Tools

Browser

Let the agent open your app, click through it, and report back.

The agent can drive a browser: open your app, try a flow, and show you what it saw. You do not install a separate browser tool to start.

What it can do

• Open a URL and follow links
• Click, type, and scroll
• Take a screenshot of the page
• Read console messages
• Watch the requests the page makes

Actions and screenshots show up in the chat, next to the browser pane.

Uses

• Walk a flow you just changed and confirm it
• Check what a keyboard or screen-reader user would hit
• Turn a screen you can see into the matching interface code
• Adjust layout from a screenshot you attach

You stay in charge

Browser actions that need a yes will ask. You can limit which sites it may open. The browser session stays separate from the accounts in your everyday browser.