# Cua Documentation > Cua is an open-source platform for computer-use automation: AI agents that operate real computers by clicking, typing, and reading the screen and accessibility tree. Drive a machine you already have with Cua Driver, provision hosted desktops with Cloud Fleets and the Sandbox SDK, or use Lume as local Apple Silicon/macOS VM infrastructure underneath your own workflows. Full text of every page: https://cua.ai/docs/llms-full.txt ## Start here - [Choose your path](https://cua.ai/docs.md): Open-source computer-use automation for real machines and isolated desktops. - [Your first Cloud Fleet](https://cua.ai/docs/tutorials/your-first-cloud-fleet.md): Provision a Cloud Fleet, claim an isolated Linux desktop, run a command, save a screenshot, and clean up the cloud resources. - [Drive your first app with Cua Driver](https://cua.ai/docs/tutorials/drive-your-first-app.md): Install Cua Driver, connect your agent, and verify a result in a desktop app on macOS, Windows, or Linux. - [Create a local macOS VM with Lume](https://cua.ai/docs/tutorials/create-your-first-lume-vm.md): Install Lume and create a vanilla macOS Tahoe VM from an Apple restore image. - [Build your first Cua-Bench task](https://cua.ai/docs/tutorials/your-first-cua-bench-task.md): Create, inspect, solve, and evaluate a simulated computer-use task. ## Sandboxes & Fleets - [Overview: local execution and Fleet](https://cua.ai/docs/concepts/how-sandboxes-work.md): How a Cua Sandbox gives an agent one isolated computer it can both run code in and drive through the GUI. - [Cloud Fleets](https://cua.ai/docs/cloud-fleets.md): Find tutorials, guides, explanations, and reference for managed cloud sandboxes. ### How-to guides - [Set up Fleet credentials](https://cua.ai/docs/how-to-guides/sandbox/set-up-fleet-credentials.md): Create a Fleet user API key, configure the Sandbox SDK, and check access before provisioning. - [Choose a sandbox image](https://cua.ai/docs/how-to-guides/sandbox/choose-an-image.md): Choose between a published Fleet image, local image customization, and preparing a Fleet boot artifact. - [Manage local sandbox lifecycle](https://cua.ai/docs/how-to-guides/sandbox/manage-local-lifecycle.md): Create, reconnect to, list, and clean up local sandboxes with the Sandbox SDK. - [Configure a sandbox pool with Terraform](https://cua.ai/docs/how-to-guides/sandbox/configure-pool-with-terraform.md): Configure reusable Linux and Windows sandbox pools on run.cua.ai with Terraform. - [Create a sandbox pool with Python](https://cua.ai/docs/how-to-guides/sandbox/create-pool-with-python.md): Create a reusable warm sandbox pool, claim a sandbox, and run commands with the Cua Sandbox SDK. - [Create a sandbox pool with TypeScript](https://cua.ai/docs/how-to-guides/sandbox/create-pool-with-typescript.md): Create a reusable Fleet pool and capture a sandbox screenshot from browser or Node.js TypeScript. - [Run a Cursor Cloud Agent worker on Cua Cloud Fleet](https://cua.ai/docs/how-to-guides/sandbox/run-cursor-cloud-agent-on-fleet.md): Start a Cursor self-hosted Team Pool worker inside a claimed Cua Cloud Fleet Linux desktop. - [Expire pools and claims automatically](https://cua.ai/docs/how-to-guides/sandbox/expire-pools-and-claims.md): Set a creation-age TTL so sandbox pools and claims delete themselves. - [Troubleshoot Fleet pools and claims](https://cua.ai/docs/how-to-guides/sandbox/troubleshoot-fleet-pools.md): Diagnose Fleet pool errors and recover interrupted claim or pool cleanup without provisioning replacements. - [Run OpenClaw on Cloud Fleet](https://cua.ai/docs/how-to-guides/sandbox/run-openclaw-on-cloud-fleet.md): Run