Docs
Quickstart

Quickstart

Set up Fume and generate your first Playwright test from a Loom recording.

In this guide you'll connect your app, record one flow, and run your first Fume‑generated Playwright test.

Add a 2–3 minute Loom walking through this quickstart end to end.

Prerequisites

  • A web app you can log into in a staging or demo environment
  • A Loom (or screen recording) of the flow you want to test

1. Create a project

  1. Sign up: https://app.fumedev.com/signup
  2. Create a project and add your app's base URL (e.g. https://staging.example.com).

2. Record your flow

Record a 30–90s Loom showing:

  • How to navigate to the page(s)
  • Inputs and interactions
  • The expected success outcome (what should be visible/changed)
Insert a screenshot of an example Loom thumbnail and the success criteria callouts.

3. Generate tests

  1. Paste the Loom link into Fume and give the test a name.
  2. Click Generate. Fume will analyze the video and your app to produce Playwright tests.
  3. Review the proposed steps and assertions. Adjust text if needed and save.

4. Run tests

  • Run in Fume Cloud: click Run to execute on Fume's runners (free).
  • Or run in your CI: use the Fume CLI/API to fetch and execute Playwright tests on your machines.
Add code example for running with GitHub Actions and a sample Playwright config snippet.

5. Handle failures and self‑healing

When a selector changes or the UI shifts, Fume can:

  1. Fall back to a browser‑use agent to complete the run, then
  2. Update the Playwright code so future runs pass.
Add a before/after code diff snippet showing a selector update.

What's next

  • Read Concepts to understand how Fume models recordings, tests, runners, and self‑healing
  • Integrate CI for automatic runs on PRs and merges
  • Migrate existing tests to Fume to consolidate maintenance