What it does
Agent S is a "computer use" agent. You describe a task in plain words, and it looks at the screen and completes the task by moving the mouse, clicking and typing in ordinary desktop and web apps.
It works on Mac, Windows and Linux and can use AI models from OpenAI, Anthropic and open-model providers. Its makers say their latest version scored higher than humans on a well-known test of computer tasks.
The README warns that the agent runs code to control your computer, so it should be used with care. The company behind it also offers a hosted version.
Who it's for
Researchers and developers exploring AI that operates a computer, and technical users who want to automate desktop tasks.
Can I use it without coding?
Developers only. It is installed as a Python package and run from the command line with API keys set up. New to this? Start here.
Can I run it?
- Cost
- Free, but needs an AI accountThe software is free, but it uses an online AI service (like OpenAI or Anthropic), which usually charges for use.
- Runs on
- Mac, Windows, Linux
- Computer
- An ordinary computer is fine
- Privacy
- Your choice: run it on your device or use an online service
- API key
- Needed. An API key is a password-like code from an online AI service. What's that?
Try it
Using Claude Code, Cursor or another AI coding assistant? Paste the prompt and it will walk you through the setup.New to this?
Before you install
- Only download from the official GitHub page. Copies with similar names exist for many popular projects.
- License (Apache-2.0): Permissive: free to use, change and share, including for business.
- Keep your API key private. Never paste it into websites or share it, and set a spending limit with the AI service.
- More on staying safe
Where people are talking about it
- Scaling Agents for Computer UseResearch paper · Oct 2025
- Agent S2 paperResearch paper · Apr 2025
- Agent S paperResearch paper · Oct 2024