Getting started

This guide gets you a running instance, a terminal signed in to it, and an empty Terraform configuration that can reach it. It takes about five minutes.

1

Create your account

Sign up at cloud.marmotdata.io and pick a plan. Free needs no card and has no end date: one Standard instance with a 500-asset catalog, enough to point Marmot at a system you actually run.

2

Launch an instance

Launching takes about a minute and continues if you close the page. You end with an address like https://acme.marmotdata.cloud. That is your instance, and almost everything from here happens there rather than in the console.

3

Sign in to your instance

Reveal the initial password next to the admin username in the console and sign in at your instance URL. Marmot asks for a new password immediately, and the generated one stops working.

Do this before anyone else signs in

Configure single sign-on now, while the instance has one user. Retrofitting it onto an instance full of local accounts means reconciling each one by hand.

4

Connect your terminal

AutomaticManual

Homebrew

brew install marmot

Install script

curl -fsSL get.marmotdata.io | sh

Verify Scripts

It's good practice to inspect the contents of any script before piping it into bash.

1

Download the binary

Download the latest Marmot binary for your platform from GitHub Releases.

2

Make it executable

chmod +x marmot

3

Move to your PATH

sudo mv marmot /usr/local/bin/

marmot login https://acme.marmotdata.cloud
marmot assets list

Your browser opens, you sign in, and the CLI stores a 24-hour token. An empty asset list is correct on a fresh instance.

5

Start a Terraform configuration

Everything you do to the instance from here is declared in Terraform. Create a repository and start with the provider block alone:

terraform {
  required_version = ">= 1.11"

  required_providers {
    marmot = {
      source = "marmotdata/marmot"
    }
  }
}

provider "marmot" {
  host = "https://acme.marmotdata.cloud"
}

terraform init
terraform plan

The provider authenticates with your CLI session, so there is no API key to create. An empty plan confirms it can reach the instance. When the configuration moves to CI it gets a service account of its own.

Your first pipeline

Pick the cloud your data lives in. Each guide walks you from the empty configuration above to a running pipeline that catalogs a native source with no credential at all, then shows how to add a source that needs a password.

Google Cloud

Catalog BigQuery with Workload Identity Federation

AWS

Catalog the Glue Data Catalog with an IAM role

Azure

Catalog Blob Storage with a federated app registration

Somewhere else

A pipeline that reads its password from your secret manager at run time

Next steps

Lock down access

Grant roles per asset, data product, glossary term and secret store

Connect your agents

Point Claude, Cursor or your own agent at your instance's MCP endpoint

Manage your instance

Sizes, stopping and starting, and your own domain

Plans and billing

What each plan includes and how sizes are billed