AWS Bedrock MCP Client and Server

AWS Bedrock MCP Client and Server

A Spring Boot application that serves as a client for the Model Context Protocol (MCP), enabling interaction with Amazon Bedrock AI models with tool-using capabilities.

Application Overview

This application provides a REST API for:

• Communicating with Amazon Bedrock models
• Using MCP tools for calculations, weather information, and web page retrieval
• Handling complex AI queries that might require multiple tools
• The application runs as an AWS Lambda function behind API Gateway, making it serverless and scalable.

Prerequisites

• Java 21 or later
• Maven 3.8+
• AWS CLI configured with appropriate credentials
• AWS CDK installed (npm install -g aws-cdk)
• Weather API key from https://www.weatherapi.com/my/

Environment Setup

Clone the Repository

git clone https://github.com/sverze/aws-bedrock-mcp-client-server
cd aws-bedrock-mcp-client-server

Configure AWS

You will credentials that has priviledges to the following services

• Lambda
• API Gateway
• Bedrock
• Bedrock Model Anthropic Sonnet 4.0

To configure local credentials run the following command:

aws configure

###Building the Application

The project follows a multi-module Maven structure. To build all modules run this from the parent directory:

mvn clean package install -DWEATHER_API_KEY=<your_api_key_here>

This will build all modules in the correct order:

• aws-bedrock-mcp-server
• aws-bedrock-mcp-client
• aws-bedrock-mcp-client-server-cdk

Deploying to AWS

The AWS deployment is managed using AWS CDK. To deploy the application, run the following commands:

cd aws-bedrock-mcp-client-server-cdk
cdk bootstrap
cdk deploy --context WEATHER_API_KEY=<your_api_key_here>

After deployment completes, CDK will output the API Gateway URL in the terminal. You will need this URL to interact with the API and perform the tests below.

Testing the API

Once deployed, you can test the API endpoints using curl or a tool like Postman:

1. Test the Hello World endpoint

   curl https://<your-api-id>.execute-api.<region>.amazonaws.com/prod/hello
   

   Expected response: Hello World from MCP Client!

2. List available tools

   curl https://<your-api-id>.execute-api.<region>.amazonaws.com/prod/tools
   

3. Perform a calculation

   curl -X POST "https://<your-api-id>.execute-api.<region>.amazonaws.com/prod/calculate?operation=add&a=5&b=3"
   

4. Get weather information

   curl -X POST "https://<your-api-id>.execute-api.<region>.amazonaws.com/prod/weather?location=New%20York&format=celsius"
   

5. Convert a web page to markdown

   curl -X POST "https://<your-api-id>.execute-api.<region>.amazonaws.com/prod/webpage?url=https://example.com"
   

6. Make a complex query - The Weather

   curl -X POST "https://<your-api-id>.execute-api.<region>.amazonaws.com/prod/query?query=What%20is%20the%20weather%20in%20London%20and%20calculate%2045%20divided%20by%209"
   

7. Make a complex query - Summarise a Webpage

   curl -X POST "https://<your-api-id>.execute-api.<region>.amazonaws.com/prod/query?query=Summarise%20the%20following%20webpage%20https://example.com"
   

Monitoring and Logs

View application logs in CloudWatch Logs:

• Navigate to AWS CloudWatch console
• Check the /aws/lambda/McpClientServerFunction log group

Clean Up

To remove all deployed resources:

cd aws-bedrock-mcp-client-server-cdk
cdk destroy

Architecture

This application implements a serverless architecture:

• API Gateway serves as the HTTP endpoint
• AWS Lambda runs the Spring Boot application
• Amazon Bedrock provides the AI model capabilities
• MCP Server provides tool implementations
• The AWS CDK stack creates and configures all the necessary AWS resources, including IAM roles with appropriate permissions for accessing Amazon Bedrock.