Getting started.

1. Install the CLI

Build from source with Go 1.22 or later. This is the path that works today:

git clone https://github.com/mrdorianh/fieldwick.git
cd fieldwick/cli
go build -o /usr/local/bin/fieldwick .

Pre-built binaries for each OS are published to the GitHub releases page with each tagged release. When one is available for your platform, download it, mark it executable, and drop it on your PATH instead of building.

Verify the install worked:

fieldwick --version

2. Sign up and log in

Create an account at app.fieldwick.com/sign-up. Google OAuth is supported. No card is required on the Starter tier.

Once your account exists, authenticate the CLI. The default flow opens a device-auth page in your browser and polls for approval:

fieldwick login

Prefer to stay in the terminal? Sign up and log in without leaving the shell:

fieldwick signup you@example.com
fieldwick login

For CI or agent environments, generate a token in the dashboard and pass it with fieldwick login --token $FIELDWICK_TOKEN.

3. Deploy your first app

From inside your project directory, run fieldwick init to detect your language and framework. It writes a fieldwick.json into the current directory.

fieldwick init

Create a service to deploy into. The service holds your live allocation, scaling settings, and deploy history:

fieldwick services create --name my-api --region us-east-1

Trigger a deploy and wait for it to reach a terminal state. The CLI exits 0 on success and 1 on failure, so it is safe to use in a CI pipeline:

fieldwick deploy --wait

Once the deploy finishes, the CLI prints your live URL. The service ID is written back into fieldwick.json so subsequent deploys from the same directory need no extra arguments.

For AI agents

The exact same three-step flow runs through the Fieldwick MCP server. Register the server with your agent host and it exposes tools for signup, login, init, and deploy that an agent can call directly without a shell.

The CLI is designed to be machine-readable. Add --json to any read command to get structured output instead of human- formatted tables. The deploy --wait flag emits a machine-readable exit code (0 = success, 1 = failure, 124 = timeout) so agents can branch on deploy outcome without parsing log lines.

# agent-friendly: structured output + reliable exit codes
fieldwick services --json
fieldwick deploy --wait --json

What is in fieldwick.json?

The config file lives at the root of your project. You can commit it to version control - it contains no secrets. A minimal example for a Node.js API looks like this:

{
  "name": "my-api",
  "region": "us-east-1",
  "build": {
    "command": "npm ci && npm run build"
  },
  "start": {
    "command": "node dist/server.js",
    "port": 3000
  }
}

After fieldwick services create runs, the CLI appends a serviceId field so subsequent deploys know which service to target.