OpenClaw's bundled CUA computer-use provider in an isolated Linux desktop on Cua Cloud Fleet. - [Run Omarchy on Fleet](https://cua.ai/docs/how-to-guides/sandbox/run-omarchy-on-cloud-fleet.md): Provision an amd64 Omarchy desktop on Cua Fleet and control it with the Cua Sandbox SDK. - [Connect to an Omarchy Fleet desktop with noVNC](https://cua.ai/docs/how-to-guides/sandbox/connect-to-omarchy-with-novnc.md): Add browser-based desktop access to an Omarchy VM running on Cua Fleet. - [Pass secrets into a sandbox](https://cua.ai/docs/how-to-guides/sandbox/secrets.md): Read secrets from the host and inject them into a sandbox at runtime. - [Run sandboxes in parallel](https://cua.ai/docs/how-to-guides/sandbox/scale-out.md): Run many sandbox jobs at once, limit concurrency, and handle failures. - [Forward a port from a sandbox](https://cua.ai/docs/how-to-guides/sandbox/tunneling.md): Reach a sandbox service through a local startup mapping, a Fleet service request, or a supported tunnel transport. - [Share a sandbox service with a signed URL](https://cua.ai/docs/how-to-guides/sandbox/share-service-with-signed-url.md): Create, list, and revoke time-limited public URLs for sandbox services with the TypeScript Fleet SDK. - [Choose and build a sandbox image](https://cua.ai/docs/how-to-guides/sandbox/images.md): Select a sandbox base image and add packages, files, setup commands, ports, and inspection. - [Prepare and reference a Fleet image](https://cua.ai/docs/how-to-guides/sandbox/prepare-and-reference-a-fleet-image.md): Publish a Fleet-compatible OCI image and reference it from the Sandbox SDK. - [Open an interactive shell in a sandbox](https://cua.ai/docs/how-to-guides/sandbox/interactive-shell.md): Open an interactive PTY session inside the code half of a sandbox from the CLI or Python SDK. - [Run Minecraft in a Windows sandbox](https://cua.ai/docs/how-to-guides/sandbox/minecraft.md): Boot a Windows sandbox, install Minecraft Java Edition, and drive it with an agent through the cua-driver MCP server running inside the sandbox. ### Concepts - [How Fleet images work](https://cua.ai/docs/concepts/how-fleet-images-work.md): How Fleet boot artifacts define the operating system, compatibility boundary, and service contract for managed sandbox capacity. - [Sandbox lifecycle](https://cua.ai/docs/concepts/sandbox-lifecycle.md): How connection lifetime, local sandbox lifetime, and Fleet pool capacity differ. ### Reference - [Sandbox SDK](https://cua.ai/docs/reference/sandbox-sdk.md): Packages, configuration, and Sandbox lifecycle APIs for Python and TypeScript. - [Pool and claim reference](https://cua.ai/docs/reference/sandbox-sdk/pool.md): Python Fleet pool configuration, claim acquisition, cleanup, and resource types in cua-sandbox 0.4.3. - [Sandbox interfaces reference](https://cua.ai/docs/reference/sandbox-sdk/interfaces.md): Python Sandbox computer-control, file, service, tunnel, and application interfaces in cua-sandbox 0.4.3. - [Image reference](https://cua.ai/docs/reference/sandbox-sdk/image.md): API reference for the Image class: the immutable, chainable image specification used to configure sandbox environments. - [OS and image catalog](https://cua.ai/docs/reference/sandbox-sdk/os-image-catalog.md): Sandbox operating systems, Linux distributions, desktop environments, window managers, and image registries. - [TypeScript Fleet reference](https://cua.ai/docs/reference/sandbox-sdk/typescript.md): Entry points, client methods, resource types, and lifecycle contracts for @trycua/fleet 0.1.1. - [Sandbox runtime support](https://cua.ai/docs/reference/sandbox-sdk/runtime-support.md): Versioned image, runtime, customization, and