---
title: Browser Automation
description: Let the agent work in a visible browser you can watch and take over.
updated: 2026-09-12
canonical: https://turnpanel.com/docs/browser-automation
markdown: https://turnpanel.com/docs/browser-automation.md
llms: https://turnpanel.com/llms.txt
---

# Browser Automation

Let the agent work in a visible browser you can watch and take over.

## One visible session

Open Browser yourself or ask the agent to open it. You both see and control the same page. Its tabs and saved sign-ins are separate from your everyday browser.


## How automation works

- Before each action, the agent reads the current page controls. It can then click, type, choose options, scroll, or handle files you approved.
- After the page changes, the agent reads it again instead of reusing old controls. It opens sites in a mobile-sized view by default and switches to desktop when a wider layout is needed.


## Your turn

For passwords, CAPTCHA, two-factor codes, payment details, or an important confirmation, TurnPanel gives you control. Finish the step in the live Browser, then choose Done, hand back. Choose Cancel to stop.


### Hand off credentials and consequential choices in Browser

**Context:** Waiting for you

The Your turn banner pauses the agent and shows handoff controls.

- **Waiting for you** [waiting]: Human-only Browser step
- **Cancel** [ready]: Deny the pending handoff
- **Done, hand back** [active]: Allow the agent to continue

## Boundaries

- Instructions shown by a website cannot change what the agent is allowed to do.
- The agent should use a direct API, integration, command-line tool, or page reader when that is simpler and reliable.
- The agent needs your clear instruction before it buys, sends, publishes, subscribes, or deletes.