mcpField
A
AWS MCP Server

AWS MCP Server

Local
github.com/RupeebwCloud & Infra0Updated 80w ago

About AWS MCP Server

AWS MCP Server connects to AWS accounts using credential and profile options similar to AWS CLI. It is for users who need to access S3, EC2, and Lambda operations through its REST APIs.

Key features

  1. 01Connect with AWS credentials
  2. 02Configure credentials by profile
  3. 03Manage S3 buckets and objects
  4. 04Inspect and control EC2 instances
  5. 05Manage and invoke Lambda functions
  6. 06Scan for exposed secrets

Use cases

  1. 01List or upload S3 objects
  2. 02Start or stop EC2 instances
  3. 03Inspect or invoke Lambda functions
  4. 04Check a codebase for potential secrets

MCP support

The project identifies itself as an MCP server, but the README describes REST API endpoints rather than specifying tools or other capabilities exposed to MCP clients.

Configuration

Local server config is the same for every client. Copy it into your client’s MCP config file.

No install method listed yet

We haven’t collected a package or remote endpoint for this server. Follow the instructions in the project README.

Repository

Tools

No tool list yet

Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.

Details

From the project README

AWS MCP Server

A server that connects to AWS accounts similar to AWS CLI.

Features

  • Connect to AWS accounts using credentials
  • Support for multiple AWS services
  • Credential management similar to AWS CLI
  • Profile-based configuration

Folder Structure

src/aws-mcp-server/
├── .env.example           # Example environment variables
├── .gitignore             # Git ignore file
├── README.md              # Documentation
├── package.json           # Project dependencies
└── src/                   # Source code
    ├── config/            # Configuration files
    │   └── aws-config.js  # AWS SDK configuration
    ├── core/              # Core functionality (empty for now)
    ├── index.js           # Main entry point
    ├── services/          # AWS service implementations
    │   ├── ec2.js         # EC2 service endpoints
    │   ├── lambda.js      # Lambda service endpoints
    │   └── s3.js          # S3 service endpoints
    └── utils/             # Utility functions
        ├── logger.js      # Logging utility
        └── session-manager.js # AWS session management

Setup

  1. Install dependencies:
npm install
  1. Configure AWS credentials:

    • Create a .env file based on .env.example
    • Or use AWS credentials file at ~/.aws/credentials
  2. Set up Git hooks to prevent committing secrets:

./scripts/setup-git-hooks.sh
  1. Start the server:
npm start

Configuration

The server supports multiple ways to configure AWS credentials:

  1. Environment variables
  2. AWS credentials file (~/.aws/credentials)
  3. AWS config file (~/.aws/config)
  4. Instance profiles (when running on EC2)
Temporary Credentials

If you're using temporary AWS credentials (Access Key ID starting with 'ASIA'), make sure to include the session token in your .env file:

AWS_ACCESS_KEY_ID=your_access_key
AWS_SECRET_ACCESS_KEY=your_secret_key
AWS_SESSION_TOKEN=your_session_token
AWS_REGION=your_region

Security

This project includes several security measures to prevent accidental exposure of credentials:

  1. .gitignore: Configured to exclude .env files, keys, certificates, and other sensitive files
  2. Git Hooks: Pre-commit hook to check for potential secrets in the codebase
  3. Secrets Checker: Script to scan for potential hardcoded secrets

To run the secrets check manually:

./scripts/check-secrets.sh

Usage

The server exposes REST APIs to interact with AWS services:

S3 Operations
  • GET /api/s3/buckets - List all S3 buckets
  • GET /api/s3/buckets/:bucket/objects - List objects in a bucket
  • POST /api/s3/buckets/:bucket/objects - Upload an object to a bucket
  • DELETE /api/s3/buckets/:bucket/objects/:key - Delete an object from a bucket
EC2 Operations
  • GET /api/ec2/instances - List all EC2 instances
  • GET /api/ec2/instances/:instanceId - Get EC2 instance details
  • POST /api/ec2/instances/:instanceId/start - Start an EC2 instance
  • POST /api/ec2/instances/:instanceId/stop - Stop an EC2 instance
Lambda Operations
  • GET /api/lambda/functions - List all Lambda functions
  • GET /api/lambda/functions/:functionName - Get Lambda function details
  • POST /api/lambda/functions/:functionName/invoke - Invoke a Lambda function
  • PATCH /api/lambda/functions/:functionName/configuration - Update Lambda function configuration
Ways to Interact with the AWS MCP Server:
  1. Using curl from the command line:

    # List S3 buckets
    curl http://localhost:3000/api/s3/buckets
    
    # List EC2 instances
    curl http://localhost:3000/api/ec2/instances
    
    # List Lambda functions
    curl http://localhost:3000/api/lambda/functions
    
  2. Using a REST client like Postman:

  3. Using a web browser (for GET requests only):

  4. Building a frontend application:

    • You can create a frontend application that makes API calls to this server
    • This would give you a GUI similar to the AWS Management Console
Troubleshooting AWS Credentials:

If you're having issues with AWS credentials, here are some options:

  1. For temporary credentials (Access Key starting with ASIA):

    • Make sure to include the AWS_SESSION_TOKEN in your .env file
    • These credentials typically expire after a few hours
  2. Use long-term credentials (Access Key starting with AKIA):

    • These don't require a session token
    • Be careful with these credentials and never commit them to version control
  3. Use AWS CLI profiles:

    • If you have AWS CLI configured, you can use a profile:
    AWS_PROFILE=your-profile-name
    
    • Remove the AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY from .env
  4. Use IAM roles if running on EC2:

    • If you deploy this to an EC2 instance with an IAM role, you don't need to specify credentials

Development

npm run dev

Testing

npm test

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.