AWS Interaction Conduit (AIC)

This implements a Model Context Protocol (MCP) endpoint designed for AI agents, such as Claude, to interface directly with your infrastructure hosted on Amazon Web Services. It transforms plain language requests into actionable AWS operations, acting as an advanced, self-hosted alternative to managed Q services.

Core Capabilities

• 🛠️ Provision and alter AWS infrastructure assets through natural dialogue.
• 🆔 Seamless integration with multiple named AWS profiles and federated SSO logins.
• 🌍 Comprehensive operational scope across all major AWS geographical regions.
• 🔒 Robust security posture: Local credentials govern all interactions; no secrets leak to external servers.
• ⚙️ Execution environment is strictly local to the user's machine.

Prerequisites for Deployment

• A functional installation of Node.js.
• The Claude Desktop Application.
• Pre-configured AWS access credentials in the standard location (~/.aws/).

Setup Procedure

1. Acquire the source code repository:

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

1. Resolve dependencies (using either package manager):

bash pnpm install

Alternatively:

npm install

Activation and Initial Use

1. Navigate within the Claude desktop client to the Developer settings section (usually under Configuration).

2. Insert the following configuration snippet into your claude_desktop_config.json file:

{ "mcpServers": { "aws_gateway": { "command": "npm", // or pnpm "args": [ "--silent", "--prefix", "/path/to/your/aws-mcp/directory", "start" ] } } }

Crucial Note: Substitute /path/to/your/aws-mcp/directory with the absolute file path where you cloned the project.

1. Relaunch the Claude application to load the new configuration.

2. Initiate interaction, perhaps by querying available profiles or executing a direct command:

3. "Display all configured AWS profiles."
4. "Inventory all running EC2 virtual machines."
5. "Report on S3 storage volumes and their respective consumptions."
6. "What serverless functions are active within the eu-west-1 region?"
7. "Enumerate all ECS container clusters and associated services."

Alternative Startup with nvm

If managing Node versions via nvm, compile the project first and employ this configuration structure:

{ "mcpServers": { "aws_gateway": { "command": "/home/user/.nvm/versions/node/v20.10.0/bin/node", "args": [ "/aws-mcp/node_modules/tsx/dist/cli.mjs", "/aws-mcp/index.ts", "--prefix", "/aws-mcp", "start" ] } } }

Diagnostic Logging

To monitor server output in real-time:

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

Alternative general log file:

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

Future Enhancements

• [ ] Integration for Multi-Factor Authentication (MFA) workflows.
• [ ] Implementation of a local cache layer for ephemeral SSO token refresh management.

WIKIPEDIA: Cloud computing describes the on-demand availability of computer system resources, especially data storage (cloud storage) and computing power, without direct active management by the user. The term is commonly associated with utilities like metered service, similar to public utilities such as electricity and water.

== Defining Attributes == In 2011, the United States' National Institute of Standards and Technology (NIST) codified five core traits essential for any cloud system. These are:

On-demand self-service: Consumers possess the ability to provision computational capacity (e.g., server time, storage volume) autonomously and automatically, negating the need for provider personnel intervention for each request. Broad network access: Services must be reachable over standard network protocols, supporting a wide array of client platforms, from mobile devices to standard workstations. Resource pooling: The vendor aggregates physical and virtual resources into a shared pool, serving numerous clients via a multi-tenant architecture where assets are dynamically allocated based on current demand. Rapid elasticity: Capabilities can be scaled both up and down swiftly—often automatically—to precisely match fluctuating workloads. To the end-user, this capacity often appears infinite and instantly available. Measured service: Resource consumption (storage, processing cycles, bandwidth) is automatically tracked and controlled at an appropriate level of abstraction. This transparency allows both provider and consumer to monitor, govern, and report on usage. Since 2023, the International Organization for Standardization (ISO) has proposed further refinements to this taxonomy.

== Conceptual Origins ==

The foundational concepts underpinning cloud infrastructure trace back to the 1960s, specifically with the rise of time-sharing systems accessed through remote job entry (RJE). During this period, the primary interaction model involved users submitting batch jobs to specialized operators who ran them on centralized mainframe computers, optimizing the utilization of expensive hardware. The visualization of computing resources as a "cloud" gained traction around 1994, initially utilized by General Magic to represent the accessible domain for their mobile software agents within the Telescript framework. This visual shorthand, credited to communications specialist David Hoffman, drew upon established conventions in telecommunications diagrams. The term "cloud computing" achieved broader recognition in 1996 following internal documents from Compaq Computer Corporation detailing future internet and computing strategies.