Skip to main content
Back to all articles
AWS

AWS MCP Servers for Claude Code: Overview of 45+ AI-Powered Tools

Justin Pursati
Justin Pursati
12 min read
Share:

The Complete AWS MCP Server Ecosystem

In an earlier post I walked through how MCP servers let a small team move faster on AWS.

That article covered just 6 servers. The full ecosystem contains 45+ specialized MCP servers across 9 categories—from infrastructure automation to AI integration.

Here's what's actually available today.

Let's Be Clear: This Builds On Your AWS Knowledge

Before diving in, a word on expectations. MCP servers are force multipliers for engineers who understand AWS fundamentals. They don't replace your expertise—they amplify it.

What you still need:

  • Understanding of AWS concepts and services
  • Clear vision of your application's architecture
  • Awareness of cost implications and trade-offs

What MCP servers provide:

  • Automated implementation of best practices you already know
  • Natural language interface to complex AWS APIs
  • Consistent application of security patterns
  • Time savings on repetitive configuration tasks
  • Protection against common misconfigurations

Think of it as having an experienced colleague who never forgets a security group rule or IAM policy detail. Your AWS knowledge directs the strategy; MCP servers handle the implementation precision.

What This Means for Your Business

  • Infrastructure tasks that took hours now take minutes
  • Junior developers can handle complex AWS architectures
  • AI handles the complexity while you focus on business logic

From 6 to 45+: The Complete AWS MCP Ecosystem

When AWS Labs released their MCP (Model Context Protocol) servers, they didn't just ship a few tools—they built out a whole catalog for managing infrastructure through natural language.

The 9 Categories That Cover Everything

🏗️ Core Infrastructure (5 servers)

CDK, Terraform, CloudFormation, Documentation, Core Management

🧠 AI & Machine Learning (5 servers)

Nova Canvas, Bedrock KB, Amazon Q, Kendra, Data Automation

⚡ Compute & Serverless (2 servers)

Lambda Tools, Serverless Framework Integration

🐳 Containers & Orchestration (3 servers)

EKS, ECS, Batch Processing

💾 Database Services (9 servers)

DynamoDB, Aurora, DocumentDB, ElastiCache, RDS, MemoryDB

🔐 Security & Identity (3 servers)

IAM, Secrets Manager, Systems Manager

📊 Monitoring & Operations (5 servers)

CloudWatch, Prometheus, Health, Cost Analysis

🔄 Integration & Messaging (8 servers)

SNS/SQS, EventBridge, Step Functions, API Gateway

🛠️ Development Tools (5 servers)

Diagrams, Frontend, Documentation, Git Research, OpenAPI

The Top 10 MCP Servers Worth Knowing

Here are the servers I've found make the biggest difference in day-to-day work:

1. 🎯 Core MCP Server - The Orchestrator

# Set your AWS credentials first
export AWS_PROFILE=your-profile
export AWS_REGION=us-east-1
export MCP_LOG_LEVEL=ERROR
 
# Add the server to your project (creates .mcp.json in your project root)
claude mcp add awslabs.core-mcp-server -s project -e FASTMCP_LOG_LEVEL=$MCP_LOG_LEVEL -- uvx awslabs.core-mcp-server@latest

What it does: Acts as the brain of your AWS infrastructure, automatically managing and coordinating all other MCP servers.

Real impact: Once you're comfortable with the natural language interface, the routine setup work that used to eat an afternoon collapses into a short conversation.

2. 💰 Cost Analysis MCP Server - The Money Saver

claude mcp add awslabs.cost-analysis-mcp-server -s project -e FASTMCP_LOG_LEVEL=$MCP_LOG_LEVEL -e AWS_PROFILE=$AWS_PROFILE -- uvx awslabs.cost-analysis-mcp-server@latest

What it does: Analyzes AWS costs in real-time and predicts expenses before deployment.

Real impact: It surfaces cost optimization opportunities in your existing infrastructure that are easy to miss by hand.

3. 🚀 AWS Serverless MCP Server - The Speed Demon

claude mcp add awslabs.aws-serverless-mcp-server -s project -e FASTMCP_LOG_LEVEL=$MCP_LOG_LEVEL -e AWS_PROFILE=$AWS_PROFILE -e AWS_REGION=$AWS_REGION -- uvx awslabs.aws-serverless-mcp-server@latest

What it does: Builds, deploys, and manages serverless applications with natural language commands.

Real impact: Deploy a complete API with authentication in a few minutes. No YAML wrestling required.

