MCP Manager

Deploy multiples and secure MCP from your private environment

1

Install MCP Manager CLI

You can install mcp-manager using either pip (for Python environments) or Docker (recommended for ease of setup and isolation).

a. Installation via pip

This method is suitable if you prefer to install Python packages directly into your environment. Open your terminal and run the following command:

pip install mcp-manager-cli

The mcp-manager-daemon runs the core server process in the background. Open a new terminal and execute:

mcp-manager-daemon

Keep this terminal open as long as you intend to use the MCP Manager CLI.

In your original terminal (or another new one), you can now interact with the CLI by typing:

mcp-manager

b. Installation via Docker (Recommended)

Ensure Docker is installed and running on your system. If not, refer to the official https://docs.docker.com/ documentation for installation instructions.

Execute the following command in your terminal. This will download the mcp-manager-cli image (if not already present), run it as a detached container, and map the necessary port and volume.

docker run -d \
  --name mcp-manager \
  -p 4123:4123 \
  -v ~/.aws:/root/.aws \
  -v /var/run/docker.sock:/var/run/docker.sock \
  -v $(pwd)/.cache:/app/.cache \
  khaitrinhxuan/mcp-manager-cli:latest

Once the container is running, you can execute mcp-manager commands within the container using docker exec:

docker exec -it mcp-manager mcp-manager

2

Understanding MCP Manager

Let's break down how MCP Manager works and how you can set it up to suit your needs.

At its core, MCP Manager uses a background process called a daemon. This daemon automatically starts up and listens for connections on port 4123 by default. When you run MCP Manager for the first time, it'll automatically create a sample configuration file for you. You'll find this file named mcp_config.json inside the ./cache/ folder. Here's the format it uses:

{
  "mcpServers": {
    "hackernews": {
      "command": "uvx",
      "args": [
        "mcp-hn"
      ],
      "auto_start": true
    }
  }
}

To see what's running, use server list (a default Hacker News MCP server is included). You can start and stop servers using server start and server stop. To check out the tools, use tools list. For a full list of commands, use help.

MCP Manager has a built-in authentication mechanism. To get the API key needed for authentication, simply run the api-key command.

3

Adding Your Own MCP Server

Let's walk through an example of how to add a new MCP server. You can begin setting up your own by modifying the mcp_config.json file. In this example, we'll show you how to add a new MCP server aws-cli-mcp-server .

{
  "mcpServers": {
    "hackernews": {
      "command": "uvx",
      "args": [
        "mcp-hn"
      ],
      "auto_start": true
    },
    "aws-cli-mcp-server": {
      "command": "uvx",
      "args": [
        "aws-cli-mcp-server"
      ],
      "env": {
        "AWS_PROFILE": "your_aws_profile"
      },
      "auto_start": true
    }
  }
}

After modifying the file, run server reload to load the new configuration, then use server list to check its status. Your MCP server is now successfully set up!

MCP Manger has a builtin authentication mechanism please run api-key to get the api-key for authentication.

To make your server public, forward port 4123. For quick testing, ngrok is perfect! Just run:

ngrok http 4123

Remember to save the public URL that ngrok provides; you'll need it later !

Find installation details https://ngrok.com/downloads/mac-os.

4

Connect your MCP(s) to CloudThinker

Navigate to https://app.cloudthinker.io/ and then click on Connectors in the left sidebar under Operate.

Selet MCP Servers tab and click Add Server button:

• For the Server URL field:

  • If connecting your main MCP Manager, use YOUR_NGROK_URL/sse.

  • If connecting a specific server within your MCP Manager, use YOUR_NGROK_URL/server_name/sse.

• In the Bearer Token field, enter the API key you obtained by running the /api-key command in your MCP Manager.

After submitting the form, your MCP server connector is successfully added !

5

Assign MCP to Agent

You assign MCP to Agent by navigating to Operate, then choosing Agents. Locate the agent card you want to configure within the Team Member section, and click its Settings button in the bottom-right corner. From there, select your desired MCP Server.

6

Complete your task with MCP

Head to our chat page. Since the agents are now equipped with the MCP server you selected, you'll be able to use it as needed for your tasks!

7

Conclusion

You've successfully navigated through the steps to set up and interact with the MCP server, including connecting it to CloudThinker and accessing it via the chat page.