I saw that AI Browser lets me and the Agent interact with the same page.
If a task reaches a QR-code login or two-factor verification step, should I ask it to pause before taking over? Once I'm done, is saying 'continue' enough, or should I ask it to check the current page again? I'd like to know a reliable handoff sequence.
Yes, you can hand off within the same Buda AI Browser: you and the Agent operate the same cloud browser page. When it reaches a QR-code login or MFA step, tell it to stop clicking and wait for your input. Open that Agent's AI Browser tab and complete the verification yourself; do not paste passwords, one-time codes or recovery codes into the chat. Confirm that the browser actually reached the signed-in page before handing control back.
Then reply in the same Session: “I completed the login. Re-observe the current page, confirm the signed-in account and current state, and only then continue. Do not repeat the form submission from before the handoff.” Re-observation matters because your manual actions changed the page while the Agent's earlier view may now be stale. If the Agent keeps acting while you need control, stop that step or run, then give a fresh instruction in the same Session. For future scheduled work, ask it to check the site is still logged in first.
The AI Browser guide explains the shared canvas. Is this a one-time QR login or MFA required for every action? The Chinese discussion covers the same handoff.
Sign in to join the discussion. Reading is always open; you only need an account to post.
Sign inI saw that AI Browser lets me and the Agent interact with the same page.
If a task reaches a QR-code login or two-factor verification step, should I ask it to pause before taking over? Once I'm done, is saying 'continue' enough, or should I ask it to check the current page again? I'd like to know a reliable handoff sequence.
Yes, you can hand off within the same Buda AI Browser: you and the Agent operate the same cloud browser page. When it reaches a QR-code login or MFA step, tell it to stop clicking and wait for your input. Open that Agent's AI Browser tab and complete the verification yourself; do not paste passwords, one-time codes or recovery codes into the chat. Confirm that the browser actually reached the signed-in page before handing control back.
Then reply in the same Session: “I completed the login. Re-observe the current page, confirm the signed-in account and current state, and only then continue. Do not repeat the form submission from before the handoff.” Re-observation matters because your manual actions changed the page while the Agent's earlier view may now be stale. If the Agent keeps acting while you need control, stop that step or run, then give a fresh instruction in the same Session. For future scheduled work, ask it to check the site is still logged in first.
The AI Browser guide explains the shared canvas. Is this a one-time QR login or MFA required for every action? The Chinese discussion covers the same handoff.
Sign in to join the discussion. Reading is always open; you only need an account to post.
Sign in