# MCP AWS Tools

This Model Context Protocol (MCP) server provides tools for inspecting AWS cloud infrastructure. It uses your local AWS credentials to securely access your AWS account.

## Overview

The `mcp-aws` server exposes three powerful tools to the AI:
1. `list_s3_buckets`: Retrieves a list of all S3 buckets in your AWS account.
2. `list_ec2_instances`: Retrieves information about your EC2 instances (IDs, states, types) for a specified region.
3. `get_cloudwatch_logs`: Fetches recent log events from a specific CloudWatch log group.

## Prerequisites

1. **Node.js** installed (v18 or higher recommended).
2. **AWS Credentials**: This server uses the standard AWS SDK for JavaScript. You must have your AWS credentials configured locally on your machine.
   - Run `aws configure` via the AWS CLI to set this up, OR
   - Export environment variables `AWS_ACCESS_KEY_ID` and `AWS_SECRET_ACCESS_KEY` before running.

## Steps to Run

1. **Install Dependencies:**
   ```bash
   npm install
   ```

2. **Build the Project:**
   ```bash
   npm run build
   ```

3. **Configure MCP:**
   Add this server to your Claude Desktop or Antigravity MCP configuration file:
   ```json
   {
     "mcpServers": {
       "aws": {
         "command": "node",
         "args": ["/absolute/path/to/mcp-aws/build/index.js"],
         "env": {
            "AWS_REGION": "us-east-1"
         }
       }
     }
   }
   ```
   *(Note: Adjust the absolute path to point to where you saved this folder. The `AWS_REGION` is optional but helpful if not set in your config).*

4. **Restart your MCP Client.** You can now ask the AI to "Check my S3 buckets" or "List my running EC2 instances in us-west-2".