June 12, 2026 Guides

AWS MCP Servers: A Complete Guide

Comprehensive guide to setting up and using AWS-powered MCP servers with Bedrock, S3, DynamoDB, and more.

By MCPConfig Team

AWSMCPBedrockS3DynamoDBcloud

Amazon Web Services offers powerful MCP integrations that let AI assistants interact with your AWS infrastructure. This guide covers the most popular AWS MCP servers and how to configure them.

Overview

AWS MCP servers bridge the gap between AI clients and AWS services. They enable natural-language interactions with:

Amazon Bedrock — Invoke and manage foundation models
Amazon S3 — Read, write, and manage storage buckets
Amazon DynamoDB — Query and manage NoSQL databases
AWS Lambda — Invoke serverless functions
Amazon CloudWatch — Monitor logs and metrics

Prerequisites

An AWS account with appropriate IAM permissions
AWS CLI configured with credentials (aws configure)
Node.js 18+ for running MCP servers
An MCP-compatible client (Claude Desktop, Cursor, Windsurf)

Setting Up AWS Credentials

AWS MCP servers use the standard AWS credential chain. Configure your credentials before starting:

aws configure
# Enter your AWS Access Key ID
# Enter your AWS Secret Access Key
# Enter default region (e.g., us-east-1)
# Enter default output format (json)

Alternatively, set environment variables:

export AWS_ACCESS_KEY_ID=AKIA...
export AWS_SECRET_ACCESS_KEY=wJal...
export AWS_REGION=us-east-1

Amazon Bedrock MCP Server

The Bedrock MCP server lets AI clients invoke foundation models directly.

Installation

npx @modelcontextprotocol/server-aws-bedrock

Configuration

Add to your MCP client config:

{
  "mcpServers": {
    "aws-bedrock": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-aws-bedrock"],
      "env": {
        "AWS_REGION": "us-east-1"
      }
    }
  }
}

Available Tools

Tool	Description
`invoke_model`	Invoke a Bedrock foundation model
`list_models`	List available Bedrock models
`list_model_versions`	Show versions of a specific model

Example Usage

Ask your AI client: “Use Bedrock to generate a summary of this document using Claude 3 Sonnet.” The client will use the invoke_model tool with the appropriate parameters.

Amazon S3 MCP Server

Interact with S3 buckets and objects through natural language.

Installation

npx @modelcontextprotocol/server-aws-s3

Configuration

{
  "mcpServers": {
    "aws-s3": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-aws-s3"],
      "env": {
        "AWS_REGION": "us-east-1"
      }
    }
  }
}

Available Tools

Tool	Description
`list_buckets`	List all S3 buckets in the account
`list_objects`	List objects in a bucket (with optional prefix filter)
`get_object`	Read the contents of an S3 object
`put_object`	Upload content to an S3 bucket
`delete_object`	Delete an object from S3

Available Resources

Resource URI	Description
`s3://{bucket}/`	Browse objects in a bucket
`s3://{bucket}/{key}`	Read object contents

Amazon DynamoDB MCP Server

Query and manage DynamoDB tables through MCP.

Installation

npx @modelcontextprotocol/server-aws-dynamodb

Configuration

{
  "mcpServers": {
    "aws-dynamodb": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-aws-dynamodb"],
      "env": {
        "AWS_REGION": "us-east-1"
      }
    }
  }
}

Available Tools

Tool	Description
`list_tables`	List all DynamoDB tables
`describe_table`	Get table schema and metadata
`get_item`	Retrieve an item by key
`query`	Query a table with key condition expression
`scan`	Scan a table (with optional filter)
`put_item`	Insert or update an item
`delete_item`	Delete an item by key

AWS Lambda MCP Server

Invoke and manage Lambda functions through MCP.

Installation

npx @modelcontextprotocol/server-aws-lambda

Configuration

{
  "mcpServers": {
    "aws-lambda": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-aws-lambda"],
      "env": {
        "AWS_REGION": "us-east-1"
      }
    }
  }
}

Available Tools

Tool	Description
`list_functions`	List all Lambda functions
`invoke_function`	Invoke a Lambda function with a payload
`get_function_config`	Get function configuration details
`list_function_versions`	Show versions of a function

Running Multiple AWS Servers

You can run multiple AWS MCP servers in a single config:

{
  "mcpServers": {
    "aws-s3": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-aws-s3"],
      "env": { "AWS_REGION": "us-east-1" }
    },
    "aws-dynamodb": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-aws-dynamodb"],
      "env": { "AWS_REGION": "us-west-2" }
    },
    "aws-lambda": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-aws-lambda"],
      "env": { "AWS_REGION": "eu-west-1" }
    }
  }
}

Each server can target a different region — perfect for multi-region workflows.

Security Best Practices

IAM Permissions

Create dedicated IAM policies with least-privilege permissions. For example, a read-only S3 policy:

{
  "Version": "2012-10-17",
  "Statement": [
    {
      "Effect": "Allow",
      "Action": [
        "s3:ListAllMyBuckets",
        "s3:ListBucket",
        "s3:GetObject"
      ],
      "Resource": "*"
    }
  ]
}

Use IAM Roles

For production, use IAM roles instead of long-lived access keys. On EC2 or ECS, attach the role to the instance. On Lambda, use the function’s execution role.

Region Locking

Restrict each server to a specific region to limit blast radius:

{
  "env": {
    "AWS_REGION": "us-east-1"
  }
}

Troubleshooting

Issue	Fix
`AccessDenied` errors	Check IAM permissions for the tool or resource
`Credentials not found`	Verify `aws configure` or env vars are set
Region mismatch	Some services require specific regions — verify with `aws --region`
Timeout on large S3 objects	The server streams content; large files may take time

Next Steps

Use the MCPConfig Builder to generate your AWS MCP configuration visually
Browse the server templates for pre-built AWS server configs
Check the server directory for all available MCP servers

← Back to Blog Open Config Builder