Skip to main content
Write, debug, and execute test steps in real-time using the test debugger.

​
Basic Workflow

1

Write and Run Test

Start by writing your test step in natural language. After running the step, the agent will process your instruction and suggest an action.
Writing a Test Step
Access keyboard shortcuts by typing / or clicking the Shortcuts button
2

Approve or Reject

Review the proposed action and either approve it to proceed or reject if it doesn’t match your intent.
Action Review and Approval
If you’re comfortable with prompting, you can write multiple steps at once and hit SAVE You don’t have to run each step in the preview mode. You can use the view of your application as a guide and knockout the steps faster!
Explore the different components that make up the test debugger.

Write and execute test steps in the left-hand panel of the test debugger.
Watch test runs get executed by Spursor (Spur’s AI agent) in real-time.
User interactions in this sandbox are not recorded. Any changes made by the user in this sandbox browser will not be incorporated into testing.
Preview Mode allows users to preview Spursor’s next action on the website, then will provide options for the user to approve or decline the action. When turned off, Spursor will execute its action without approval.

Preview Mode On: Spursor asks for approval

View Spursor’s thought process when evaluating and executing test steps.
View real-time console logs when running tests.
View real-time network logs when running tests.
View extracted or generated values that are saved in the testing environment. These values can be accessed and used in subsequent test steps.
Access page and window dimensions of the Spursor Sandbox to aid test step creation. Also, scroll inside the Spursor Sandbox and access the page position.
common use case: scroll down to desired point in the site, access the page position in scroll, and use that in subsequent test cases (eg. scroll down 150px)

scroll offset estimator demo

​
Why Spur Test Debugger?

The Test Debugger allows users to walk through the agent’s exact through process, resolving any issues with the test steps.
The Spur Test Debugger emulates the actual thought process of Spursor (our AI Agent), giving you complete visibility into how it interprets and executes your instructions.
By seeing how the AI responds to different instructions in real-time, you can improve your prompting skills and learn the most effective ways to communicate with the agent.
The interactive approve/reject workflow ensures that each step is executed exactly as intended, reducing errors and improving test reliability. You have full control over the agent’s actions before they’re executed.

​
Best Practices

Write specific, unambiguous instructions for the best results. For example: “Click the ‘Submit’ button” is better than “Click the button”
Include verification steps to ensure your test is working as expected:
  • Verify text content
  • Validate form submissions
For complex interactions, break them down into smaller, clearer steps:
  1. Hover over menu
  2. Wait for dropdown
  3. Click specific option
The agent’s performance for open-ended actions in this mode is still in beta. We recommend using specific, well-defined instructions for the most reliable results.

​
Current Limitations

Sequential Execution

Steps must run in sequence to maintain test integrity. (this is how we have it structured for now, but we are working on allowing you to run any step!)

Beta Features

Open-ended actions and verifications are currently in beta
Once a step is approved and completed, it cannot be edited in the current version of the session. However, you can edit any step as long as it has not been run yet
⌘I