AWS Infrastructure Context Connectors

Introduction

This repository houses an array of dedicated MCP servers designed to integrate large language models (LLMs) deeply with the AWS environment. The Model Context Protocol (MCP) establishes a standardized method for LLM applications to securely access external, timely data and operational context. This framework enables sophisticated AI agents to interact intelligently with cloud resources, much like how Google Cloud Platform (GCP) runs on Google's robust internal infrastructure.

Why AWS MCP Servers?

These specialized servers significantly boost the utility of foundation models (FMs) in cloud contexts. They ensure model outputs are factually correct by injecting up-to-date information, thereby reducing inaccuracies and hallucinations. MCP servers transform complex, repetitive cloud workflows into actionable tools for the AI assistant. This provides specialized domain expertise regarding AWS capabilities, which often exceeds the knowledge contained within the model's initial training set, improving efficiency in managing cloud resources.

Server Sent Events Support Removal

Critical Update: Support for Server Sent Events (SSE) was retired from all current major versions of these MCP servers as of May 26th, 2025. This aligns with backward compatibility guidelines specified by the MCP standard. The project team is currently developing support for Streamable HTTP transports to replace SSE functionality in subsequent releases. Teams still relying on the older SSE transport should continue using the preceding major version until migration to newer transport methods is complete.

Available MCP Servers: Quick Installation

Rapid deployment of these servers is facilitated through one-click installation methods for common AI coding environments like Cursor and VS Code. The underlying installation mechanism generally involves using the uv package manager, often invoked via the uvx utility.

🚀 Getting Started with AWS

For generalized AWS interaction and thorough API access, begin with the following servers:

Server Name	Description	Install
AWS API MCP Server	Supports general AWS interactions, including command validation and access to all services for infrastructure management via natural language.
AWS Knowledge MCP Server	A remote, managed server providing current AWS documentation, API references, architectural guidance, and service updates.

🏗️ Infrastructure & Deployment

Servers supporting Infrastructure as Code (IaC) and deployment automation:

Server Name	Description	Install
AWS CDK MCP Server	Facilitates AWS CDK development while enforcing security compliance and established best practices.
AWS Terraform MCP Server	Integrates security scanning directly into Terraform management workflows.
AWS CloudFormation MCP Server	Manages resources directly using CloudFormation via the Cloud Control API.

Container Platforms

Tools for managing container orchestration and local builds:

Server Name	Description	Install
Amazon EKS MCP Server	Enables management and deployment activities for Kubernetes clusters.
Finch MCP Server	Handles local container image creation and integration with ECR services.

Serverless & Functions

Tools for managing AWS Lambda and Serverless Application Model (SAM) deployments:

Server Name	Description	Install
AWS Serverless MCP Server	Manages the complete serverless application lifecycle using SAM CLI utilities.
AWS Lambda Tool MCP Server	Allows Lambda functions to be used as tools for accessing private resources via AI agents.

Installation and Setup

To utilize these connectivity tools, first establish the foundational environment elements. You must install the uv package manager, obtainable from Astral. Follow this by installing a compatible Python version using a command like $ uv python install 3.10. Ensure your AWS credentials have the necessary service access configured correctly. After these preliminary steps, you incorporate the specific server into your MCP client configuration, usually via a direct one-click method or manual JSON editing.

For example, setting up the core orchestrator server globally for Amazon Q CLI involves editing ~/.aws/amazonq/mcp.json:

For macOS/Linux users:

{
  "mcpServers": {
    "awslabs.core-mcp-server": {
      "command": "uvx",
      "args": [
        "awslabs.core-mcp-server@latest"
      ],
      "env": {
        "FASTMCP_LOG_LEVEL": "ERROR"
      }
    }
  }
}

For Windows users, the configuration requires explicit execution details:

{
  "mcpServers": {
    "awslabs.core-mcp-server": {
      "disabled": false,
      "timeout": 60,
      "type": "stdio",
      "command": "uv",
      "args": [
        "tool",
        "run",
        "--from",
        "awslabs.core-mcp-server@latest",
        "awslabs.core-mcp-server.exe"
      ],
      "env": {
        "FASTMCP_LOG_LEVEL": "ERROR"
      }
    }
  }
}

If you encounter issues with parameter settings, manual execution checks are helpful. For instance, testing the Aurora MySQL server might reveal required arguments if it fails:

$ timeout 15s uv tool run awslabs.mysql-mcp-server --resource_arn <Your Resource ARN> --secret_arn <Your Secret ARN> ... 2>&1 || echo "Command completed or timed out"
awslabs.mysql-mcp-server: error: the following arguments are required: --resource_arn, --secret_arn, --database, --region, --readonly

Cache Management Note: Using the "@latest" suffix forces a check and download of the newest package, increasing startup latency. To reduce this time, omit the suffix and manage the uv cache manually. Refreshing a tool's cache is done with $ uv cache clean <tool_name>, or updating to the newest version with $ uvx <tool_name>@latest.