transport support for the Sandbox SDK. ## Cua Driver ### How-to guides - [Install Cua Driver](https://cua.ai/docs/how-to-guides/driver/install.md): Install Cua Driver on macOS, Windows, or Linux with a one-line script. - [Use Cua Driver with](https://cua.ai/docs/use-cua-with.md): Choose a model provider, agent harness, or local runtime for Cua Driver. - [Use Cua Driver in process](https://cua.ai/docs/how-to-guides/driver/use-sdk-in-process.md): Call the typed Cua Driver SDK directly from Python or TypeScript without a daemon. - [Verify a desktop action](https://cua.ai/docs/how-to-guides/driver/verify-a-desktop-action.md): Check a Python or TypeScript desktop action against an independent postcondition. - [Use Cua Driver with Claude Agent SDK](https://cua.ai/docs/how-to-guides/driver/use-claude-agent-sdk.md): Choose same-process native callbacks or the MCP agent boundary in Python and TypeScript. - [Use Cua Driver with Codex SDK](https://cua.ai/docs/how-to-guides/driver/use-codex-sdk.md): Configure Cua Driver MCP for the Python and TypeScript Codex SDKs. - [Expose MCP from a desktop app](https://cua.ai/docs/how-to-guides/driver/expose-mcp-from-desktop-app.md): Host a private Cua Driver daemon when an external agent must reuse your app's desktop permissions. - [Install the Cua Driver agent skill](https://cua.ai/docs/how-to-guides/driver/install-agent-skill.md): Install Cua Driver instructions for OpenClaw from ClawHub, or install them directly for another supported agent. - [Run Cua Driver in a macOS Lume VM](https://cua.ai/docs/how-to-guides/driver/run-in-macos-lume-vm.md): Pull the published Tahoe image, install Cua Driver, grant macOS consent, and connect an agent to the VM desktop. - [Run Cua Driver macOS tests in a Lume VM](https://cua.ai/docs/how-to-guides/driver/run-tests-in-macos-lume-vm.md): Build a private Tahoe test seed, run the complete GUI E2E gate from a disposable Lume clone, and retain its evidence. - [Connect your agent to Cua Driver](https://cua.ai/docs/how-to-guides/driver/connect-your-agent.md): Connect an agent to Cua Driver through MCP or the Cua Driver skill and CLI. - [Run a local model with Cua Driver](https://cua.ai/docs/how-to-guides/driver/run-with-local-model.md): Connect Muse Glimmer to Cua Driver through Claude Code using Ollama or llama.cpp while keeping the model's tool context small. - [Drive a Web Page](https://cua.ai/docs/how-to-guides/driver/drive-a-web-page.md): Bind an exact browser window, inspect its active tab, and navigate, click, or type without foregrounding it. - [Record and render a Cua Driver trajectory](https://cua.ai/docs/how-to-guides/driver/record-and-render-a-trajectory.md): Capture an agent's Cua Driver actions and turn the trajectory into a zoom-on-click MP4. - [Keep Cua Driver running](https://cua.ai/docs/how-to-guides/driver/keep-running.md): Register Cua Driver as a persistent daemon that starts automatically and survives reboots. - [Restrict tool access with permission policies](https://cua.ai/docs/how-to-guides/driver/restrict-tool-access.md): Use YAML or Rego policies to control which MCP tools an agent can call and what arguments it may pass. - [Write a capability manifest](https://cua.ai/docs/how-to-guides/driver/write-a-bounded-manifest.md): Narrow any Cua Driver permission profile to explicit tools and resources. - [Personalize the agent cursor](https://cua.ai/docs/how-to-guides/driver/personalize-cursor.md): Use Cua Driver's semantic default cursor or compile and install a safe custom dotLottie theme. - [Update Cua Driver](https://cua.ai/docs/how-to-guides/driver/update.md): Check whether a newer Cua Driver