Scale computer-use 2.0 with open-source drivers, cross-OS sandboxes, and benchmarks for training, evaluation, and data generation.
Rust agentai-agentapplecomputer-usecomputer-use-agentcontainerizationcuadesktop-automationhacktoberfestlumemacosmanusoperatorswiftvirtualizationvirtualization-frameworkwindowswindows-sandbox
README: lead with Cua Spaces, retire Cua Fleets (#4417)
Francesco Bonacci989cc76cec26
diff --git a/README.md b/README.mdindex 162e9650eb..1ca67fc967 100644--- a/README.md+++ b/README.md@@ -7,9 +7,9 @@</picture></a>- <p align="center"><strong>Give AI agents computers they can use.</strong><br>Cua provides open-source desktop automation, isolated cloud desktops, local macOS VMs, specialist decision models, and benchmarks for evaluating computer-use agents.</p>+ <p align="center"><strong>Give AI agents computers they can use.</strong><br>Cua Spaces gives your agents full desktops on your Mac and on machines you own. This repository also holds Cua Driver for desktop automation, Lume for local VMs, CUA-S1 decision models and Cua Bench for evaluating computer-use agents.</p>- <p align="center"><strong><a href="https://run.cua.ai/?utm_source=github&utm_medium=referral&utm_campaign=fleet_activation&content_id=repo_readme" target="_blank" rel="noopener noreferrer">Try Cua Fleets now at run.cua.ai</a></strong></p>+ <p align="center"><strong><a href="#cua-spaces">Get Cua Spaces for macOS</a></strong></p><p align="center"><a href="https://cua.ai" target="_blank" rel="noopener noreferrer"><img src="https://img.shields.io/badge/cua.ai-0ea5e9" alt="cua.ai"></a>@@ -28,8 +28,8 @@<table width="100%"><tr><td colspan="2" align="center" valign="top" width="66.66%">- <a href="#sandboxes">- <img src="img/card-cua-fleets-wide.gif" alt="Cua Fleets: isolated cloud desktops for your agents" width="100%">+ <a href="#cua-spaces">+ <img src="img/card-cua-spaces-wide.gif" alt="Cua Spaces: full desktops for your agents" width="100%"></a></td><td align="center" valign="top" width="33.33%">@@ -58,96 +58,124 @@</table></div>-- **Cua Fleets:** [Provision a Linux desktop, run a command, and save a screenshot](https://cua.ai/docs/fleets/quickstart).-- **Cua SDK, CLI and Spaces:** [Install `cua`, create a sandbox locally or in the cloud, and share desktops as Spaces](#cua-sdk-cli-and-spaces).-- **CUA-S1:** [Explore small, specialized models for computer-use decisions](#cua-s1).+- **Cua Spaces:** [Install the app and run your first Space](https://cua.ai/docs/spaces/quickstart).- **Cua Driver:** [Operate Calculator and verify its result](https://cua.ai/docs/cua-driver/quickstart).- **Lume:** [Create a Tahoe VM and connect over SSH](https://cua.ai/docs/lume/quickstart).+- **Cua SDK and CLI:** [Install `cua` and create a local sandbox](#cua-sdk-and-cli).+- **CUA-S1:** [Explore small, specialized models for computer-use decisions](#cua-s1).- **Cua Bench:** [Create and verify a simulated task](https://cua.ai/docs/cua-bench/quickstart).-Bring your own agent and model, or explore CUA-S1 for specialized decisions. Cua provides the computer and automation tools. [Computer-Use 2.0](https://cua.ai/docs/cua-driver/concepts/what-is-computer-use) describes an agent moving between code, APIs, and graphical interfaces within the same task.--## See Cua Driver in action+Bring your own agent and model, or explore CUA-S1 for specialized decisions. Cua provides the computer and automation tools. [Computer-Use 2.0](https://cua.ai/docs/cua-driver/concepts/how-cua-driver-works) describes an agent moving between code, APIs, and graphical interfaces within the same task.-Two Cua Driver sessions select cells in LibreOffice Calc and objects in Inkscape on an Omarchy desktop while a terminal stays in the foreground. Watch the 50-second demo, then explore [Omarchy on Fleet](https://cua.ai/docs/cua-sdk/guides/examples/run-omarchy-on-cloud-fleet).