Availability

What's real today, what isn't yet.

“Available” should mean something specific: installed, working, and ready for your test suite. Web automation runs today against real Chromium and Firefox browsers, Android automation runs on real devices and emulators, and iOS automation runs natively on Simulators and registered devices. Native macOS application automation runs through XCUITest, and native Windows application automation runs through UI Automation — each with the actions, locators, and retrying assertions everyday tests rely on. This page is the detailed, continuously updated picture behind the status pills you see elsewhere on the site — surface by surface, capability by capability, and language by language, including what's still missing.

Surfaces

Web, Mobile Android, Mobile iOS, Desktop macOS, and Desktop Windows have real, installable plugins today. The rest have a reserved place in the plugin catalog but no runtime build yet — installing them isn't possible until that changes.

Web

Available now

Chromium and Firefox, with the actions, locators, state reads, and retrying assertions everyday web tests need.

Mobile

Available now

Android plus iOS Simulators and registered physical devices are available with automatic runtime and app setup, using the same Playwright-style client shape.

Desktop

macOS + Windows available

Native macOS applications through XCUITest and native Windows x64 applications through UI Automation are available. Linux remains planned.

API

Not yet available

Backend checks in the same test run.

Cross-surface feature parity

The core action and locator model is shared everywhere. Features tied to browsers, mobile operating systems, or desktop windowing still differ; this matrix makes those boundaries explicit.

Capability	Web	Android	iOS	macOS	Windows
Click, focus, fill, key input, text, count, and waits	Available	Available	Available	Available	Available
Semantic locators	Available	Available	Available	Available	Available
Accessibility snapshots and AI references	Available	Available	Available	Available	Available
Viewport screenshots	Available	Available	Available	Available	Available
Full-page / scroll-and-stitch screenshots	Available	Available	Available	Available	Available
URL navigation and deep links	Available	Available	Available	Available	Available
File chooser and download hooks	Available	Available	Available	Not yet	Not yet
Dialogs and newly opened contexts	Available	Not yet	Not yet	Not yet	Not yet
Input, checked, attribute, and box reads	Available	Available	Available	Available	Available
Selected-option and selected-text reads	Available	Not yet	Not yet	Not yet	Not yet
Retrying text, count, attached/hidden, enabled/disabled, focused, and editable assertions	Available	Available	Available	Available	Available
Retrying value, checked-state, attribute, and box assertions	Available	Available	Available	Available	Available
Hover, highlight, double-click, and right-click	Available	Not yet	Not yet	Not yet	Not yet
DOM, frames, and in-page JavaScript	Available	Accessibility tree only	Accessibility tree only	Native tree only	Native tree only
Install the application under test	Not applicable	Available	Available	Not yet	Not yet

Allwright Inspector

allwright inspector opens a local UI that shows a live app, its accessibility tree, and locator code in TypeScript, Python, Java, Go, and Rust. It is available for Android, iOS, macOS, and Windows apps; web pages are not covered yet. Inspect mode selects elements without activating them, and Interact mode forwards clicks, swipes, scrolling, typing, and paste to the live app. Read the walkthrough →

Web, capability by capability

“Available now” means the web plugin is installable and everything below works against Chromium and Firefox today — enough to cover the core of a real web test suite. The list on the right is what hasn't landed yet, so you can check whether your suite depends on any of it.

Available now

Launch a real Chromium or Firefox browser — no separate driver to install or version-match
Open and close tabs within a browser session
Work with nested and cross-origin iframes as pages, with automatic load waits and configurable timeouts
Register a typed hook before an action and wait for the new tab that action opens
Register a typed file-chooser hook and upload one or multiple local files
Register a typed download hook and save the completed download to a local path
Handle JavaScript alerts, confirms, and prompts with typed hooks — accept, dismiss, or enter prompt text
Navigate to a URL
Click, double-click, middle-click, or right-click an element with configurable click counts
Type into a field
Hover over an element
Press a key on an element
Focus an element
Highlight matching elements for debugging
Find elements by role, text, label, placeholder, alt text, title, or test ID, with filters, exclusions, and chained locators
Count matching elements
Read native accessibility snapshots as JSON or YAML and act through cached AI element references
Read visible or raw text from an element
Read input values, checked state, native accessibility attributes, and on-screen bounding boxes
Capture the current page URL, live input values, selected options and text, checkbox/radio state, attributes, and element bounding boxes
Wait for an element to appear or become visible
Capture viewport or full-page screenshots
Read page accessibility snapshots as JSON or standard YAML, with queryable element references in AI mode
Retrying URL, value, selection, checked state, text, attribute, bounding-box, count, attached/hidden, enabled/disabled, focused, and editable assertions, including negation (via @allwright.dev/vitest)

