fly machines run

The Fly.io command line interface is a higher level interface that enables you to do much of what you can do via the Machines API. It is suitable for scripting and ad hoc exploration and updates. This guide presumes that you have flyctl installed, and have successfully run either fly auth signup or fly auth login. The first step is intended to be run in an empty directory.

Create your app

Now use the fly apps create command:

fly apps create --generate-name --save

If you prefer, you can replace --generate-name with a name of your choice. You can also specify the organization by passing in a --org parameter, or let it prompt you.

Create IP addresses for your application

The following will create a shared IPv4 address and a dedicated IPv6 address:

fly ips allocate-v4 --shared
fly ips allocate-v6

If your application is going to be public, but instead is only going to be accessed by other applications within your organization, run the following instead:

fly ips allocate-v6 --private

Create a volume

This demo uses a volume. If your application doesn’t use a volume skip this step.

fly volumes create data --region iad --yes

Adjust the region as necessary.

Create a machine

This next part contains a lot of flags, so first an overview:

The first parameter specifies the image we will be running. Fly.io provides an image capable of running npx and uvx, which is sufficient to run many MCPs. If you have a custom MCP with unique requirements, you can provide your own image.
The next parameter specifies the command we will be running.
--entrypoint invokes fly mcp wrap passing the command we specified.
If you are using a volume, the --region selected must match a region in which you have an allocated but unattached volume.
--volume specifies the volume, and where it is to be mounted.
The --vm-* parameters specify the size of the machine desired.
--auto* and --port define what network services your application provides.

fly machine run flyio/mcp:latest \
  "npx -f @modelcontextprotocol/server-filesystem /data/" \
  --entrypoint "/usr/bin/flyctl mcp wrap --" \
  --region iad --volume data:/data \
  --vm-cpu-kind shared --vm-cpus 1 --vm-memory 1024 \
  --autostart=true --autostop=stop \
  --port 80:8080/tcp:http --port 443:8080/tcp:http:tls

Note that this creates and starts a machine. If you want to only create the machine, use fly machine create instead. Once this command completes, you can update your fly.toml to include this new information using the following command:

fly config save --yes

Accessing the MCP via an inspector

As the MCP inspector is a Node.js application, you need to Download and install Node.js first. MacOS users can use brew install node.

You are test out your MCP server using the MCP inspector:

fly mcp proxy -i

Navigate to http://127.0.0.1:6274 ; click Connect; then List Tools; select any tool; fill out the form (if any) and click Run tool.

Configure your LLM

Here’s an example claude_desktop_config.json:

{
  "mcpServers": {
    "filesystem": {
      "command": "/Users/rubys/.fly/bin/flyctl",
      "args": [
         "mcp",
         "proxy",
         "--url=https://mcp.fly.dev/"
       ]
    }
  }
}

Adjust the flyctl path and the value of the —url, restart your LLM (in this case, Claude) and try out the tools.

​Create your app

​Create IP addresses for your application

​Create a volume

​Create a machine

​Accessing the MCP via an inspector

​Configure your LLM

Create your app

Create IP addresses for your application

Create a volume

Create a machine

Accessing the MCP via an inspector

Configure your LLM