Run tests

A run executes test cases on one environment. Each test case becomes one test in the run, and the tests execute at the same time, each in its own browser.

1. Click Run test in the top right corner of any page, or press Cmd+R (Ctrl+R on Windows and Linux).

2. Under Environment, choose where to run. The project’s default environment is preselected; a PROD badge marks production.

3. Under What to run, choose All test cases, Test plan or Single test, and pick the plan or test case.

4. Click Run.

Testmode opens the new run’s page, where you can follow it live. Starting runs needs the Editor role or above.

Shortcuts into the dialog:

• On a plan’s page, Run test opens with that plan chosen.
• On Test cases, Run in a test case’s ⋯ menu opens with that test case chosen.
• In Runs, Re-run in a run’s ⋯ menu starts the same run again.

You choose	The run includes
All test cases	Every enabled test case in the project
Test plan	Every enabled test case in the plan
Single test	That one test case, which must be enabled

Left out, and not charged:

• Switched-off test cases. Switch one on to include it.
• Test cases with no steps and no Expected Outcome. There is nothing to check; the dialog says how many will be skipped.
• Test cases marked Changes data, on a production environment that doesn’t allow them. The run page says how many were left out. See Pause testing and protect data.

A run uses one credit for each action its tests take. The dialog says nothing about credits unless few are left, or the run needs more than you have: then it says how many it needs and how many are left, and disables Run. See How credits are counted.

• Schedules run a plan automatically. See Schedule test plans.
• Re-run: in Runs, the ⋯ menu on a run starts a new run with the same test cases on the same environment. On a test’s page, Re-run runs that one test again.
• CI pipelines start runs with an API key. See Run tests from CI.
• AI agents start runs over MCP. See Connect an AI agent over MCP.
• Self-driving runs tests too: its re-runs and drafts when it investigates a failure, and a trial of each test it writes or fixes. They show in Runs as started by Self-driving.

All of them use credits the same way, and their runs appear in Runs with who or what started them.

While a run is queued or running, Cancel run on its page stops it, after you confirm:

• Tests that haven’t started don’t run and use no credits.
• Tests that are running are stopped within a minute, and use the credits of the actions they already ran.
• Tests that have finished keep their results.

The run ends Canceled. Cancelling needs the Editor role or above. To keep one test out of a run that is still waiting, switch its test case off instead: its queued tests are skipped and not charged.

The run’s page fills in live: each test’s status, its steps as they complete, and, for editors, the live browser. See Read run results.