Not yet available

Network mocking or request interception
Cookies and saved session state
Geolocation and other device permissions
Mobile viewport and device emulation
Drag and drop
Multiple isolated browser profiles per session
Video recording, tracing, coverage, or PDF export
Safari / WebKit (Chromium and Firefox only today)

Every client language exposes this same capability set — there is no language-exclusive functionality on the web surface today.

Mobile — Android, capability by capability

“Available now” means the mobile-android plugin is real, installable, and drives a genuine app over adb — no Appium, no separate driver server. It covers the core actions, reads, hooks, and the shared retrying assertion set. URL and selection matchers remain web-only.

Available now

Connect to a running emulator or a real device over adb — no separate driver server to run
Install and launch a real app from a local APK or a URL
Open universal links and custom URL schemes with app.goto / app.navigate
Click an element
Type into a field
Focus an element
Press a key
Read visible or raw text from an element
Wait for an element to appear
Count matching elements
Read native accessibility snapshots as JSON or YAML and act through cached AI element references
Capture screenshots, including a full-page scroll-and-stitch capture
Register typed file-chooser and public-download hooks with files streamed to and from the test machine
Text, partial-text, resource id, class name, XPath, and state-based (e.g. clickable) selectors
Playwright-style getByRole, getByText, getByLabel, and getByTestId native accessibility locators
Retrying text, value, checked-state, attribute, bounding-box, count, attached/hidden, enabled/disabled, focused, and editable assertions, including negation (via @allwright.dev/vitest)

Not yet available

Hover and highlight (web-only for now)
Explicit swipe, scroll, drag-and-drop, pinch, and multi-touch gesture APIs
Selected-option and selected-text reads
Typed dialog and newly opened app-context hooks (file chooser and download hooks are available)
Multiple isolated app sessions or switching among several foreground app contexts
General CSS/DOM, iframe, or arbitrary in-app JavaScript sessions for WebViews
Network mocking, request interception, cookies, permissions, tracing, and video recording

Every client language exposes this same Android capability set — same as web, nothing here is language-exclusive.

Mobile — iOS, capability by capability

The mobile-ios plugin is installable on macOS and provisions the runtime and app under test automatically, so users do not install a separate driver or sample IPA manually.

Available now

Connect to an available iOS Simulator with no separate driver setup
Connect to a registered physical device; the prebuilt agent is discovered, re-signed, installed, and forwarded automatically
Download a simulator .ipa or ZIP, or use a local .app bundle, then install and launch it automatically
Install and launch a signed physical-device .ipa, ZIP, or .app bundle without a manual install step
Click, focus, fill, press keys, read text, wait for selectors, count elements, and capture screenshots
Read input values, checked state, native accessibility attributes, and on-screen bounding boxes
Open universal links and custom URL schemes with app.goto / app.navigate
Capture full-page scroll-and-stitch screenshots
Read native accessibility snapshots as JSON or YAML and act through cached AI element references
Register typed file-chooser and download hooks with files streamed to and from the test machine
Automate WebView content exposed through the native XCTest accessibility tree
Accessibility-id, text, XCTest element type, and basic XPath selectors
Playwright-style getByRole, getByText, getByLabel, and getByTestId native accessibility locators
Playwright-style action auto-waiting and retrying text, value, checked-state, attribute, bounding-box, count, attached/hidden, enabled/disabled, focused, and editable assertions
The same Rust, Go, Java, Python, and TypeScript client shape used by Android