+----https://github.com/user-attachments/assets/b4e5517c-d2db-4758-b4cf-07131b0753b2+## Cua Spaces----+Cua Spaces is a desktop app that gives your agents full desktops. Each desktop is a Space: a macOS VM built locally on your Mac, or a Linux or Omarchy image. Spaces run on your Mac, on other machines you own, and in [your own cloud account](https://cua.ai/docs/cua-sdk/guides/your-cloud) (AWS, Google Cloud or Modal), and the app keeps them in your menu bar and notch.-## Sandboxes+- **Teleport.** Move a signed-in app, such as Chrome or Slack, into a Space and it opens there still signed in. Your sessions stay in the Cua Keyvault, encrypted on your Mac, and reach a Space only after you approve.+- **Multiplayer.** You and your agents work on the same desktop, each with your own cursor. Step in to make a choice, then hand the desktop back.+- **Agent-ready images.** Spaces images ship with [cua-spacesd](libs/cua-spacesd/README.md), so agents get processes, files, screenshots and input as soon as a Space starts.-Run any image as an isolated computer or container, on your machine or in the cloud at [run.cua.ai](https://run.cua.ai/?utm_source=github&utm_medium=referral&utm_campaign=fleet_activation&content_id=repo_readme). Only where it runs changes:+https://github.com/user-attachments/assets/a2ccc86b-10d0-48c0-bee6-7ae652ba1c59-```python-from cua_sandbox import Image, Sandbox+**Install on macOS 26 or later**-async with Sandbox.ephemeral(Image.linux(), local=True) as sb: # local=False for the cloud- print((await sb.shell.run("uname -a")).stdout)+```sh+curl -fsSL https://cua.ai/install.sh | sh```-Commands, environment, named services, readiness probes, public URLs, port forwards and MCP servers work the same on both. See the [backend notes](https://cua.ai/docs/cua-sdk/reference/sandbox/runtime-support) for limits.+On macOS the installer selects the Cua Spaces app by default and adds the `cua` CLI. You can also [download the signed `.dmg`](https://github.com/trycua/cua/releases/download/cua-spaces-v0.1.0/cua-spaces-0.1.0-darwin-universal.dmg), or get the `.pkg` from the [Cua Spaces 0.1.0 release](https://github.com/trycua/cua/releases/tag/cua-spaces-v0.1.0).++Spaces is free for individuals. Pro and Teams plans are coming soon. The app is source-available under [FSL-1.1-MIT](#licensing).-**[Sandboxes](https://cua.ai/docs/cua-sdk)** | **[Quickstart](https://cua.ai/docs/cua-sdk/quickstart)** | **[Your first cloud sandbox](https://cua.ai/docs/fleets/quickstart)** | **[Sandbox SDK reference](https://cua.ai/docs/cua-sdk/reference/sandbox)**+**[Quickstart](https://cua.ai/docs/spaces/quickstart)** | **[Teleport an app](https://cua.ai/docs/spaces/guides/teleport-an-app)** | **[Share a Space](https://cua.ai/docs/spaces/guides/share-a-space)** | **[Host Spaces on a spare Mac (relay, Tailscale or SSH)](https://cua.ai/docs/start-here/host-spaces-on-your-spare-mac)** | **[Your own cloud](https://cua.ai/docs/cua-sdk/guides/your-cloud)** | **[App source](apps/cua-spaces-macos/README.md)**----## Cua SDK, CLI and Spaces+## Cua Driver-One SDK and one `cua` command for every sandbox: cloud desktops, local VMs and containers, or any machine that runs [cua-spacesd](libs/cua-spacesd/README.md).+Give your agent tools to inspect and operate native desktop apps and browsers on macOS, Windows, and Linux. Connect through the CLI, MCP, or typed SDKs. Background delivery lets agents work without moving your pointer or taking focus when the app and platform support it; see [platform support](https://cua.ai/docs/cua-driver/concepts/platform-support) for the boundaries.**macOS / Linux**```sh-curl -fsSL https://cua.ai/install.sh | sh+/bin/bash -c "$(curl -fsSL https://cua.ai/driver/install.sh)"```**Windows (PowerShell)**```powershell-irm https://cua.ai/install.ps1 | iex+irm https://cua.ai/driver/install.ps1 | iex```-In a terminal the script shows a short checklist: the `cua` CLI, the Cua Spaces app (default on macOS), the cua-driver MCP and skill for your agents, and hosting this machine. It then runs `cua auth login`: it signs you in and offers to install cua skills and the cua MCP server into your AI coding agents (Claude Code, Codex, Cursor, and others). Preselect items with `sh -s -- --select cua-driver`, or skip the checklist with `--only cua-driver`. The Spaces app installers (dmg/pkg, msi/exe, deb/AppImage) do the same graphically. See [the installer options](scripts/install/README.md).