Introduction to MCP and Amazon Connect

MCP, or Model Context Protocol, is an open-source protocol developed by Anthropic (the creators of Claude) as a means of providing a standard way to connect LLMs to other systems. Those other systems can be publicly accessible, such as weather information, commercial, like payment processors, or private for systems built and run within organisations.

With MCP, we are able to give your AI of choice superpowers so that they are highly skilled in certain things. This is important as AIs are built using data that is available at the time, but things change and not all information is available during the training cycle.

So, having an AI that is fluent in Python is great, but giving it the ability to then use that language to interact with your APIs and perform tasks and automations is where the superpowers come in.

Companies that have an API-first strategy are early adopters of MCP, and in just a few short months these have exploded. In this blog, we're going to take a look at using the AWS MCP to build and configure Amazon Connect. That's right, we will use natural language to

Create an Amazon Connect instance
Sync resources between 2 instances

So lets get stuck in:

Pre Reqs

If you are going to try out these experiments then this section is required for you. Otherwise, skip over.

For my set-up I am using

Install Claude Code
1. https://www.claude.com/product/claude-code
2. Follow the onscreen prompts to set-up claude code when you;re ready move on
Install uv
1. brew install uv or https://github.com/astral-sh/uv
Install AWS CLI
1. brew install awscli or https://docs.aws.amazon.com/cli/latest/userguide/getting-started-install.html
2. Ensure your aws credentials are set-up however you need them. Run aws s3 ls and see the AWS guide linked to troubleshoot any access issues.
Open terminal and create a new folder blog-mcp-connect
1. In the folder create a file called mcp.json and paste in the below (changing the AWS_REGION as required)

1{
2  "mcpServers": {
3    "awslabs.core-mcp-server": {
4      "command": "uvx",
5      "args": ["awslabs.core-mcp-server@latest"],
6      "env": {
7        "FASTMCP_LOG_LEVEL": "ERROR",
8        "aws-foundation": "true",
9        "solutions-architect": "true",
10        "AWS_REGION": "eu-west-2"
11      },
12      "autoApprove": [],
13      "disabled": false
14    }
15  }
16}
17