Agents use the same browser tab, which creates a conflict

Where does the bug appear (feature/product)?

Cursor IDE

Describe the Bug

When agents are working in chat mode, and in the process of completing tasks, they open a browser to check the web application, and often use the same tab, which creates a conflict. For example, one agent opened a page page1.html , and examines it, at the same time another agent in the same browser opens the page page2.html to study according to your own task. The first agent sees that his page page1.html switched to page2.html and returns to page1.html and the second agent does the same thing, only returns the browser to its page and so they can interfere with each other for a long time, although sometimes I ask to work with the browser in a new tab, sometimes it helps, but it still happens

Steps to Reproduce

In a chat mode give a task to Agent (Opus 4.5, Opus 4.6) to investigate a http://localhost:3000/page1.html, open a new tab in the chat and give another task to another agent (Opus 4.5, Opus 4.6) to investigate http://localhost:3000/page2.html.

Expected Behavior

Both agents will use the same browser and each of agent will try to open its page.

Operating System

MacOS

Version Information

Version: 2.4.31 (Universal)
VSCode Version: 1.105.1
Commit: 3578107fdf149b00059ddad37048220e41681000
Date: 2026-02-08T07:42:24.999Z
Build Type: Stable
Release Track: Early Access
Electron: 39.2.7
Chromium: 142.0.7444.235
Node.js: 22.21.1
V8: 14.2.231.21-electron.0
OS: Darwin arm64 25.2.0

For AI issues: which model did you use?

Opus 4.5, Opus 4.6 + thinking mode sometimes

Does this stop you from using Cursor

No - Cursor works, but with this issue

1 Like

Hey, thanks for the report. This is a known limitation. Right now, parallel agents use the same browser instance, so they end up fighting over the active tab.

As a temporary workaround, try having one agent use the browser while the other works on non-browser tasks, or take turns using the browser between agents. Adding explicit instructions like “open a new browser tab, don’t reuse existing tabs” in the prompts can sometimes help (like you noticed), but it’s not fully reliable because the root cause is architectural.

Let me know if you run into anything else.

Thanks for your reply. Yes, I ask agents to use a new browser and lock it to prevent to use it by other agents. Sometimes it works.

1 Like