+**Your first result:** connect your agent, ask it to compute 6 × 7 in Calculator, and have it verify that the app displays 42. The tutorial covers platform setup, permissions, and agent connection.-```sh-cua sb create ubuntu --name dev # local: a gVisor container, set up on first use-cua sb exec dev uname -a-cua sb screenshot dev-cua sb rm dev-cua sb create ubuntu --on cloud # the same image in the Cua cloud-cua config set default.on cloud # make the cloud the default (--on local still works)+**[Drive your first app](https://cua.ai/docs/cua-driver/quickstart)** | **[Installation](https://cua.ai/docs/cua-driver/quickstart)** | **[CLI Reference](https://cua.ai/docs/cua-driver/reference/cli)**++Using Claude Code, Codex, Cursor, OpenClaw, or another agent? [Find your integration](https://cua.ai/docs/cua-driver/guides/connect-your-agent). Source documentation and architecture notes live in [`libs/cua-driver/README.md`](libs/cua-driver/README.md).++### See Cua Driver in action++Two Cua Driver sessions select cells in LibreOffice Calc and objects in Inkscape on an Omarchy desktop while a terminal stays in the foreground. Watch the 50-second demo.++https://github.com/user-attachments/assets/b4e5517c-d2db-4758-b4cf-07131b0753b2++---++## Lume++Create and manage local macOS and Linux VMs on Apple Silicon using Apple's Virtualization.Framework.++```bash+/bin/bash -c "$(curl -fsSL https://cua.ai/lume/install.sh)"```-- **SDK:** the same API in Python (`pip install cua`), TypeScript (`@trycua/cua`), Swift (`Cua`) and Kotlin (generated bindings), running embedded in your process or through a shared `cua daemon`.-- **Sandboxes need no agent inside.** Readiness comes from the runtime and optional port probes. Images that ship cua-spacesd (port 3211) add processes, files, screenshots, input through cua-driver, and low-latency video and audio streaming.-- **Spaces** are sandboxes or machines with cua-spacesd that you and your agents share: live desktop and window streams, presence, file send, teleporting an app session from your machine, and agent threads. `cua host setup` makes this machine reachable through the cua.ai relay with no port forwarding.+**Your first result:** create a vanilla macOS Tahoe VM from an Apple restore image, start it, and connect over SSH. The tutorial uses the Lume CLI directly and explains the unattended setup defaults.-**[SDK README](libs/cua/README.md)** | **[CLI reference](https://cua.ai/docs/cua-cli/reference/cli)** | **[Example Spaces app](https://cua.ai/docs/spaces/examples/cua-bots)**+**[Create your first Lume VM](https://cua.ai/docs/lume/quickstart)** | **[Installation](https://cua.ai/docs/lume/quickstart)** | **[CLI reference](https://cua.ai/docs/lume/reference/cli)**----## Cua Driver+## Cua SDK and CLI-Give your agent tools to inspect and operate native desktop apps and browsers on macOS, Windows, and Linux. Connect through the CLI, MCP, or typed SDKs. Background delivery lets agents work without moving your pointer or taking focus when the app and platform support it; see [platform support](https://cua.ai/docs/cua-driver/concepts/platform-support) for the boundaries.+One SDK and one `cua` command for local VMs and containers, and for any machine that runs [cua-spacesd](libs/cua-spacesd/README.md).**macOS / Linux**```sh-/bin/bash -c "$(curl -fsSL https://cua.ai/driver/install.sh)"+curl -fsSL https://cua.ai/install.sh | sh```**Windows (PowerShell)**```powershell-irm https://cua.ai/driver/install.ps1 | iex+irm https://cua.ai/install.ps1 | iex```-**Your first result:** connect your agent, ask it to compute 6 × 7 in Calculator, and have it verify that the app displays 42. The tutorial covers platform setup, permissions, and agent connection.