Not yet available

Hover and highlight (web-only for now)
Explicit swipe, scroll, drag-and-drop, pinch, and multi-touch gesture APIs
Selected-option and selected-text reads
Typed dialog and newly opened app-context hooks (file chooser and download hooks are available)
Multiple isolated app sessions or switching among several foreground app contexts
Arbitrary in-page JavaScript and general CSS/DOM sessions inside WebViews
Network mocking, request interception, cookies, permissions, tracing, and video recording

Desktop — macOS, capability by capability

The desktop-mac plugin is installable on macOS 14 or newer and starts its bundled XCUITest runner automatically. Xcode must be installed, UI automation permission must be enabled for the process running Xcode, and the app under test must already be installed.

Available now

Connect to the local macOS desktop with an automatically started XCUITest runner
Launch or terminate an application by bundle identifier
Open universal links and custom URL schemes with app.goto / app.navigate
Click, focus, fill, press keys, read text, wait for selectors, and count elements
Read input values, checked state, native accessibility attributes, and on-screen bounding boxes
Capture the application window or a full scroll-and-stitch screenshot of its primary scrollable region
Read native accessibility snapshots as JSON or YAML and act through cached AI element references
Use accessibility-id, text, XCTest element type, basic XPath, and Playwright-style semantic locators
Use retrying text, value, checked-state, attribute, bounding-box, count, attached/hidden, enabled/disabled, focused, and editable assertions through @allwright.dev/vitest
Use the same Rust, Go, Java, Python, and TypeScript server-only client shape as the other surfaces

Not yet available

Application installation or distribution — launch targets must already be installed
File-chooser, download, dialog, and newly opened window hooks
Selected-option and selected-text reads
Hover, highlight, drag-and-drop, multi-click, and other pointer gesture APIs
Multiple isolated desktop sessions or switching among several application windows
Web-style DOM, iframe, arbitrary JavaScript, network, cookie, permission, tracing, and video APIs

Desktop — Windows, capability by capability

The desktop-windows plugin is installable on Windows x64 and starts its bundled, self-contained UI Automation agent automatically — no Appium, WinAppDriver, Developer Mode, or separate .NET install. The app under test must already be installed.

Available now

Connect to the local Windows desktop with an automatically started UI Automation agent
Launch or terminate an application by executable path, a command Windows can resolve, or a packaged-app ID
Open universal links and custom URL schemes with app.goto / app.navigate
Automate Win32, WinForms, WPF, UWP, and WinUI applications
Click, focus, fill, press keys, read text, wait for selectors, and count elements
Read input values, checked state, native UI Automation attributes, and on-screen bounding boxes
Capture the application window or a full scroll-and-stitch screenshot of its primary scrollable region
Read native accessibility snapshots as JSON or YAML and act through cached AI element references
Use native UI Automation selectors and Playwright-style semantic locators
Use retrying text, value, checked-state, attribute, bounding-box, count, attached/hidden, enabled/disabled, focused, and editable assertions through @allwright.dev/vitest
Use the same Rust, Go, Java, Python, and TypeScript server-only client shape as the other surfaces

Not yet available

Windows on ARM and 32-bit Windows
Application installation or distribution — launch targets must already be installed
File-chooser, download, dialog, and newly opened window hooks
Selected-option and selected-text reads
Hover, highlight, drag-and-drop, multi-click, and other pointer gesture APIs
Multiple isolated desktop sessions or switching among several application windows
Web-style DOM, iframe, arbitrary JavaScript, network, cookie, permission, tracing, and video APIs

Still to come

The Linux desktop plugin plus API testing remain part of the direction, but there is no installable runtime for them yet.

Desktop — Linux

Not yet available

The shared desktop contract exists, but there is no installable Linux automation runtime yet.

API

Not yet available

Backend checks in the same test run.

Client languages

“Published” means a normal package-manager install. “From source” means the client is complete and working, but you build it from the repository instead of pulling it from a package registry.