --Scope of Tools: Some servers expose dozens of tools while others focus on a few. Fewer, well-designed tools are usually better than a bloated tool set that confuses the model.
  • --Security: MCP servers run with your permissions. A database server has access to your data. Always review what permissions a server requires and limit access where possible.
  • --Maintenance: Check the server's repository for recent commits, issue response times, and compatibility with the latest Claude Code version.
  • --Performance: Some servers add latency to every Claude Code interaction. Prefer servers that use local connections and minimize network calls.

Frequently Asked Questions

What are MCP servers?

MCP (Model Context Protocol) servers are standardized plugins that give Claude Code access to external data sources and services. They enable Claude Code to interact with databases, APIs, cloud platforms, and other tools through a unified protocol.

How do I install an MCP server for Claude Code?

MCP servers are installed by adding them to your Claude Code configuration. You can add them globally via claude mcp add or per-project in your .mcp.json file. Most servers are distributed as npm or Python packages.

Can I build my own MCP server?

Yes. The Model Context Protocol is an open standard with official SDKs for TypeScript and Python. You can build custom MCP servers to connect Claude Code to any internal API, database, or service your team uses. The SDK provides helpers for tool definitions, resource management, and protocol handling.

Are MCP servers safe to use?

MCP servers run locally on your machine and require explicit permission to be used. Claude Code will ask for confirmation before using any MCP server tool for the first time. Always review the source code of community-built MCP servers before installing them, especially those that access sensitive resources like databases or cloud accounts.

Discover More MCP Servers

Browse the full MCP server directory with ratings, installation guides, and community reviews.

Browse MCP Servers