+In a terminal the script shows a short checklist: the `cua` CLI, the Cua Spaces app (default on macOS), the cua-driver MCP and skill for your agents, and hosting this machine. It then runs `cua auth login`, which signs you in and offers to install cua skills and the cua MCP server into your AI coding agents (Claude Code, Codex, Cursor, and others). Preselect items with `sh -s -- --select cua-driver`, or skip the checklist with `--only cua-driver`. See [the installer options](scripts/install/README.md).++```sh+cua sb create ubuntu --name dev # a gVisor container, set up on first use+cua sb exec dev uname -a+cua sb screenshot dev+cua sb rm dev+```-**[Drive your first app](https://cua.ai/docs/cua-driver/quickstart)** | **[Installation](https://cua.ai/docs/cua-driver/guides/install)** | **[CLI Reference](https://cua.ai/docs/cua-driver/reference/cli)**+The same sandbox from Python:-Using Claude Code, Codex, Cursor, OpenClaw, or another agent? [Find your integration](https://cua.ai/docs/cua-driver/guides/connect-your-agent). Source documentation and architecture notes live in [`libs/cua-driver/README.md`](libs/cua-driver/README.md).+```python+from cua_sandbox import Image, Sandbox++async with Sandbox.ephemeral(Image.linux(), local=True) as sb:+ print((await sb.shell.run("uname -a")).stdout)+```++- **SDK:** the same API in Python (`pip install cua`), TypeScript (`@trycua/cua`), Swift (`Cua`) and Kotlin (generated bindings), running embedded in your process or through a shared `cua daemon`.+- **Sandboxes need no agent inside.** Readiness comes from the runtime and optional port probes. Images that ship cua-spacesd (port 3211) add processes, files, screenshots, input through cua-driver, and low-latency video and audio streaming.+- **Your own machines.** `cua host setup` makes this machine reachable through the cua.ai relay with no port forwarding.++**[SDK README](libs/cua/README.md)** | **[Quickstart](https://cua.ai/docs/cua-sdk/quickstart)** | **[CLI reference](https://cua.ai/docs/cua-cli/reference/cli)** | **[Sandbox SDK reference](https://cua.ai/docs/cua-sdk/reference/sandbox)**---@@ -165,20 +193,6 @@ The project includes Python model code, synthetic-data generation, training, and----## Lume--Create and manage local macOS and Linux VMs on Apple Silicon using Apple's Virtualization.Framework.--```bash-/bin/bash -c "$(curl -fsSL https://cua.ai/lume/install.sh)"-```--**Your first result:** create a vanilla macOS Tahoe VM from an Apple restore image, start it, and connect over SSH. The tutorial uses the Lume CLI directly and explains the unattended setup defaults.--**[Create your first Lume VM](https://cua.ai/docs/lume/quickstart)** | **[Installation](https://cua.ai/docs/lume/quickstart)** | **[CLI reference](https://cua.ai/docs/lume/reference/cli)**------## Cua BenchBuild computer-use tasks, evaluate agents, and export trajectories for training. Start with a simulated task that requires no VM, Docker, or model API key.@@ -201,13 +215,14 @@ uv tool run --from 'cua-bench[browser]' playwright install chromium| Package | Description || ------------------------------------------------------- | ------------------------------------------------------------------------------------ || [cua-driver](libs/cua-driver/README.md) | Background computer-use agent for macOS, Windows, and Linux |-| [cua SDK and CLI](libs/cua/README.md) | Rust core and the `cua` command: sandboxes, Fleet, local runtimes, images, Spaces |+| [cua SDK and CLI](libs/cua/README.md) | Rust core and the `cua` command: sandboxes, local runtimes, images, Spaces || [cua (Python)](libs/cua/python/README.md) | The cua SDK for Python (`pip install cua`) || [@trycua/cua](libs/cua/typescript/README.md) | The cua SDK for Node and the browser, plus `@trycua/cua/spaces` || [Cua (Swift)](libs/cua/swift/README.md) | The cua SDK for Swift (SwiftPM, XCFramework) || [cua-sandbox](libs/python/cua-sandbox/README.md) | High-level Python `Sandbox`/`Image`/`Pool` API, a thin wrapper over the cua SDK || [cua-spacesd](libs/cua-spacesd/README.md) | In-sandbox daemon on port 3211: processes, files, desktop, streaming (gRPC) |-| [Cua Spaces](apps/cua-spaces/README.md) | Desktop app for Spaces: live streams, PiP, teleport, agent threads |+| [Cua Spaces for macOS](apps/cua-spaces-macos/README.md) | The Spaces app on macOS: menu bar and notch, live streams, teleport, agent threads |+| [Cua Spaces core](apps/cua-spaces/README.md) | The shared Spaces app core, and the Tauri app for Linux and Windows || [cua-agent](libs/python/agent/README.md) | AI agent framework for computer-use tasks || [cua-bench](libs/cua-bench/README.md) | Benchmarks and RL environments for computer-use || [lume](https://cua.ai/docs/lume/reference/cli) | macOS/Linux VM management on Apple Silicon |@@ -241,11 +256,11 @@ For reproducibility, include the Cua release or commit used in your experiments.