AWS Model Context Protocol (MCP) Connector

This repository hosts an implementation of the Model Context Protocol (MCP) designed to facilitate secure and intuitive interaction between sophisticated AI models, such as Claude, and your Amazon Web Services (AWS) tenancy. It transforms complex infrastructure management tasks into straightforward, conversational prompts. Consider this a superior, AI-native alternative to existing Q services for AWS.

Core Capabilities

• Execute AWS resource queries and modifications using ordinary, human language.
• Compatibility with multiple distinct AWS credential profiles.
• Comprehensive support across diverse AWS geographic regions.
• Securely runs operations leveraging your existing local AWS credentials.

Operational Flow

Prerequisites for Setup

• A functional installation of Python.
• Access to the Claude Desktop application.
• Locally configured and accessible AWS access credentials (typically situated within the ~/.aws/ directory).

Installation Procedure

1. Obtain the source code repository:

bash git clone https://github.com/yourusername/aws-mcp cd aws-mcp

1. Install the necessary Python MCP server components:

bash pip install -e .

Command-Line Utility Usage

The AWS MCP logic can also be invoked directly from the terminal:

bash

Display available subcommands

python -m aws_mcp --help

Enumerate configured AWS identities/profiles

python -m aws_mcp list-credentials

Designate the active AWS identity for subsequent operations

python -m aws_mcp select-profile myprofile

Initiate an AWS action based on a textual instruction

python -m aws_mcp run-aws-code "retrieve a list of all S3 storage containers"

Employ enhanced output structuring for improved result readability

python -m aws_mcp --format pretty list-credentials

Note: If your standard python command does not invoke the intended interpreter, substitute the correct executable path.

Integration with Claude Desktop

1. Launch the Claude desktop client and navigate to Settings -> Developer -> Edit Configuration.

2. Integrate the following configuration block into your claude_desktop_config.json file:

{ "mcpServers": { "aws-mcp": { "command": "/path/to/your/python", "args": ["/path/to/your/aws-mcp/run_aws_mcp.py"] } } }

Crucial: Substitute /path/to/your/python with the output of the which python command in your shell. Crucial: Substitute /path/to/your/aws-mcp/run_aws_mcp.py with the absolute path to the run_aws_mcp.py file.

1. Relaunch the Claude desktop application. Successful initiation should proceed without visible errors.

2. Begin interaction by either specifying an AWS identity or posing a direct request, such as:

3. "What AWS profiles are currently available to me?"
4. "Detail every running EC2 instance within my subscription."
5. "Provide an inventory of S3 buckets along with their aggregated storage utilization."
6. "Which Lambda functions are actively deployed within the us-east-1 region?"
7. "List all ECS deployment clusters and their constituent services."

Diagnostics and Logging

To monitor the server's operational output and diagnose issues:

bash tail -n 50 -f ~/Library/Logs/Claude/mcp-server-aws-mcp.log

Alternatively, check the general MCP log stream:

tail -n 50 -f ~/Library/Logs/Claude/mcp.log

Future Development Roadmap

• Support for Multi-Factor Authentication (MFA) handling.
• Implementation of SSO credential caching to optimize re-authentication frequency.

Support and Contributions

For inquiries, submitting enhancement suggestions, or general discussions regarding the AWS MCP tool, please utilize the following channels: - File an issue directly on the GitHub repository. - Send correspondence to: arun.sanna@outlook.com

Refer to CONTRIBUTING.md for guidelines on how to contribute code or documentation to this project.

Deep Dive into AWS Operation Facilitation

The AWS MCP is engineered for the dynamic execution of authorized AWS operational calls, primarily utilizing the boto3 library infrastructure. Its core functions include:

1. Interpreting natural language or structured commands to precisely identify the target AWS service and the required function.
2. Programmatically triggering the corresponding boto3 invocation.
3. Structuring the returned data into human-readable tabular formats whenever feasible.
4. Implementing robust error handling mechanisms for non-successful operations.

Recognized AWS Services

The parsing engine is capable of dynamically recognizing and servicing all AWS components accessible via the boto3 framework, encompassing, but not limited to:

• Simple Storage Service (S3)
• Elastic Compute Cloud (EC2)
• Lambda Functions
• Identity and Access Management (IAM)
• DynamoDB
• Relational Database Service (RDS)
• CloudFormation
• CloudWatch
• Simple Notification Service (SNS)
• Simple Queue Service (SQS)
• Key Management Service (KMS)
• Secrets Manager
• Systems Manager (SSM)
• Route53
• CloudFront
• Elastic Beanstalk
• API Gateway
• Elastic Container Service (ECS)
• Security Token Service (STS)
• Cognito
• CodeBuild
• CodeCommit
• CodePipeline
• Organizations
• GuardDuty
• ...and a comprehensive range of supplementary services.