awslabs/mcp

What it solves

This project provides a collection of Model Context Protocol (MCP) servers that bridge the gap between Large Language Models (LLMs) and the AWS ecosystem. It solves the problem of LLMs having outdated knowledge of AWS services, lacking access to real-time documentation, and being unable to perform actual cloud infrastructure tasks or workflow automation within a development environment.

How it works

It implements the Model Context Protocol, an open standard that allows AI applications (like Cursor, Windsurf, or Claude Desktop) to connect to specialized servers. These servers act as intermediaries that can fetch the latest AWS documentation, API references, and best practices, or execute specific AWS operations. The servers in this repository primarily use the stdio transport mechanism to communicate with the host AI application.

Who it’s for

This is designed for cloud developers, DevOps engineers, and AI architects who use agentic AI coding assistants or chatbots to build, deploy, and manage AWS infrastructure and applications.

Highlights

• Real-time Documentation: Provides servers for immediate access to official AWS docs, API references, and Well-Architected guidance.
• Infrastructure as Code (IaC): Includes a toolkit for CloudFormation and CDK, offering best practices and security validation.
• Service-Specific Servers: Dedicated servers for various AWS domains including EKS (Kubernetes), AI/ML, Data & Analytics, and Healthcare.
• Broad Client Support: Compatible with popular AI IDEs and tools such as Cursor, VS Code, Windsurf, and Claude Code.