> ## Documentation Index
> Fetch the complete documentation index at: https://docs.watchtower.systems/llms.txt
> Use this file to discover all available pages before exploring further.

# Running Patrol

> Start test runs, watch them stream live, and review run history.

Patrol runs your Playwright or Cypress tests and syncs the results straight back into Test Center. It's Watchtower's on-demand test runner, with no separate CI dashboard to check. On a hosted workspace, Patrol dispatches the run to your connected repository's GitHub Actions workflow or GitLab CI pipeline, using a Playwright or Cypress workflow that matches whichever framework your test source uses.

## Starting a run

From **Test Center → Patrol**:

1. Choose a **Test Suite**: **All Tests**, or a specific folder.
2. Optionally expand **Advanced** to narrow things further:
   * **Project**: run against a specific Playwright project or browser target.
   * **Grep Filter**: filter by test title using a chip-based search; add as many terms as you need. Switching to a different test suite clears any grep terms left over from the previous one.
3. On a hosted workspace, pick an **environment** from the selector next to Run Patrol, or leave it on **No specific environment**. The chosen environment is passed through to the CI run and made available to your tests.
4. Click **Run Patrol**.

You can start more than one run at a time. Click **Run Another** to launch a second suite while the first is still going. If a suite is already running, its button is disabled until that run finishes.

## While a run is active

The **Active Runners** panel shows every in-progress run, with:

* A live `done/total` count and the test currently executing.
* A progress bar. For runs dispatched to CI, Watchtower can't see per-test progress until results are reported back, so the bar shows an indeterminate sweeping animation instead of a percentage until the first results land, then switches to a real progress bar.
* Running pass/fail counts.
* A **Stop** button to cancel that runner individually, or **Cancel All** to stop everything at once. Cancellation stops the underlying CI run, rather than only removing its card in Watchtower.

The console panel streams timestamped log lines as the run progresses, so you can watch output in real time instead of waiting for the run to finish.

## After a run

* **Run History** shows a chart of total vs. passed tests across recent runs, for the current suite or all suites.
* **Past Runs** lists every run, paginated, with its date, environment, framework (`PW` for Playwright, `CY` for Cypress), what was run (folder or "All tests," tagged with the grep pattern if one narrowed it), branch, commit, and pass/fail/skip counts and duration.

If a run's suite or grep filter didn't match any tests, or every matched test was skipped, Watchtower shows **"No tests ran"** rather than treating it as a failed run.

Results from a run also update the corresponding tests in Test Center immediately, and roll up into the [Dashboard](https://docs.watchtower.systems/v0.7.0/dashboard-and-reports.md), including the [Go/No-Go release verdict](https://docs.watchtower.systems/v0.7.0/dashboard-and-reports.md#release-readiness).

> [!TIP]
>
> Use **Dedup** in Patrol if a sync has produced duplicate entries for the same test. It removes the duplicates in one click.

## Release email

If **Email release notifications** is turned on under **Configuration → Codebase → Notifications**, a completed Patrol run can also send an email summarizing the result. See [Configuration](https://docs.watchtower.systems/v0.7.0/configuration.md#notifications).

## Flaky-test checks

Choose **2x to 5x** repeats when starting a Patrol to check whether the same tests produce consistent outcomes. **Past Runs** groups the repeats and labels them **Flaky · N**, **Failing · N**, or **Same outcome**. Open the group for a **Flaky check** view of the tests whose outcomes changed.

Under **Settings → Flaky-test repeats**, enable **Run repeats in parallel** to dispatch each repeat as a separate CI job. This finishes sooner but consumes more CI minutes. Leave it off when repeats share a test account or remote environment and could interfere with each other.

Use [Signals](https://docs.watchtower.systems/v0.7.0/signals.md) to review flaky outcomes and regressions across recorded runs.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.