## Contributing-We welcome contributions! See our [Contributing Guidelines](CONTRIBUTING.md) for details.+We welcome contributions. See our [Contributing Guidelines](CONTRIBUTING.md) for details.## Licensing-Cua is open source under the MIT License ([LICENSE](LICENSE.md)): the cua SDK and its Python, TypeScript, Swift and Kotlin packages, the `cua` command and `cua daemon`, Cua Driver, Lume and the rest of this repository unless a directory says otherwise.+Everything outside Cua Spaces is MIT-licensed ([LICENSE](LICENSE.md)): the cua SDK and its Python, TypeScript, Swift and Kotlin packages, the `cua` command and `cua daemon`, Cua Driver, Lume and the rest of this repository unless a directory says otherwise.Cua Spaces is source-available under [FSL-1.1-MIT](https://fsl.software): the Spaces apps, cua-spacesd, the Cua Keyvault, teleport, Cua Volume (`cua-volume`) and the streaming client, codecs and viewers. The streaming wire protocol stays MIT. It is free to use, self-host and build on, with no competing hosted service, and each release becomes MIT two years after it ships.The MIT parts never depend on the FSL parts. Some subdirectories carry their own licence; [LICENSING.md](LICENSING.md) lists each one and how the two fit together.Offering Spaces as a hosted or managed service? See [COMMERCIAL.md](COMMERCIAL.md). For use of our names and logo, see [TRADEMARKS.md](TRADEMARKS.md).@@ -281,7 +296,7 @@ Apple, macOS, Ubuntu, Canonical, and Microsoft are trademarks of their respectiv<div align="center">-Thank you to all our [GitHub Sponsors](https://github.com/sponsors/trycua)!+Thank you to all our [GitHub Sponsors](https://github.com/sponsors/trycua).| [Adam Cohen Hillel](https://github.com/adamcohenhillel) | [CodeRabbit](https://github.com/coderabbitai) | [Zephyr Cloud IO](https://github.com/ZephyrCloudIO) || :------------------------------------------------------------------------------------------------------------------------------------------------------------------: | :-----------------------------------------------------------------------------------------------------------------------------------------------------------: | :-----------------------------------------------------------------------------------------------------------------------------------------------------------------: |diff --git a/img/card-cua-bench.gif b/img/card-cua-bench.gifindex 2a3beb19ec..e9677a76c8 100644Binary files a/img/card-cua-bench.gif and b/img/card-cua-bench.gif differdiff --git a/img/card-cua-driver.gif b/img/card-cua-driver.gifindex c4673f03a3..a804809923 100644Binary files a/img/card-cua-driver.gif and b/img/card-cua-driver.gif differdiff --git a/img/card-cua-fleets-wide.gif b/img/card-cua-fleets-wide.gifdeleted file mode 100644index 04397734f0..0000000000Binary files a/img/card-cua-fleets-wide.gif and /dev/null differdiff --git a/img/card-cua-lume.gif b/img/card-cua-lume.gifindex 1aea6c0170..adc751d4c4 100644Binary files a/img/card-cua-lume.gif and b/img/card-cua-lume.gif differdiff --git a/img/card-cua-s1.gif b/img/card-cua-s1.gifindex 3506a4c1b5..12a7b47ac6 100644Binary files a/img/card-cua-s1.gif and b/img/card-cua-s1.gif differdiff --git a/img/card-cua-spaces-wide.gif b/img/card-cua-spaces-wide.gifnew file mode 100644index 0000000000..a59aef3e23Binary files /dev/null and b/img/card-cua-spaces-wide.gif differdiff --git a/img/cua-spaces-demo.mp4 b/img/cua-spaces-demo.mp4new file mode 100644index 0000000000..0591b080fbBinary files /dev/null and b/img/cua-spaces-demo.mp4 differ