AWS Model Context Protocol Server

AWS Model Context Protocol (MCP) Server

A lightweight service that enables AI assistants to execute AWS CLI commands through the Model Context Protocol (MCP).

Overview

The AWS MCP Server provides a bridge between MCP-aware AI assistants (like Claude Desktop, Cursor, Windsurf) and the AWS CLI. It enables these assistants to:

1. Retrieve AWS CLI documentation (aws_cli_help) - Get detailed help on AWS services and commands
2. Execute AWS CLI commands (aws_cli_pipeline) - Run commands with Unix pipes and receive formatted results optimized for AI consumption

Demo

Demo

The video demonstrates using Claude Desktop with AWS MCP Server to create a new AWS EC2 instance with AWS SSM agent installed.

Features

• Command Documentation - Detailed help information for AWS CLI commands
• Command Execution - Execute AWS CLI commands and return human-readable results
• Unix Pipe Support - Filter and transform AWS CLI output using standard Unix pipes and utilities
• AWS Resources Context - Access to AWS profiles, regions, account information, and environment details via MCP Resources
• Prompt Templates - Pre-defined prompt templates for common AWS tasks following best practices
• Docker Integration - Simple deployment through containerization with multi-architecture support (AMD64/x86_64 and ARM64)
• AWS Authentication - Leverages existing AWS credentials on the host machine

Requirements

• Docker (default) or Python 3.13+ (and AWS CLI installed locally)
• AWS credentials configured

Getting Started

Note: For security and reliability, running the server inside a Docker container is the strongly recommended method. Please review the Security Considerations section for important considerations.

Run Server Option 1: Using Docker (Recommended)

The Docker image supports both AMD64/x86_64 (Intel/AMD) and ARM64 (Apple Silicon M1-M4, AWS Graviton) architectures.

Note: The official image from GitHub Packages is multi-architecture and will automatically use the appropriate version for your system.

# Use the latest stable version docker pull ghcr.io/alexei-led/aws-mcp-server:latest # Or pin to a specific version (recommended for production) docker pull ghcr.io/alexei-led/aws-mcp-server:1.0.0

Docker Image Tags:

• latest: Latest stable release
• x.y.z (e.g., 1.0.0): Specific version
• sha-<commit-sha>: Development builds, tagged with Git commit SHA (e.g., sha-gb697684)

Run Server Option 2: Using Python

Use with Caution: Running natively requires careful environment setup and carries higher security risks compared to the recommended Docker deployment. Ensure you understand the implications outlined in the Security Considerations section.

Configuration

The AWS MCP Server can be configured using environment variables:

Important: Securely manage the AWS credentials provided to the server, whether via mounted ~/.aws files or environment variables. Ensure the credentials follow the principle of least privilege as detailed in the Security Considerations section. When running via Docker, ensure these variables are passed correctly to the container environment (e.g., using docker run -e VAR=value ...).

Security Considerations

Security is paramount when executing commands against your AWS environment. While AWS MCP Server provides functionality, you are responsible for configuring and running it securely. Please adhere strictly to the following:

1. Recommended Deployment: Docker Container

• Isolation: Running the server inside a Docker container is the strongly recommended and default deployment method. Containerization provides crucial filesystem and process isolation. Potentially destructive Unix commands (like rm, mv) executed via pipes, even if misused, will be contained within the ephemeral Docker environment and will not affect your host machine's filesystem. The container can be easily stopped and recreated.
• Controlled Environment: Docker ensures a consistent environment with necessary dependencies, reducing unexpected behavior.

2. AWS Credentials and IAM Least Privilege (Critical)

• User Responsibility: You provide the AWS credentials to the server (via mounted ~/.aws or environment variables).
• Least Privilege is Essential: The server executes AWS CLI commands using the credentials you provide. It is absolutely critical that these credentials belong to an IAM principal (User or Role) configured with the minimum necessary permissions (least privilege) for only the AWS actions you intend to perform through this tool.
  • Do Not Use Root Credentials: Never use AWS account root user credentials.
  • Regularly Review Permissions: Periodically audit the IAM permissions associated with the credentials.
• Impact Limitation: Properly configured IAM permissions are the primary mechanism for limiting the potential impact of any command executed via the server, whether intended or unintended. Even if a command were manipulated, it could only perform actions allowed by the specific IAM policy.

