mcp -aws

This AWS MCP Server implements Model Context Protocol support for AWS operations, focused on S3 and DynamoDB. It exposes a set of tools that allow Claude (via MCP clients) to perform common AWS actions such as creating and listing S3 buckets, uploading and reading objects, and managing DynamoDB tables and items, including batch and TTL operations. All operations are logged through the audit endpoint at audit://aws-operations, enabling traceability of actions performed through Claude.

To use it, configure your MCP client to connect to the server using the mcp-server-aws entry (as defined in the mcp_config). Then issue requests for the available tools, for example creating a bucket (s3_bucket_create) or creating a DynamoDB table (dynamodb_table_create). The server will execute these AWS operations on your account and return results to Claude, along with audit data for each operation.

Prerequisites:

• Access to a machine with Python and uv (Python UV) installed, or you can run via Smithery as shown below.
• AWS credentials with IAM permissions to S3 and DynamoDB.
• Network access to AWS services from your environment.

Option A: Install via Smithery (recommended for Claude Desktop)

1. Install the MCP server via Smithery:

npx -y @smithery/cli install mcp-server-aws --client claude

2. Ensure AWS credentials are configured in your environment or via AWS CLI.
3. In Claude Desktop, add or configure the mcp-server-aws entry in claude_desktop_config.json as described in the README, then load Claude and test an operation (e.g., create an S3 bucket).

Option B: Manual installation (local development)
1. Clone this repository:
```bash
git clone https://github.com/modelcontextprotocol/servers.git
cd servers/mcp-server-aws

2. Set up AWS credentials:

• Environment variables:

export AWS_ACCESS_KEY_ID=your_access_key_id
export AWS_SECRET_ACCESS_KEY=your_secret_access_key
export AWS_REGION=us-east-1

• Or configure via AWS CLI with aws configure.

3. Run the MCP server using uv:

uv --directory /path/to/repo/mcp-server-aws run mcp-server-aws

4. Connect Claude Desktop using the guidance in the README to load the server and verify by performing an AWS operation (e.g., create an S3 bucket).

Tips and common issues:

• Ensure your AWS credentials have the necessary permissions for both S3 and DynamoDB operations.
• If you encounter authentication errors, verify the environment variables are loaded in the same shell session that starts the MCP server, or rely on the AWS CLI credential store.
• When using the manual uv approach, ensure the path provided in --directory points to the root of the mcp-server-aws repository.
• Audit logs are available at audit://aws-operations; consult them to verify operation provenance and troubleshoot failed operations.
• If running behind a proxy or restrictive network, ensure outbound access to AWS endpoints is allowed.
• DynamoDB and S3 operations may incur charges; consider using test resources or a dedicated sandbox account.