release is available and install it. - [Use Computer History](https://cua.ai/docs/how-to-guides/driver/use-computer-history.md): Enable, inspect, query, and delete the encrypted Computer History preview in Cua Driver nightly builds. - [Troubleshoot stale macOS permissions](https://cua.ai/docs/how-to-guides/driver/troubleshoot-stale-macos-permissions.md): Reset stale app-scoped TCC grants when Cua Driver reports missing permissions that appear enabled in System Settings. - [Drive a Windows app over SSH](https://cua.ai/docs/how-to-guides/driver/windows-ssh.md): Use Cua Driver to drive GUI apps on a remote Windows machine from an SSH session. - [Automate a legacy Windows app behind a VPN](https://cua.ai/docs/how-to-guides/recipes/automate-a-legacy-windows-app-behind-a-vpn.md): Use Cua Driver on a VPN-connected Windows machine to drive a legacy desktop app that has no API. - [Export contacts overnight](https://cua.ai/docs/how-to-guides/recipes/export-contacts-overnight.md): Extract LinkedIn or X contacts overnight from a logged-in browser session into a clean CSV. - [Build a report in a native app](https://cua.ai/docs/how-to-guides/recipes/build-a-report-in-a-native-app.md): Generate a financial report in Numbers on macOS by driving the native app with Cua Driver. - [Fill a form from a local file](https://cua.ai/docs/how-to-guides/recipes/fill-a-form-from-a-local-file.md): Drive a browser form from a PDF or CSV file on your local machine. ### Concepts - [Computer use: how AI agents operate computers](https://cua.ai/docs/concepts/what-is-computer-use.md): Understand computer use, computer-use agents and models, their observe-decide-act loop, and how they differ from APIs, RPA, and browser automation. - [Choose a Cua Driver integration](https://cua.ai/docs/concepts/choose-a-cua-driver-integration.md): Choose between MCP, the SDK, a private worker, and an app-hosted service based on who owns the runtime. - [SDK, MCP, and process hosting](https://cua.ai/docs/concepts/sdk-mcp-and-hosting.md): How Cua Driver separates its typed application contract, agent protocol, and desktop process identity. - [Best-effort background](https://cua.ai/docs/concepts/the-no-foreground-contract.md): How Cua Driver tries to operate apps without taking focus, moving the cursor, or raising windows, and when it must fall back to foreground. - [Capture and Delivery Modalities](https://cua.ai/docs/concepts/capture-and-delivery-modalities.md): How Cua Driver observes and acts on an app. Perception returns both the accessibility tree and a screenshot; each action chooses its target, ax or px rung, and delivery mode. - [Linux desktops and computer use](https://cua.ai/docs/concepts/linux-desktops-and-computer-use.md): How X11, Wayland, desktop environments, and compositors affect Cua Driver's capture and input routes. - [Browser Targeting and Background Delivery](https://cua.ai/docs/concepts/browser-targeting-and-background-delivery.md): Why browser automation needs an exact native-window-to-tab binding, and how CDP adds a full-background rung to Cua Driver. - [How permission policies work](https://cua.ai/docs/concepts/how-permission-policies-work.md): How the Cua Driver permission policy engine evaluates YAML and Rego policies, the trust model, and the guarantees the engine makes. - [How we continuously validate Cua Driver](https://cua.ai/docs/concepts/how-cua-driver-is-validated.md): How source-built desktop harnesses, independent application oracles, and retained evidence validate Cua Driver between releases. ### Reference - [CLI Reference](https://cua.ai/docs/reference/cua-driver/cli-reference.md): Command-line interface specification