3. Trusted User Model

• The server assumes the end-user interacting with the MCP client (e.g., Claude Desktop, Cursor) is the same trusted individual who configured the server and provided the least-privilege AWS credentials. Do not expose the server or connected client to untrusted users.

4. Understanding Execution Risks (Current Implementation)

• Command Execution: The current implementation uses shell features (shell=True in subprocess calls) to execute AWS commands and handle Unix pipes. While convenient, this approach carries inherent risks if the input command string were manipulated (command injection).
• Mitigation via Operational Controls: In the context of the trusted user model and Docker deployment, these risks are mitigated operationally:
  • The trusted user is assumed not to provide intentionally malicious commands against their own environment.
  • Docker contains filesystem side-effects.
  • Crucially, IAM least privilege limits the scope of any AWS action that could be executed.
• Credential Exfiltration Risk: Despite containerization and IAM, a sophisticated command injection could potentially attempt to read the mounted credentials (~/.aws) or environment variables within the container and exfiltrate them (e.g., via curl). Strict IAM policies remain the most vital defense to limit the value of potentially exfiltrated credentials.

5. Network Exposure (SSE Transport)

• If using the sse transport (which implies a network listener), ensure you bind the server only to trusted network interfaces (e.g., localhost) or implement appropriate network security controls (firewalls, authentication proxies) if exposing it more broadly. The default stdio transport does not open network ports.

6. Shared Responsibility Summary

• AWS MCP Server provides the tool.
• You, the user, are responsible for:
  • Running it within the recommended secure Docker environment.
  • Providing and securely managing least-privilege AWS credentials.
  • Ensuring only trusted users interact with the server/client.
  • Securing the network environment if applicable.

By strictly adhering to Docker deployment and meticulous IAM least-privilege configuration, you establish the necessary operational controls for using the AWS MCP Server securely with its current implementation.

Integrating with Claude Desktop

Configuration

To manually integrate AWS MCP Server with Claude Desktop:

1. Locate the Claude Desktop configuration file:
   • macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
   • Windows: %APPDATA%\Claude\claude_desktop_config.json
2. Edit the configuration file to include the AWS MCP Server:
   { "mcpServers": { "aws-mcp-server": { "command": "docker", "args": [ "run", "-i", "--rm", "-v", "/Users/YOUR_USER_NAME/.aws:/home/appuser/.aws:ro", "ghcr.io/alexei-led/aws-mcp-server:latest" ] } } }
3. Restart Claude Desktop to apply the changes
   • After restarting, you should see a hammer 🔨 icon in the bottom right corner of the input box
   • This indicates that the AWS MCP Server is available for use

Example Interactions

Getting AWS CLI Documentation:

Executing AWS CLI Commands:

Using Command Pipes:

Accessing AWS Resources:

Using Prompt Templates:

Available Prompt Templates

The AWS MCP Server includes the following pre-defined prompt templates:

Core Operations

Security & Compliance

Cost & Performance

Infrastructure & Architecture

Reliability & Monitoring

Security

The AWS MCP Server implements a comprehensive multi-layered approach to command validation and security:

Command Validation System

The server validates all AWS CLI commands through a three-layer system:

1. Basic Command Structure:
   • Verifies commands start with 'aws' prefix and contain a valid service
   • Ensures proper command syntax
2. Security-Focused Command Filtering:
   • Dangerous Commands: Blocks commands that could compromise security
   • Safe Patterns: Explicitly allows read-only operations needed for normal use
   • Regex Pattern Matching: Prevents complex security risks with pattern matching
3. Pipe Command Security:
   • Validates Unix commands used in pipes
   • Restricts commands to a safe allowlist
   • Prevents filesystem manipulation and arbitrary command execution

Default Security Configuration

The default security configuration focuses on preventing the following attack vectors:

1. Identity and Access Management (IAM) Risks

2. Audit and Logging Tampering

3. Sensitive Data Access and Protection

4. Network Security Risks

Many read-only operations that match these patterns are explicitly allowed via safe patterns:

• All get-, list-, and describe- commands
• All help commands (--help, help)
• Simulation and testing commands (e.g., aws iam simulate-custom-policy)