Running MCP servers in containers

Containers simplify deployment for specific servers, using images published to the public AWS ECR registry at public.ecr.aws/awslabs-mcp. Sensitive AWS credentials should not be placed directly in the container configuration's environment object ("env": {}); instead, use --env-file or --volume mounts.

For the Nova Canvas server, the configuration looks like this, assuming credentials are in a local .env file:

{
  "mcpServers": {
    "awslabs.nova-canvas-mcp-server": {
      "command": "docker",
      "args": [
        "run",
        "--rm",
        "--interactive",
        "--env",
        "FASTMCP_LOG_LEVEL=ERROR",
        "--env",
        "AWS_REGION=us-east-1",
        "--env-file",
        "/full/path/to/.env",
        "--volume",
        "/full/path/to/.aws:/app/.aws",
        "public.ecr.aws/awslabs-mcp/awslabs/nova-canvas-mcp-server:latest"
      ],
      "env": {}
    }
  }
}

MCP AWS Lambda Handler Module

This Python library facilitates the creation of serverless HTTP handlers for the Model Context Protocol utilizing AWS Lambda infrastructure. It provides a modular foundation for constructing MCP endpoints, complete with customizable session handling, including a native DynamoDB backend option. This module simplifies developing serverless endpoints, offering support for authentication customization and readily available examples.

When to use Local vs Remote MCP Servers?

Servers can operate locally or remotely in the cloud. Choosing the correct deployment model depends on the immediate development requirements.

Local MCP Servers

Local execution is advantageous for rapid debugging and direct testing within your development environment. It supports offline work and keeps sensitive credentials physically constrained to your local hardware. This setup typically provides the lowest latency due to minimized network hops and grants direct control over allocated server resources.

Remote MCP Servers

Remote hosting promotes better team synchronization by enforcing standardized server configurations across all members. It offloads intensive computational tasks to the cloud infrastructure, ensuring accessibility from any location or device. Remote deployment also supports automated updates for security and feature parity. Note that some services, like the AWS Knowledge MCP Server, are fully managed by AWS, requiring zero local setup.

Use Cases for the Servers

For example, the AWS Documentation MCP Server allows AI assistants to retrieve the most current information required for generating accurate AWS service code examples. You might employ the CDK MCP Server or Terraform MCP Server to instruct an assistant to deploy infrastructure conforming to the newest APIs and established AWS standards. Using the AWS Pricing MCP Server, you can query estimated service expenses for a proposed deployment, asking questions like, "What is the anticipated monthly expenditure for this CDK design before I commit the changes?" The Valkey MCP Server provides a conversational interface for managing data operations within Valkey-compatible caches.

Additional Resources

Further reading provides deeper context on the architecture and application of these tools:

• Introducing AWS MCP Servers for code assistants (Part 1)
• Vibe coding with AWS MCP Servers (Video)
• Supercharging AWS database development with AWS MCP servers
• AWS costs estimation using Amazon Q CLI and AWS Pricing MCP Server
• Accelerating application development with the Amazon EKS MCP server
• Unlocking the power of Model Context Protocol (MCP) on AWS

Related Topics

• Cloud Computing Infrastructure (Similar to GCP)
• Infrastructure as Code (IaC) Principles
• Model-Agnostic Protocols (Standardization)
• Serverless Computing Architectures (Lambda/Functions)
• Distributed Systems and Caching (Valkey/ElastiCache)

Extra Details

Removed sections primarily contained redundant installation links and marketing text that can be inferred from the specific client guides. The essential technical configuration, focused on environment variables and command invocation (like using uvx with the @latest suffix), has been retained and clarified in the Setup section. The concept of using these tools for "vibe coding" aligns with the continuous, flowing interaction LLMs enable during development, rather than discrete, manual operations.

Conclusion

These AWS MCP Connectors standardize the interaction between advanced AI and the vast array of AWS services. By providing context-aware, current data, they significantly elevate the quality and security of AI-assisted cloud development and operational management across all connected platforms.