4. 🗄️ DynamoDB MCP Server - The Data Wizard

claude mcp add awslabs.dynamodb-mcp-server -s project -e FASTMCP_LOG_LEVEL=$MCP_LOG_LEVEL -e AWS_PROFILE=$AWS_PROFILE -e AWS_REGION=$AWS_REGION -- uvx awslabs.dynamodb-mcp-server@latest

What it does: Manages NoSQL databases with English commands instead of complex queries.

Real impact: Complex database operations that typically require deep DynamoDB expertise become accessible to any developer who understands basic database concepts.

5. 🎨 Nova Canvas MCP Server - The Creative Engine

claude mcp add awslabs.nova-canvas-mcp-server -s project -e FASTMCP_LOG_LEVEL=$MCP_LOG_LEVEL -e AWS_PROFILE=$AWS_PROFILE -e AWS_REGION=$AWS_REGION -- uvx awslabs.nova-canvas-mcp-server@latest

What it does: Generates images for your applications using Amazon's Nova AI model.

Real impact: Create branded assets, diagrams, and UI mockups without leaving your development environment.

6. 🏗️ CDK MCP Server - The Infrastructure Artist

claude mcp add awslabs.cdk-mcp-server -s project -e FASTMCP_LOG_LEVEL=$MCP_LOG_LEVEL -- uvx awslabs.cdk-mcp-server@latest

What it does: Writes infrastructure as code with built-in security best practices and compliance.

Real impact: Generate CDK code with sensible security defaults baked in, so there's far less to fix when it's reviewed.

7. 🔍 Git Repo Research MCP Server - The Code Detective

claude mcp add awslabs.git-repo-research-mcp-server -s project -e FASTMCP_LOG_LEVEL=$MCP_LOG_LEVEL -e AWS_PROFILE=$AWS_PROFILE -e AWS_REGION=$AWS_REGION -- uvx awslabs.git-repo-research-mcp-server@latest

What it does: Uses AI to understand and navigate large codebases with semantic search.

Real impact: Semantic search finds relevant code much faster than traditional grep/find commands, especially in unfamiliar codebases.

8. 📊 CloudWatch Logs MCP Server - The Insight Generator

claude mcp add awslabs.cloudwatch-logs-mcp-server -s project -e FASTMCP_LOG_LEVEL=$MCP_LOG_LEVEL -e AWS_PROFILE=$AWS_PROFILE -e AWS_REGION=$AWS_REGION -- uvx awslabs.cloudwatch-logs-mcp-server@latest

What it does: Queries logs with natural language and automatically identifies patterns.

Real impact: In my experience, diagnosing production issues takes far less time when an AI can read and summarize the logs for you.

9. 🔐 IAM MCP Server - The Security Guard

claude mcp add awslabs.iam-mcp-server -s project -e FASTMCP_LOG_LEVEL=$MCP_LOG_LEVEL -e AWS_PROFILE=$AWS_PROFILE -e AWS_REGION=$AWS_REGION -- uvx awslabs.iam-mcp-server@latest

What it does: Creates least-privilege IAM policies automatically, ensuring security without the complexity.

Real impact: Implement zero-trust security policies without becoming a security expert.

10. 🌐 Frontend MCP Server - The UI Builder

claude mcp add awslabs.frontend-mcp-server -s project -e FASTMCP_LOG_LEVEL=$MCP_LOG_LEVEL -e AWS_PROFILE=$AWS_PROFILE -e AWS_REGION=$AWS_REGION -- uvx awslabs.frontend-mcp-server@latest

What it does: Creates React applications with AWS Amplify integration using conversational commands.

Real impact: Prototype and deploy full-stack applications in a fraction of the usual time.

Where It Gets Interesting: Server Combinations

Things get more useful when you combine multiple MCP servers:

🎯 The Startup Stack

Core + Serverless + DynamoDB + Cost Analysis
= Complete serverless infrastructure with cost optimization

🏢 The Enterprise Suite

CDK + IAM + CloudWatch + EKS
= Secure, monitored, containerized infrastructure

🚀 The Rapid Prototype Pack

Frontend + Lambda + API Gateway + Nova Canvas
= Full-stack application with AI-generated assets

Real-World Impact: What Actually Changes

Less setup
Spinning up infrastructure
Fewer surprises
Cost flagged before you deploy
Faster
Feature deployment, start to finish

Getting Started in About 5 Minutes

Step 1: Install Claude Code (if you haven't already)

Download Claude Code from claude.ai/code for your operating system.