Configuration Options

• Security Modes:
  • strict (default): Enforces all security validations
  • permissive: Logs warnings but allows execution (use with caution)
• Custom Configuration:
  • Override default security rules via YAML configuration file
  • Configure service-specific dangerous commands
  • Define custom safe patterns and regex rules
  • Environment variable: AWS_MCP_SECURITY_CONFIG
• Execution Controls:
  • Timeouts prevent long-running commands (default: 300 seconds)
  • Output size limits prevent memory issues
  • Environment variables: AWS_MCP_TIMEOUT, AWS_MCP_MAX_OUTPUT

Custom Security Rules Example

You can create custom security rules by defining a YAML configuration file:

Security Examples

The system follows IAM best practices, focusing on preventing escalation of privilege:

Security Best Practices

• Always use the default strict security mode in production
• Follow the deployment recommendations in Security Considerations
• Run with least-privilege AWS credentials
• For custom configurations, focus on your security requirements

Development

Setting Up the Development Environment

Makefile Commands

The project includes a Makefile with various targets for common tasks:

For a complete list of available commands, run make help.

Code Coverage

The project includes configuration for Codecov to track code coverage metrics. The configuration is in the codecov.yml file, which:

• Sets a target coverage threshold of 80%
• Excludes test files, setup files, and documentation from coverage reports
• Configures PR comments and status checks

Coverage reports are automatically generated during CI/CD runs and uploaded to Codecov.

Integration Testing

Integration tests verify AWS MCP Server works correctly with actual AWS resources. To run them:

1. Set up AWS resources:
   • Create an S3 bucket for testing
   • Set the environment variable: export AWS_TEST_BUCKET=your-test-bucket-name
   • Ensure your AWS credentials are configured
2. Run integration tests:
   # Run all tests including integration tests make test-all # Run only integration tests make test-integration

Or you can run the pytest commands directly:

Troubleshooting

• Authentication Issues: Ensure your AWS credentials are properly configured
• Connection Errors: Verify the server is running and AI assistant connection settings are correct
• Permission Errors: Check that your AWS credentials have the necessary permissions
• Timeout Errors: For long-running commands, increase the AWS_MCP_TIMEOUT environment variable

Why Deploy with Docker

Deploying AWS MCP Server via Docker is the recommended approach, offering significant security and reliability advantages that form the core of the tool's secure usage pattern:

Security Benefits

• Isolation (Primary Mitigation): The Docker container provides essential filesystem and process isolation. AWS CLI commands and piped Unix utilities run in a contained environment. Accidental or misused commands affecting the filesystem are limited to the container, protecting your host machine.
• Controlled Credential Access: When mounting credentials, using the :ro (read-only) flag limits the container's ability to modify your AWS configuration files.
• No Local Installation: Avoids installing the AWS CLI and its dependencies directly on your host system.
• Clean Environment: Each container run starts with a known, clean state.

Reliability Advantages

• Consistent Configuration: All required tools (AWS CLI, SSM plugin, jq) are pre-installed and properly configured
• Dependency Management: Avoid version conflicts between tools and dependencies
• Cross-Platform Consistency: Works the same way across different operating systems
• Complete Environment: Includes all necessary tools for command pipes, filtering, and formatting

Other Benefits

• Multi-Architecture Support: Runs on both Intel/AMD (x86_64) and ARM (Apple Silicon, AWS Graviton) processors
• Simple Updates: Update to new versions with a single pull command
• No Python Environment Conflicts: Avoids potential conflicts with other Python applications on your system
• Version Pinning: Easily pin to specific versions for stability in production environments

Versioning

This project uses setuptools_scm to automatically determine versions based on Git tags:

• Release versions: When a Git tag exists (e.g., 1.2.3), the version will be exactly that tag
• Development versions: For commits without tags, a development version is generated in the format: <last-tag>.post<commits-since-tag>+g<commit-hash>.d<date> (e.g., 1.2.3.post10+gb697684.d20250406)

The version is automatically included in:

• Package version information
• Docker image labels
• Continuous integration builds

Creating Releases

To create a new release version:

The CI/CD pipeline will automatically build and publish Docker images with appropriate version tags.

For more detailed information about the version management system, see VERSION.md.

License

This project is licensed under the MIT License - see the LICENSE file for details.