for Cua Driver - [SDK reference](https://cua.ai/docs/reference/cua-driver/sdk-reference.md): Python and TypeScript packages, constructors, methods, results, and errors for the typed Cua Driver SDK. - [Telemetry and privacy](https://cua.ai/docs/reference/cua-driver/telemetry.md): Exact Cua Driver telemetry controls, event schema, and privacy boundaries. - [Computer History Agent Integration](https://cua.ai/docs/reference/cua-driver/computer-history.md): Tool discovery, permissions, query semantics, privacy boundaries, and consultation behavior for agent hosts. - [macOS Permissions](https://cua.ai/docs/reference/cua-driver/macos-permissions.md): The macOS-only `cua-driver permissions` command for inspecting and requesting Accessibility and Screen Recording TCC grants. - [App-hosted daemon reference](https://cua.ai/docs/reference/cua-driver/embedding.md): Run a private Cua Driver daemon when a desktop app must expose MCP using its permission identity. - [MCP Tools](https://cua.ai/docs/reference/cua-driver/mcp-tools.md): Reference for every MCP tool Cua Driver exposes - [MCP Tool Notes](https://cua.ai/docs/reference/cua-driver/mcp-tool-notes.md): Cross-cutting MCP tool contracts: shared parameters, required-parameter rules, platform-specific parameters, and the action response shape. - [Browser Profile Attachment](https://cua.ai/docs/reference/cua-driver/browser-profile-attachment.md): Contract, lifetimes, refusals, and platform scope for attaching Cua Driver to an existing Chromium profile. - [Browser semantic snapshots](https://cua.ai/docs/reference/cua-driver/browser-semantic-snapshots.md): Reference for semantic_v2 browser state, refs, scopes, and continuation capabilities. - [Agent action policy](https://cua.ai/docs/reference/cua-driver/action-selection-policy.md): Behavior an agent should follow when choosing element, pixel, page, and foreground actions in Cua Driver. - [Permission modes](https://cua.ai/docs/reference/cua-driver/permission-modes.md): Choose standard, bounded, or unrestricted authorization for Cua Driver. - [Permission policies](https://cua.ai/docs/reference/cua-driver/permission-policies.md): YAML and Rego permission policy schema, environment variable, evaluation rules, and Rego input interface for Cua Driver. - [Interface Contracts](https://cua.ai/docs/reference/cua-driver/contracts.md): The contracts behind the CLI and MCP surfaces: lifecycle sessions, per-call targets, config persistence, and action routing. - [Platform Support](https://cua.ai/docs/reference/cua-driver/platform-support.md): Current operating-system and window-system support, proven surfaces, and known capability boundaries. - [Platform Roadmap](https://cua.ai/docs/reference/cua-driver/platform-roadmap.md): Remaining Cua Driver platform work, evidence gaps, and operating-system boundaries without implied delivery dates. - [Process model](https://cua.ai/docs/reference/cua-driver/process-model.md): How Cua Driver maps one typed surface onto same-process SDK and daemon-backed execution. - [Known Limits](https://cua.ai/docs/reference/cua-driver/limits.md): Documented behavioral limits of Cua Driver and available alternatives - [Development](https://cua.ai/docs/reference/cua-driver/development.md): Source layout, harnesses, and test entry points for contributors working on cua-driver. ## Cua CLI - [CLI reference](https://cua.ai/docs/reference/cua-cli/cli-reference.md): Command reference for the cua CLI: authentication, sandboxes, images, platforms, one-shot automation, skills, and trajectories. - [CLI authentication](https://cua.ai/docs/reference/cua-cli/authentication.md): Authenticate the cua CLI, choose