Amplify Data API

You can interact with your AWS Amplify Gen2 data through natural language using this MCP server. It lets you view schemas, run queries, and modify data with plain English, streamlining AI-assisted data work while respecting your access permissions.

How to use

When you want the AI to operate on your Amplify data, connect your MCP client and direct requests to the Amplify Data API MCP Server. Use natural language queries and actions to view models, inspect fields and relationships, run GraphQL queries, and perform create, update, or delete mutations. Ensure you authenticate with a user who has appropriate permissions, ideally in the ADMINS group, so you can exercise full CRUD capabilities.

Prefix your messages with the MCP server tag to route requests correctly. For example, send your questions or commands starting with the Amplify server identifier you configured (e.g., @amplify_data_api). The server will translate your natural language requests into the appropriate GraphQL operations against your Amplify data models.

How to install

Prerequisites: you need Node.js v16 or higher and an AWS Amplify Gen2 app with data models. You also require an amplify_outputs.json file from your project and Cognito authentication set up in Amplify. Ensure you have a user with permissions in Cognito (ideally in the ADMINS group).

Install and set up the MCP server with these steps.

Step 1: Clone the repository

git clone https://github.com/dbsmith7491/aws-amplify-data-mcp.git
cd aws-amplify-data-mcp

Step 2: Install dependencies

npm install

Step 3: Build the project

npm run build

If you are configuring for Claude Desktop, prepare the desktop configuration file as described in the configuration section below.

Configuration for Claude Desktop

To connect Claude Desktop to the Amplify Data API MCP Server, add a server entry that runs the built MCP index and points to your amplify_outputs.json file. The configuration below shows exactly how the stdio-based command is defined.

{
  "mcpServers": {
    "amplify_data_api": {
      "command": "node",
      "args": [
        "/ABSOLUTE/PATH/TO/aws-amplify-data-mcp/build/index.js",
        "--amplify-outputs",
        "/ABSOLUTE/PATH/TO/YOUR/amplify_outputs.json"
      ],
      "env": {
        "AMPLIFY_USERNAME": "your-admin-username",
        "AMPLIFY_PASSWORD": "your-admin-password"
      }
    }
  }
}

Replace the absolute paths with your actual paths. For full functionality, use credentials for a user in the ADMINS group. After saving, restart Claude Desktop to apply the changes.

Security and access control

This MCP server uses Cognito authentication to access your Amplify data. Your Amplify project must be configured to use Cognito-based authentication. The server will only access data models and records that the authenticated user can access.

Best practice is to create an ADMINS user group and add appropriate users. Grant full create, read, update, and delete permissions to that group for your data models. Without proper admin permissions, the MCP server is limited in what it can do.

Troubleshooting

Authentication issues: verify Cognito credentials and user pool settings. Missing or outdated amplify_outputs.json can prevent model access. Check logs for permission errors and ensure your user belongs to the ADMINS group if you expect full permissions. If you encounter connection problems, review the Claude Desktop logs for errors.

Available tools

list_models

List all available data models in your Amplify app.

model_details

Show detailed information about a specific model's fields and relationships.

run_queries_mutations

Execute custom GraphQL queries and mutations to list, retrieve, create, update, or delete data.

relationships

Inspect relationships between models to understand how data is connected.

list_enums

List all enum types defined in your data models.

manual_login

Authenticate manually if automatic authentication fails.