# mcp-server-browser [Health: Active]

**Category:** 📂 Browser Automation  
**Repository:** https://github.com/bytedance/UI-TARS-desktop  
**GitHub Stars:** 39369  
**npm Downloads (last month):** 1886  
**Views:** 2  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/mcp-server-browser

## Description
The Open-Source Multimodal AI Agent Stack: Connecting Cutting-Edge AI Models and Agent Infra

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `npx` (confidence: high):

```json
"mcpServers": {
  "mcp-server-browser": {
    "command": "npx",
    "args": ["-y","@agent-tars/cli@latest"]
  }
}
```

## Documentation

## What mcp-server-browser MCP server does

The provided source repository is the home of TARS, a multimodal AI agent stack containing Agent TARS and UI-TARS Desktop. Its documented focus is visual and GUI-based task completion across terminals, computers, browsers, and products. Agent TARS is exposed through a command-line interface and a Web UI, while UI-TARS Desktop is a native desktop application built around the UI-TARS model.

For browser-oriented work, the repository describes local and remote browser operators. It also describes local and remote computer operators. These capabilities are intended for agents that need to interact with visible interfaces, such as completing booking workflows or working with browser pages.

## How it works

Agent TARS combines multimodal models, GUI-agent behavior, and vision-oriented interaction. Its primary usage surfaces are a CLI and Web UI, with support for both headful Web UI execution and headless server execution. UI-TARS Desktop supplies a native interface and operators for local or remote computer and browser control.

The repository’s examples show agents receiving natural-language instructions for tasks such as searching for flights, booking a hotel, and preparing a transportation guide. Another example uses additional MCP servers to generate a weather chart. These examples indicate that browser operation and external MCP tools can participate in broader agent workflows, but they do not define the individual tools exposed by the mcp-server-browser MCP server.

## Setup and configuration

It also references model setup for local and remote operation.

Consult the project’s linked documentation before choosing a deployment method or relying on a particular MCP client.

## Tools and capabilities

Documented capabilities include:

- Browser operation through local and remote browser operators.
- Computer operation through local and remote computer operators.
- GUI-agent interaction supported by vision and multimodal models.
- Agent TARS CLI and Web UI usage modes.
- Headful Web UI and headless server execution.
- Integration with additional real-world MCP tools through Agent TARS.

## Limitations and notes

The repository name and README describe a broader multimodal agent stack, not a clearly documented standalone browser MCP server. The name mcp-server-browser MCP server therefore should not be treated as proof of a published package or a fixed set of MCP tools. Browser and computer control are documented at the UI-TARS Desktop and Agent TARS feature level.

Those details require verification in the linked project documentation.

_Full upstream README: https://allmcps.com/mcp/mcp-server-browser/readme_