Create a new Playwright test file. Generate the full test code and pass it here.

List all test files with their last known status. Supports filtering by status and limit.

Read the contents of a test file.

Update a test — script, display name, description, schedule, or enabled state.

Permanently delete a test and its history. Irreversible — prefer update_test with enabled: false to deactivate.

Run all tests or a specific test file. Respects pause state. Sends a webhook notification if one is configured.

Run a test repeatedly, applying automatic fixes on each failure, until it passes or max attempts is reached.

Get overall suite status, or per-test status with failure details and fix suggestion when test_name is provided.

Pause all tests for up to 60 minutes. Tests resume automatically. Use before deploys to avoid false alerts.

Configure a webhook URL (Discord, Slack, or a generic JSON endpoint) for test notifications.

Create or update a named credential set (e.g. admin, user_comum). Fields are key-value pairs like { email, password, token }.

List all saved credential sets with masked values.

Delete a named credential set.

Start a pair-debugging session: opens a visible browser at the given URL for the human to drive by hand. Records console messages, network errors, page exceptions, navigations, and their clicks/fills with timestamps as they go, so you can correlate what they did with what broke. Use when the user wants to walk through a flow live and have you spot the bug, rather than writing a scripted test upfront.

Get the recorded timeline of the active pair-debugging session so far (console, network, errors, navigations, actions). Call this whenever the user says something looks wrong, to see exactly what just happened.

End the active pair-debugging session: closes the browser and returns the full recorded timeline plus a draft Playwright test built from the actions the human performed. Save it for real with create_test if it holds up.