credential storage, and configure workload identity. - [MCP server](https://cua.ai/docs/reference/cua-cli/mcp-server.md): The cua serve-mcp stdio server, its permission grammar, and the tools each permission exposes. ## Lume ### How-to guides - [Install and run Lume](https://cua.ai/docs/how-to-guides/lume/install-lume.md): Install Lume, Cua's local Apple Silicon VM manager for macOS and Linux guests. - [Create a vanilla Tahoe VM](https://cua.ai/docs/how-to-guides/lume/create-vanilla-vm.md): Create and manage a vanilla macOS Tahoe VM from a local Apple restore image. - [Run Omarchy on Apple Silicon](https://cua.ai/docs/how-to-guides/lume/run-omarchy-arm64.md): Build an experimental ARM64 Omarchy VM from official source with Lume, Arch Linux ARM, and optional Rosetta translation. - [Run OpenClaw in a local macOS VM](https://cua.ai/docs/how-to-guides/lume/run-openclaw.md): Install OpenClaw in an isolated Lume VM and keep its gateway running headlessly. - [Manage local Lume VMs](https://cua.ai/docs/how-to-guides/lume/manage-vms.md): Run, inspect, clone, share files with, and remove local Lume VMs. - [Expand a VM disk](https://cua.ai/docs/how-to-guides/lume/expand-vm-disk.md): Increase a stopped Lume VM's disk and verify the guest capacity. - [Change SIP on a macOS VM](https://cua.ai/docs/how-to-guides/lume/change-sip.md): Disable or enable System Integrity Protection from paired macOS Recovery and verify the result. - [Metal capability unlock](https://cua.ai/docs/how-to-guides/lume/gpu-passthrough.md): Enable a configured, tested Metal capability profile for one workload inside a Lume macOS VM. - [Serve the Lume API](https://cua.ai/docs/how-to-guides/lume/serve-api.md): Run Lume's local HTTP API for tools and scripts. - [Use Lume with MCP](https://cua.ai/docs/how-to-guides/lume/use-mcp-server.md): Connect an MCP client to Lume for local VM management. ### Concepts - [How Lume creates local macOS sandboxes](https://cua.ai/docs/concepts/how-lume-unattended-setup-works.md): How Lume turns Apple restore images into unattended local macOS sandboxes. - [How SIP works in Lume VMs](https://cua.ai/docs/concepts/how-sip-works-in-lume-vms.md): Understand signed LocalPolicy state, paired Recovery, and why Lume verifies SIP after reboot. - [How Lume expands macOS disks](https://cua.ai/docs/concepts/how-lume-expands-macos-disks.md): Understand RecoveryOS relocation, APFS growth, and rollback during a Lume disk resize. ### Reference - [CLI Reference](https://cua.ai/docs/reference/lume/cli-reference.md): Command Line Interface reference for Lume - [API Reference](https://cua.ai/docs/reference/lume/http-api.md): HTTP API reference for Lume server - [MCP tools](https://cua.ai/docs/reference/lume/mcp-tools.md): Tool reference for Lume's stdio MCP server. - [Telemetry and privacy](https://cua.ai/docs/reference/lume/telemetry.md): Exact Lume telemetry controls, schema, and privacy boundaries. - [Limits and VM states](https://cua.ai/docs/reference/lume/limits.md): Host requirements, VM states, storage behavior, and common Lume failures. ## Cua-Bench ### How-to guides - [Run and validate a Cua-Bench task](https://cua.ai/docs/how-to-guides/cua-bench/run-a-task.md): Inspect a task, verify its oracle, and exercise a variant manually. ### Concepts - [What is Cua-Bench?](https://cua.ai/docs/concepts/what-is-cua-bench.md): Understand how Cua-Bench defines, runs, and scores verifiable computer-use tasks. - [The Cua-Bench task lifecycle](https://cua.ai/docs/concepts/cua-bench-task-lifecycle.md): Understand how task variants, setup, solutions, and evaluators form a repeatable experiment. ### Reference - [CLI