Step 2: Set up your AWS credentials

# Option 1: Use AWS CLI profile
export AWS_PROFILE=your-profile-name
export AWS_REGION=us-east-1
export MCP_LOG_LEVEL=ERROR
 
# Option 2: Use access keys (less secure)
export AWS_ACCESS_KEY_ID=your-key
export AWS_SECRET_ACCESS_KEY=your-secret
export AWS_REGION=us-east-1
export MCP_LOG_LEVEL=ERROR

Step 3: Add the Core MCP Server

# Add to your project (recommended - creates .mcp.json in project root)
claude mcp add awslabs.core-mcp-server -s project -e FASTMCP_LOG_LEVEL=$MCP_LOG_LEVEL -- uvx awslabs.core-mcp-server@latest
 
# Or add globally (for all projects)
claude mcp add awslabs.core-mcp-server -s global -e FASTMCP_LOG_LEVEL=$MCP_LOG_LEVEL -- uvx awslabs.core-mcp-server@latest

💡 Pro Tip: Use -s project to create a .mcp.json file in your project root. This makes your MCP server configuration:

  • Version controlled - Track changes in git
  • Shareable - Team members get the same setup
  • Visible - See all configured servers in one place
  • Project-specific - Different configs for different projects

Step 4: Let AI guide you

claude "Help me set up a serverless API with DynamoDB"

That's it. The AI will:

  • Suggest the right MCP servers
  • Install them automatically
  • Guide you through the entire setup
  • Write the code for you
  • Deploy everything

What Makes This Different?

Traditional AWS Development

  • Spend hours reading documentation
  • Write hundreds of lines of configuration
  • Debug permission issues for days
  • Worry about security best practices
  • Manually optimize for cost

With AWS MCP Servers

  • Describe what you want in plain English
  • AI generates secure, optimized code
  • Built-in best practices by default
  • Automatic cost optimization
  • Focus on business logic, not infrastructure

The Business Case: Payback You Can Feel Early

💰 Typical ROI Timeline

Week 1-2: Learning curve - your team gets familiar with the tools Week 3-4: First productivity gains - routine tasks get faster Month 2: Noticeable improvements - infrastructure work moves quicker Month 3+: Full adoption - team develops patterns and best practices What to expect: real time savings on routine infrastructure tasks, plus cost wins from catching better resource configurations earlier. How much depends on your stack and how well you know AWS.

Security First: Enterprise-Ready by Design

Every MCP server includes:

  • 🔐 Read-only modes for production safety
  • 🛡️ Built-in security scanning (CDK Nag, Checkov)
  • 📝 Audit trails for compliance
  • 🔑 IAM best practices enforced automatically
  • 🚦 Approval workflows for sensitive operations

FAQ: Your Questions Answered

Q: Do I need to be an AWS expert to use these?
A: You need basic AWS knowledge. MCP servers help with configuration and best practices, but you should understand what resources you're creating and their cost implications. It's like using a GPS—you still need to know how to drive.

Q: Is this replacing my DevOps team?
A: Absolutely not. It's a productivity tool that handles repetitive tasks so your team can focus on architecture, optimization, and strategic decisions. Think of it as augmentation, not replacement.

Q: What about vendor lock-in?
A: MCP servers generate standard AWS CloudFormation, CDK, or Terraform code. Everything is transparent, version-controlled, and modifiable. You can stop using MCP servers anytime and keep all your infrastructure code.

Q: Is it production-ready?
A: Yes, but start carefully. Use read-only modes, test in development environments first, and always review generated configurations. Many teams use it for development/staging and gradually adopt it for production with proper safeguards.

Q: What if the AI makes mistakes?
A: It can and sometimes will. That's why you need AWS knowledge to review outputs. MCP servers include safety features like dry-runs, cost estimates, and security scanning, but human oversight is essential.

Q: How much does it cost?
A: MCP servers themselves are free and open-source. You only pay for the AWS resources you create and any AI API usage (if using Claude). Most teams see net savings from optimized configurations.

A New Approach to AWS Development

The traditional approach of manually writing configurations and debugging policies isn't going away—it's evolving. Teams using MCP servers still write YAML and debug IAM policies, but they do it faster and with fewer errors.

This isn't about replacing skills or cutting corners. It's about using tools that respect your expertise while eliminating the tedious parts of AWS development.

The 45+ MCP servers represent a comprehensive toolkit for teams who want to focus on solving business problems rather than syntax problems.

Ready to Try It Out?

Get Started Today

See how MCP servers fit into your AWS development workflow.

Resources & Next Steps

📖 Read the Complete Technical Guide

My AWS MCP Servers implementation guide covers:

  • Detailed setup instructions for each category
  • Security best practices and production configurations
  • Real-world implementation patterns
  • Troubleshooting common issues

📥 A Quick Reference Is Coming

Want all 45+ MCP servers in a handy reference format? I'm putting together a PDF guide with:

  • Complete server directory with descriptions
  • Installation commands and configuration examples
  • Useful server combinations for common use cases
  • Troubleshooting checklist

Coming soon.


PS: In the earlier post I covered just a handful of these servers. This is the full picture.