Skip to main content
The wizard signs you in, installs Momentic, runs a sample test, and connects your coding agent. The manual setup does the same steps one at a time.

Prerequisites

The wizard and the manual setup both need:
  • Node.js 22.12.0+ or 24.0.0+ (node -v)
  • A package.json in your project. Run npm init -y if you do not have one. Commit it with the lockfile, so CI and local runs use the same Momentic version.
  • A Momentic account. Sign up to create one.

Set up with the wizard

Run the wizard in your terminal:
It signs you in through the browser, installs momentic, writes momentic.config.yaml and a sample test under web/, installs Chromium, runs the sample test, and asks which coding agents to set up.

Set up with AI

Copy this prompt and paste it into Claude Code, Cursor, Codex, Copilot, or any other agent. It runs the wizard with an API key because an agent cannot complete the browser sign-in. Then it writes your first test.

Manual setup

1

Install

The examples below use npx. With pnpm, use pnpm exec momentic; with yarn, use yarn momentic.
2

Authenticate

Sign in with your browser:
The CLI saves the credentials to ~/.momentic/auth.json and uses them when MOMENTIC_API_KEY is not set.In CI, or in a shell without a browser, use an API key instead. Create one in the dashboard and export it:
Check which account and organization the CLI uses:
3

Initialize

This writes momentic.config.yaml to your project root. In an interactive terminal, it also offers to install the coding agent skills and the MCP server. init --yes accepts the defaults without prompts, so it works in agent shells and CI.init does not create tests. You write your first one in the editor below. For a sample test, use the wizard above instead.
4

Connect your coding agent (optional)

Skip this step if init already did it. The two commands are independent: skills are local instruction files. The MCP server is a live process that gives the agent a browser and Momentic tools.
Both commands detect the coding agents on your machine and ask which ones to configure. Pass --yes to skip the prompt. See Building with AI for what each command does.
5

Install browsers

Momentic runs your tests in a headless browser. Install Chromium:
6

Write and run your first test

Open the local editor:
Create a test with the URL of your app as the base URL. Write one natural-language step for a flow that works. If you have no app running, use https://practicetestautomation.com/practice-test-login/ with the step Log in with username "student" and password "Password123", then confirm the success page loads.Save the test. The editor writes the .test.yaml file to your project. Run it from the editor, then run it from the CLI by name:
A passing run prints each test, the totals, and the dashboard link:
--upload-results sends the result to the dashboard. Without it, the run stays on your machine.
7

Verify

  • Check that the .test.yaml file exists in your project.
  • Open the dashboard and check Runs for the result.

Browsers

Momentic runs web tests on Chromium, Google Chrome, or Chrome for Testing. Firefox and Safari are not supported. Chromium is the default. To use another browser, install it, then set the project default in momentic.config.yaml:
momentic.config.yaml
To override the browser for one run, pass --browser chromium, chrome, or chrome-for-testing:
The flag overrides the per-test and project defaults. See browsers for the differences between the three.

Next steps

Core concepts

Finding elements, writing assertions, modules, variables

Building with AI

Skills, the MCP server, and a prompt for your coding agent

CI/CD

Run Momentic on every pull request

Goal-based testing

Build resilient tests around user outcomes
Clone momentic-ai/examples for a full working example.