# Cua

> Open computer-use drivers, sandboxes, MCP, and CUA-S1 decision models

**Category:** AI Agents  
**Pricing:** open-source  
**URL:** https://cua.ai  
**GitHub:** https://github.com/trycua/cua  
**Added:** 2026-09-20

---

## The Hook

Cua gives agents a computer-use control plane: OS and desktop drivers, sandboxes, MCP, and CUA-S1 decision models for acting beyond a browser tab.

## What's Good

MIT (trycua/cua). The stack covers computer-use drivers across OS/desktop surfaces, local sandboxes, MCP integration, and the CUA-S1 model line when an agent needs a decision loop rather than a text-only tool call. It is a useful bridge from browser automation to the wider desktop.

## The Catch

Vs BrowserSkill, Stagehand, and Browser Use: Cua is the broader computer-use driver/sandbox/model stack, not only browser automation or a browser-session attach bridge. Optional `omni` extras pull AGPL Ultralytics; cloud fleets are paid. Audit extras and treat a fleet as infrastructure spend, not as part of the MIT core.

## Verdict

MIT computer-use stack with CUA-S1. Reach for desktop/OS control and sandboxes; audit the AGPL omni extra and paid cloud fleet boundary.