# Agent TARS — CLI Desktop UI Automation Agent

> Agent TARS is a CLI-first desktop automation agent with browser support and MCP integration, so you can script real UI workflows without glue scripts.

## Install

Copy the content below into your project:

## Quick Use

1. Install (Node.js >= 22):
   ```bash
   npm install @agent-tars/cli@latest -g
   ```
2. Launch the CLI:
   ```bash
   npx @agent-tars/cli@latest
   ```
3. Verify:
   - Open the Quick Start docs and complete one guided run end-to-end.

## Intro

Agent TARS is a CLI-first desktop automation agent with browser support and MCP integration, so you can script real UI workflows without glue scripts.

- **Best for:** developers who want a repeatable, CLI-driven way to automate desktop/browser UI tasks
- **Works with:** Node.js, your chosen model provider, desktop/browser targets; can mount MCP servers
- **Setup time:** 20–45 minutes

## Practical Notes

- Quant: start with 1 UI task (login/search/export) and measure success rate over 10 runs.
- Quant: track setup time (Node + provider key) and typical run latency per task.

## How to use it safely

Desktop automation is powerful and risky. Treat Agent TARS like a privileged robot:

- Start in a disposable test account and a non-production environment.
- Make the agent produce *observable checkpoints* (e.g., screenshots, extracted text, saved files) after each major step.
- Keep a narrow allowlist of domains/apps, and require confirmation before any destructive action.

## Where it shines

Use it when the UI is the API: internal admin panels, vendor dashboards, and legacy tools that have no stable endpoint.

If you already have an API, prefer API automation first and use UI automation only for gaps.

### FAQ

**Q: Do I need a specific model provider?**
A: No. The CLI supports multiple providers; pick one you already use and budget for.

**Q: How do I keep runs reproducible?**
A: Pin the prompt/plan, log each step, and replay the same task 10 times before expanding scope.

**Q: When should I avoid UI automation?**
A: Avoid it for critical, irreversible actions in production; use APIs or add human confirmation gates.

## Source & Thanks

> Source: https://github.com/bytedance/UI-TARS-desktop
> License: Apache-2.0
> GitHub stars: 33,274 · forks: 3,300

---

<!-- ZH -->

## 快速使用

1. 安装（需要 Node.js >= 22）：
   ```bash
   npm install @agent-tars/cli@latest -g
   ```
2. 启动 CLI：
   ```bash
   npx @agent-tars/cli@latest
   ```
3. 验证：
   - 按 Quick Start 文档跑通一次完整流程（含一次真实 UI 任务）。

## 简介

Agent TARS 提供 CLI 优先的桌面自动化 Agent，支持浏览器/桌面混合执行并可挂载 MCP 工具，让你把真实 UI 流程写成可复用、可回放的脚本，减少零散胶水代码与手工操作。

- **适合谁：** 想用 CLI 把桌面/浏览器 UI 任务做成可复用自动化流程的开发者
- **可搭配：** Node.js、你选择的模型供应商、桌面/浏览器目标；可挂载 MCP servers
- **准备时间：** 20–45 分钟

## 实战建议

- 量化建议：先落地 1 个 UI 任务（登录/搜索/导出），连续跑 10 次统计成功率。
- 量化建议：记录准备时间（Node + provider key）与每次任务的典型耗时/延迟。

## 安全使用方式

桌面自动化既强大也有风险，把 Agent TARS 当作“高权限机器人”来管理：

- 先在测试账号与非生产环境里验证。
- 让 agent 在关键步骤产出可观察的检查点（截图/提取文本/落盘文件）。
- 维护域名/应用白名单；遇到潜在破坏性动作必须二次确认。

## 适用场景

当 UI 就是唯一入口时最有价值：内部后台、供应商控制台、没有稳定 API 的老系统。

如果已有 API，优先用 API 自动化；UI 自动化只补缺口。

### FAQ

**必须用特定模型供应商吗？**
答：不需要。CLI 支持多种 provider，选择你现有的并做好预算评估即可。

**如何保证可复现？**
答：固定 prompt/计划，记录每一步证据，并在扩展前把同一任务重复跑 10 次。

**什么时候不适合用 UI 自动化？**
答：生产环境里不可逆/高风险操作尽量不要全自动；优先 API 或加人工确认闸门。

## 来源与感谢

> Source: https://github.com/bytedance/UI-TARS-desktop
> License: Apache-2.0
> GitHub stars: 33,274 · forks: 3,300


---
Source: https://tokrepo.com/en/workflows/agent-tars-cli-desktop-ui-automation-agent
Author: Script Depot