Open-source infrastructure for Computer-Use Agents. Sandboxes, SDKs, and benchmarks to train and evaluate AI agents that can control full desktops (macOS, Linux, Windows).

Agent S: an open agentic framework that uses computers like a human

A research prototype of a human-centered web agent

The first open-source Artificial Narrow Intelligence generalist agentic framework Computer-Using-Agent that fully operates graphical-user-interfaces (GUIs) by using only natural language. Uses Visualization-of-Thought and Chain-of-Thought reasoning to elicit spatial reasoning and perception, emulates, plans and simulates synthetic HID interactions.

[NeurIPS'25 D&B] Mind2Web-2 Benchmark: Evaluating Agentic Search with Agent-as-a-Judge

This is the official website for TuriX Computer-use-Agent

A lightweight Python library to create, run and manage macOS and Linux virtual machines (VMs) natively on Apple Silicon.

Opensource benchmark evaluating web operators/agents performance

Browser using agent (bua) notte model on any browser environment

A Human-Centered Agentic System for Validating Research Papers

CoAct-1: Computer-using Agents with Coding as Actions

Agent-CE is a containerized continuous evaluation (CE) platform for web browsing agents. It provides production-ready Docker images and CI/CD pipelines for running and evaluating multiple agent frameworks including Browser Use, Notte, Anthropic Computer Use, and OpenAI Computer Use.

Your agent in your terminal, equipped with local tools: writes code, uses the terminal, browses the web, vision.