BrowserOS neo

BrowserOS neo is a local, open-source browser that lets AI agents such as Claude, Cowork, Codex, and Cursor work inside your logged-in accounts. It is designed for interactive, watchable browser tasks with session replay and parallel agent execution.

AI智能体开发编程代理 AI智能体目录 AI效率工具

访问网站

A browser for agents working in your own accounts

BrowserOS neo is a local, open-source browser that AI agents can drive while signed into your existing accounts. It is designed as a separate browser for interactive work, so your agent can operate in one browser while you keep using another.

The product supports logged-in workflows such as Gmail, GitHub, X, banking, and other authenticated sites, with live tab viewing and session replay. BrowserOS positions neo as an application you open and use, not a framework you embed in code.

Core features

Agent-driven browser control

BrowserOS neo gives an existing agent a browser to operate through MCP. The site names Claude, Cowork, Codex, and Cursor as examples of agents that can drive it.

Chrome login import

Setup includes one-click Chrome import so neo can use your logged-in sessions in a separate browser environment.

Live tabs and session replay

You can watch the browser while work is happening, and every session is saved as replayable video by default.

Parallel agent execution

The homepage says neo can run agents in parallel, letting several agents work at once in separate tabs inside the browser.

Separate browser for account-based tasks

neo is presented as a second browser that runs beside your main browser, keeping agent work isolated from your day-to-day browsing.

Open-source and free

BrowserOS neo is described as local, open-source, and free, with users bringing their own model keys.

Practical ways to use BrowserOS neo

Delegating logged-in web tasks
Use neo when an agent needs to act inside accounts you are already signed into, such as email, GitHub, or other personal services.
Watching and reviewing agent work
Use the live browser view and saved session replay when you want to inspect exactly what the agent did before deciding whether the result is acceptable.
Running more than one browser task
Use parallel agents when you have several independent errands to complete and want them to run at the same time in separate tabs.
Keeping agent work separate from your main browser
Use neo as a second browser when you want an agent to work in its own space rather than sharing the browser you use for your own browsing.
Using an existing coding agent without building a framework
Choose neo when you already have an agent like Claude, Cowork, Codex, or Cursor and want a browser for it to drive instead of adding a framework to your codebase.

Pros and Cons

Pros

Works with logged-in accounts instead of asking you to recreate sessions from scratch.
Shows browser activity live and keeps sessions replayable by default.
Supports parallel agent work in separate tabs.
Open-source and free.
Designed as a separate browser beside your main one, which helps keep agent activity isolated.

Cons

The site describes neo as interactive desktop work, not unattended automation.
BrowserOS states that neo is Chromium-only.
It is not presented as a framework, test runner, or app-embedding library.

FAQ

What is BrowserOS neo?

BrowserOS neo is a local, open-source browser that lets your AI agents drive a browser signed in with your own accounts.

Which agents can use it?

The site explicitly names Claude, Cowork, Codex, and Cursor, and says any agent that speaks MCP can drive neo.

Can it use my existing logins?

Yes. BrowserOS says neo supports one-click Chrome import so it can run in a separate browser with your logged-in accounts.

Does it save what the agent did?

Yes. The homepage says every session is recorded and replayable as video by default.

Is BrowserOS neo meant for CI or unattended jobs?

No. The comparison pages describe neo as interactive work on your own machine rather than a server-side or scheduled automation tool.

Quick Facts

Category: AI Browser
Product type: Local browser application
Primary users: People using AI agents for authenticated web tasks
Supported workflow: Logged-in browsing with live viewing and replay
Notable compatibility: Claude, Cowork, Codex, Cursor, and MCP-speaking agents
Source domain: browseros.com
Pricing shape: Free; no pricing page was available at the provided URL