reference](https://cua.ai/docs/reference/cua-bench/cli-reference.md): Cua-Bench command groups for tasks, datasets, benchmark runs, environments, agents, and traces. - [Task definition reference](https://cua.ai/docs/reference/cua-bench/task-definition.md): Python types, decorators, and lifecycle signatures for Cua-Bench task modules. ## Docs for agents - [Connect your agent to Cua docs](https://cua.ai/docs/how-to-guides/agent-context/connect-your-agent-to-cua-docs.md): Point your coding agent at the hosted Docs & Code MCP Server so it can search Cua's docs and source. - [Docs & Code MCP](https://cua.ai/docs/reference/docs-code-mcp.md): Reference for the hosted Cua Docs & Code MCP Server: endpoint, the four read-only query tools, database tables, and version indexing. ## Browse by type - [Start here](https://cua.ai/docs/tutorials.md): Learn Cua by following end-to-end builds. - [Concepts](https://cua.ai/docs/concepts.md): Understand the design ideas behind Cua and how its main pieces fit together. - [How-to guides](https://cua.ai/docs/how-to-guides.md): Follow recipes for specific Cua goals. - [Reference](https://cua.ai/docs/reference.md): Index of reference material for the Cua CLI, Cua Driver, Lume, the Sandbox SDK, and Cua-Bench. ## More documentation - [Anthropic / Claude](https://cua.ai/docs/use-cua-with/anthropic.md): Use Claude models with Cua Driver through an agent harness. - [Antigravity CLI](https://cua.ai/docs/use-cua-with/antigravity-cli.md): Connect Google's Antigravity CLI to Cua Driver over MCP. - [Claude Code](https://cua.ai/docs/use-cua-with/claude-code.md): Connect Claude Code to Cua Driver over MCP. - [OpenAI Codex](https://cua.ai/docs/use-cua-with/codex.md): Connect Codex to Cua Driver over MCP. - [Google / Gemini](https://cua.ai/docs/use-cua-with/google.md): Use Gemini models with Cua Driver through Antigravity CLI. - [Grok Bot](https://cua.ai/docs/use-cua-with/grok-bot.md): Let Grok Bot operate desktop apps through Cua Driver on your local computer or on its own cloud computer. - [Grok Build](https://cua.ai/docs/use-cua-with/grok-build.md): Connect xAI's Grok Build coding agent to Cua Driver over MCP. - [Hermes](https://cua.ai/docs/use-cua-with/hermes.md): Use Cua Driver through Hermes Agent's built-in Computer Use toolset. - [Kimi](https://cua.ai/docs/use-cua-with/kimi.md): Use Kimi models with Cua Driver through Kimi Code CLI. - [Kimi Code](https://cua.ai/docs/use-cua-with/kimi-code.md): Connect Kimi Code CLI to Cua Driver over MCP. - [llama.cpp and Unsloth](https://cua.ai/docs/use-cua-with/llama-cpp.md): Serve a local GGUF model with llama.cpp and connect it to Cua Driver. - [Meta](https://cua.ai/docs/use-cua-with/meta.md): Run Meta computer-use models locally with Cua Driver. - [MiniMax](https://cua.ai/docs/use-cua-with/minimax.md): Use MiniMax M-series models with Cua Driver through Codex or Claude Code. - [Ollama](https://cua.ai/docs/use-cua-with/ollama.md): Serve a local model with Ollama and connect it to Cua Driver through an agent harness. - [OpenAI](https://cua.ai/docs/use-cua-with/openai.md): Use OpenAI models with Cua Driver through Codex. - [OpenClaw](https://cua.ai/docs/use-cua-with/openclaw.md): Connect OpenClaw to Cua Driver over MCP. - [Qwen](https://cua.ai/docs/use-cua-with/qwen.md): Use Qwen models with Cua Driver through Qwen Code or a local runtime. - [Qwen Code](https://cua.ai/docs/use-cua-with/qwen-code.md): Connect Qwen Code to Cua Driver over MCP. - [T3 Code](https://cua.ai/docs/use-cua-with/t3-code.md): Use Cua Driver from an agent harness managed in T3 Code. - [xAI / Grok](https://cua.ai/docs/use-cua-with/xai.md): Use Grok Build or Grok Bot with Cua Driver.