aws

This AWS MCP server provides a Model Context Protocol (MCP) interface that exposes a wide range of AWS services for use by GenAI applications. It supports EC2, S3, Lambda, DynamoDB, RDS, CloudFormation, IAM, CloudWatch, SQS/SNS, and container services (ECS/EKS), with authentication through IAM keys, IAM roles, AWS SSO, and STS temporary credentials. Use the included Python package (aws_mcp) to initialize an AWSMCPServer and start serving MCP requests. The server exposes a collection of tools (e.g., aws_ec2_list_instances, aws_s3_upload_object, aws_lambda_invoke, aws_dynamodb_query, aws_cloudformation_create_stack, etc.). You can drive these tools via the MCP interface by sending JSON requests that specify the tool name and arguments, enabling programmatic control of your AWS account from GenAI workflows. To configure access, set environment variables (AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY, AWS_DEFAULT_REGION, and optional role/session information). The Quick Start example demonstrates starting the server, and the Claude Desktop configuration shows how to expose the AWS MCP server under the mcpServers.aws entry for seamless integration. Explore the Available Tools sections in the README to find the specific tool names and argument schemas for your use cases, such as listing EC2 instances, creating S3 objects, invoking Lambda functions, or deploying CloudFormation stacks.

Prerequisites:

• Python 3.8+ installed on the host
• Git installed
• Network access to AWS IAM credentials (or IAM role permissions if using roles)

Install from PyPI:

pip install aws-mcp-server

Or install from source:

git clone https://github.com/LokiMCPUniverse/aws-mcp-server.git
cd aws-mcp-server
pip install -e .

Run the server (example):

export AWS_ACCESS_KEY_ID=your_access_key
export AWS_SECRET_ACCESS_KEY=your_secret_key
export AWS_DEFAULT_REGION=us-east-1
python -m aws_mcp.server

Alternatively, use the Claude Desktop configuration path shown in the README to expose the MCP server under your Claude setup:

{
  "mcpServers": {
    "aws": {
      "command": "python",
      "args": ["-m", "aws_mcp.server"],
      "env": {
        "AWS_ACCESS_KEY_ID": "your_access_key",
        "AWS_SECRET_ACCESS_KEY": "your_secret_key",
        "AWS_DEFAULT_REGION": "us-east-1"
      }
    }
  }
}

Environment variable configuration is essential for AWS access. Consider using IAM roles or temporary credentials (STS) for enhanced security in production. If you enable multi-account or cross-region usage, ensure proper permission scopes for each account/region. The AWS MCP server supports MFA, session tokens, and profile-based configurations; use AWS_PROFILE to switch between named profiles. For production deployments, monitor retries and timeouts using AWS_MAX_RETRIES and AWS_TIMEOUT to balance reliability and latency. If you encounter authentication errors, verify that the environment variables are correctly exported in the runtime environment and that the credentials have the necessary permissions for the requested AWS operations.