#Quick Start

#Introduction

#Prerequisites

#Node.js

Agent TARS CLI requires Node.js as the runtime, you will need to install Node.js >= version 22, it is recommended to use the Node.js LTS version.

Check the current Node.js version with the following command:

node -v

If you do not have Node.js installed in current environment, or the installed version is too low, you can use nvm to install.

Here is an example of how to install via nvm:

# Install Node.js LTS
nvm install --lts
# Switch to Node.js LTS
nvm use --lts

#Chrome

By default, Agent TARS CLI launches and controls your local browser, you'll need install Chrome.

#Installation

Install latest version:

npm install @agent-tars/cli@latest -g

TIP

Agent TARS is under rapid development, and its version iteration follows Semantic Version. You can install the current beta version using @next:

npm install @agent-tars/cli@next -g

Current version information:

Tag	Version
Latest
Next

#Quick Start

#1. Choose a Model

The Agent TARS framework has designed a Model Provider mechanism that allows you to freely use different models.

#Overview

The current model compatibility status for Agent TARS is as follows:

#Seed 1.5 VL ByteDance

Seed1.5-VL is a powerful and efficient vision-language foundation model designed for advanced general-purpose multimodal understanding and reasoning, Seed1.5-VL has been deployed on Volcano Engine, The Model ID is doubao-1-5-thinking-vision-pro-250428.

Once you obtain the API_KEY, you can start Agent TARS with a single command:

agent-tars \
--provider volcengine \
--model doubao-1-5-thinking-vision-pro-250428 \
--apiKey {apiKey}

#claude-3.7-sonnet Anthropic

Claude 3.7 Sonnet is the first Claude model with hybrid reasoning capabilities released by Anthropic in February 2025. Once you obtain the API_KEY, you can quickly start Agent TARS:

agent-tars \
--provider anthropic \
--model claude-3-7-sonnet-latest \
--apiKey {apiKey}

#gpt-4o OpenAI

GPT-4o is high-intelligence flagship model shipped by OpenAI, once you obtain the API_KEY, you can start with a single command:

agent-tars \
--provider openai \
--model gpt-4o \
--apiKey {apiKey}

For more details about model support, please move Model Provider.

#2. Start Your First Task

When you start with agent-tars:

agent-tars [...flags]

You will see the following output in the console:

Open the link in the console: http://localhost:8888 , you can see the Web UI:

Enter this prompt:

Tell me the top 10 for Humanity's Last Exam

Congratulations 🎉! At this point, you have successfully started Agent TARS!

#3. Create a global workspace

While you can start quickly using the CLI, we still recommend creating a Global Workspace. This allows you to maintain configurations via config files, as well as store your File System. Let's get started:

agent-tars workspace --init   # Follow the prompts to complete creation
agent-tars workspace --open   # Open the Workspace

Next, you can maintain your configuration in the Global Workspace using TypeScript instead of dealing with CLI parameters, and benefit from complete type checking:

// agent-tars.config.ts
import { defineConfig } from '@agent-tars/interface';

/**
 * @see {@link https://beta.agent-tars.com/api/config/agent.html}
 */
export default defineConfig({
  model: {
    provider: 'volcengine',
    // ... other configs
  },
  // ... other configs
});

For complete configuration details, please go to Config, and for workspace, please go to Workspace.

#Next Step