Lesson 6 4 min

Browser control and takeover

An agent with browser control works in its own tab group, shows what it's doing, and lets you take over at any moment.

Its own tab group

With Browser control, an agent opens pages, reads them, clicks, types, chooses options and scrolls, using the logins you already have in this browser. It works in its own tab, grouped and named after it (or after the team, when a team shares one tab group), and never uses the tabs you opened yourself unless its rules allow it.

You stay in charge

While it works, a banner on the page shows what it is doing, with:

• Pause / Resume: hold the next action.
• Take over: pause and bring the tab to the front so you can do a step yourself.
• Stop: end the agent’s turn.

Elements it acts on are highlighted, and each action is listed under the reply, such as Clicked “Reserve”.

Prompt injection

Pages sometimes contain text written to manipulate agents, such as “ignore your instructions and…”. Page text is always treated as content, never as instructions, and so is everything else an agent gets from outside: web search results, MCP tools, workflows and teammates’ answers. When a page tries to instruct the agent, the step list shows Instructions in the page were ignored, and the rules still apply to everything it does next.

Key takeaways

1. The agent works in its own tab group, named after it; your own tabs only if its rules allow it.
2. Pause, Take over and Stop are on the page while it works.
3. Page text is always content to read, never instructions.