The features in this guide are only available in the Subframe desktop app. Download the desktop app for macOS or
Windows ↗.
Capturing a page
- Click Capture in the browser toolbar
- Add the page to a canvas:
- Click Add to canvas in the confirmation message, then select a canvas
- Or open a canvas and press Cmd + V to paste it
Asking the agent to capture a flow
In Agent chat, ask the agent to capture a flow, such as “capture every state of the checkout flow”, or click the Capture every state of a flow… starter prompt in a new chat. The agent opens the pages in browser tabs, clicks through each step, and captures every screen, dialog, and error state it finds. While the agent uses a tab, the tab shows Agent is using this tab. Click Take control to take the tab back. The agent then stops acting in it and asks you how to continue. The agent can use any site, including ones that aren’t running on your computer, and fills in and submits forms to reach each state. It asks you in chat before an action it can’t undo outside the page, like paying, sending a message as you, deleting data, or changing a password or sharing settings, unless you asked for that action. If a page needs you to sign in, the agent asks you to sign in in the tab or to give it credentials to enter.Capturing an element
Capture an element when you only need part of the page.- Click Inspect in the browser toolbar
- Click on an element
- Press Cmd + C or click Copy selection in the Inspector panel
- Paste the element onto a canvas or into Agent chat