Tasks

Open a project and write in the message box on the dashboard. Choose Browser task, API task or Code task. Browser and API tasks use the project's URLs. Code tasks need a connected repository and a branch.

Click + to add a file, give the task one of the project's connectors or open Mentions. Type @ to point the agent at a test case, bug, endpoint, page, requirement or integration.

Tasks also start from Slack, the CLI, schedules and new pull requests.

• Say where to start and who the user is: the page, the account and its role.
• List the steps, then the result you expect to see.
• Give exact data: emails, amounts, coupon codes, file names.
• Keep to one flow. Start another task for the next one.

As a signed-in customer, start on /checkout, apply coupon SAVE10 and pay with the test card 4242 4242 4242 4242. Check that the confirmation page shows an order number and a 10% discount.

The testing guide has more examples.

In a task, type /plan or turn on Plan mode in the + menu. The agent looks around first, writes a plan and waits for your approval before it acts.

The task view shows the live browser, each step and every tool call. Send a message at any time to add detail or change direction. Click Stop agent to end the task.

When the agent needs you, for example at a CAPTCHA, it hands you the browser. Do what is needed, then click Resume agent.

• Report Document: a PDF with a summary, what was covered, observations, bugs and recommendations. Custom reporting instructions change what it focuses on.
• Recording of the browser for browser tasks. We keep recordings for 7 days on the Free plan and 30 days on paid plans.
• Bugs with a severity, steps to reproduce, and the expected and actual result.
• Test cases, when you ask the agent to save a flow so you can run it again.

To show a task to someone outside your team, open the … menu in the tab bar and click Share test. We copy a public link to it.

Each task uses your plan's included usage or your credit balance. See pricing.