Browser support

Marriska runs tests with Playwright, which means three real browser engines. All three are available on every plan, Free included — paid tiers add run volume, parallelism and history, not basic engine coverage.

How does this compare to other AI test tools? See Safari/WebKit coverage across AI test tools.

Supported browsers

Name	Engine	Approximates
`chromium`	Chromium	Chrome, Edge
`firefox`	Gecko	Firefox
`webkit`	WebKit	Safari

These are the only valid values anywhere a browser is named — the run config, the CLI --browsers flag, and screenshot URLs all accept exactly chromium, firefox, and webkit. There’s no Internet Explorer; it’s 2026.

Selecting browsers

In the app — the run-config popover has a checkbox per browser. Default is chromium.
On the CLI — --browsers chromium,firefox,webkit (comma-separated; defaults to chromium). See the CLI reference.

Each selected browser is a separate execution. Two browsers means two runs of the test, and two counts against your monthly run quota. By default they run one after another; the CLI --parallel flag runs them concurrently.

Headless vs. headed

Cloud runs are headless. You see step screenshots in the report, not a live window.
Local runs can be headed. Run the local agent or marriska-runner with --headed to watch the browser drive itself — handy for debugging a flaky test.

Browsers and the local agent

The local agent runs whatever browsers are installed on its machine, and reports them as capabilities when it connects. Install them with Playwright:

npx playwright install chromium   # or firefox / webkit

If no browsers are installed, the agent exits and tells you to run the install command.

Browsers and visual baselines

Each browser keeps its own visual baseline. Chromium, Firefox, and WebKit render real differences — especially in form controls and SVG — so a Firefox run compares against the Firefox baseline, never Chromium’s.