Maestro Reviews & Product Details

The simplicity of writing E2E mobile tests in clean YAML files and the new Maestro MCP server are game changers for our workflow. Setting it up took minutes compared to traditional frameworks like Espresso or XCTest. Having built in tolerance for UI flakiness and letting our AI tools interact directly with the running emulator makes writing and maintaining UI tests effortless. Review collected by and hosted on G2.com.

Testing hybrid screens with embedded WebViews or custom canvas elements can still be tricky compared to native UI components. Review collected by and hosted on G2.com.

Maestro MCP lets my AI assistant test the real app instead of just writing code and hoping it works. It can open the app on an emulator or simulator, look at what's on screen, tap through the flows and take screenshots, then tell me what broke. It writes the test flows as it goes, so I don't have to write them by hand.

The flows are plain YAML, so they're easy to read, rerun and tweak. Setup was quick, and it works the same on Android and iOS, which saves a lot of time when you have to check both. When a step fails, I get a screenshot of the failure, so I can see what went wrong without digging through logs. Review collected by and hosted on G2.com.

Timing is the main pain. Flows sometimes fail because a screen hadn't finished loading, and I have to add waits before they pass reliably. Things that get in the way, like dev warning banners or the keyboard on iOS, also needed small extra steps to dismiss. And when a tap doesn't register, it's hard to tell whether the app or the test is at fault, so I end up debugging with the screen hierarchy. Android and iOS don't always behave the same, so a flow that passes on one sometimes needs a tweak for the other. Review collected by and hosted on G2.com.

We wanted a simple way to write end-to-end tests, and Maestro has been that for us. I really like how easy it is to write tests and use Maestro with AI through the Maestro MCP. Maestro was also very easy to set up. Review collected by and hosted on G2.com.

I wish Maestro had a more agentic way of writing tests that could adapt to changes. I’d like to describe what I want a test to do in an abstract way, like checking out or adding something to the cart, without having to tell it exactly how. Review collected by and hosted on G2.com.

I've used Maestro for over four years for UI and end-to-end testing of our iOS apps, and the Maestro MCP server has changed how I work with it. I connect it to Claude Code, so Claude can drive the iOS simulator through Maestro's tools: launch the app, inspect the view hierarchy, tap, type, take screenshots and run flows. I describe a user journey in plain language, and Claude explores the real screen, writes the YAML flow, checks the syntax and runs it until it passes. Covering a new feature with E2E tests now takes minutes instead of hours.

Debugging got much better too: when a flow fails, Claude looks at the actual screen state and proposes the fix right in my codebase, instead of me guessing from logs. And because everything stays plain YAML, the flows Claude writes are easy for anyone on the team to review and maintain.

The fundamentals are still excellent: built-in waiting keeps tests far less flaky than other tools I've tried, setup is just installing the CLI, and the same flows run in CI and on Maestro Cloud to catch regressions before every release. Review collected by and hosted on G2.com.

My main frustration is debugging failed flows in CI. Locally, the MCP server and Claude Code make it easy to look at the real screen and find the cause, but in CI that context is gone: error messages can be generic, and I sometimes have to rerun the flow locally, add screenshots or dig through logs to tell whether it was a selector issue, a timing problem or a real bug in the app. Richer failure reports, such as the view hierarchy and a screenshot at the moment of failure plus more descriptive errors, would save a lot of time and would also give AI agents much better input to fix tests automatically.

Some advanced use cases are also hard to discover and often require digging through examples or GitHub issues. The learning curve for more complex flows (JavaScript, data-driven tests or custom logic) is a bit steep, and the MCP tooling is still young, so documentation and examples of AI-assisted workflows are limited. Better docs for edge cases and for the MCP server would make Maestro easier to adopt across the whole team.

That said, these are minor issues compared to how much time Maestro saves us overall. Review collected by and hosted on G2.com.

I like Maestro’s control and customizability. It offers per-flow control and a real nice DX compared to tools like Appium. I can modify it out of the box and plug in some of our own tools, like rewriting tests in memory based on screen dimensions to change click coordinates. Since Maestro plugs into adb, I get a robust log trace. It helped us fix UI problems before we moved to prod. Setup had very little to no friction, and the documentation was pretty good. Review collected by and hosted on G2.com.

My main issue is that Maestro still can’t fully “see” an element as-is. I’ve had inconveniences with buttons and dynamic text elements. When dates and names change, we have to rewrite tests based on the day and data, which gets extremely tiring at scale, especially across 100 devices. Maestro doesn’t have a way to treat it as something generic like “Text here.” Review collected by and hosted on G2.com.

I find Maestro easy to use, with a low learning curve—anybody can use it and read the code, including non-technical colleagues. It helps me automate tests faster and increase our test automation coverage. I can use it for Android and iOS on any device, including extended edge-case tests for permissions, locations, and audio controls. The MCP helps me create targeted end-to-end tests, debug and fix test automation issues, and onboard people to our multi-team automation framework. Setup is easy on both Mac and Windows, and installing Maestro Studio was easy too. Review collected by and hosted on G2.com.

I haven’t had any Maestro-specific issues. Sometimes the app I’m testing has elements without IDs, so I have to rely on coordinates. When I run the same test on another device, the coordinates can change because the screen size or aspect ratio is different, which can lead to false negatives or broken tests. Review collected by and hosted on G2.com.

I like how easy Maestro is to use and how straightforward it is to implement and maintain tests. I also appreciate the continuous addition of new features, which helps improve the quality and capabilities of our test suite. The cloud test runs have also been stable and reliable, making it a great tool for our E2E automation. Review collected by and hosted on G2.com.

On Android, test execution can be a bit slow, which impacts overall feedback time. The cloud-generated reports(videos) could also be clearer, making it easier to identify failures and investigate issues quickly. Review collected by and hosted on G2.com.

I like how fast Maestro is compared with other mobile Android testing solutions like Appium. I could easily connect Claude Code to Maestro through its MCP support, so Claude could run mobile tests automatically. That helped me run test cases quickly and do in-depth testing. I find the YAML test scripts easier to understand, and Maestro Studio lets me write test scripts manually by clicking the mobile UI rendered in the desktop app. Setup was easy and straightforward. Review collected by and hosted on G2.com.

I use pytest for web testing, but Maestro doesn’t allow similar tools for writing full test suites and workflows. It’s not that bad of an issue, but it would be nice to have. Review collected by and hosted on G2.com.

What I like most is its simplicity. Maestro Studio really helps me write test cases quickly and efficiently, and I can easily replay the scenario afterward. There are also many pre-built methods I can use to control apps, which makes the whole process smoother. Review collected by and hosted on G2.com.

What I don’t like is that the tests run too slowly. Interaction with the UI is generally efficient, but it can still feel very slow in practice. For example, when I enter text in a field, it has to remove the old text character by character before typing the new one, which makes the whole process drag. Review collected by and hosted on G2.com.

I really love how simple it is to write tests in YAML, and the MCP is fantastic for speeding up development. Review collected by and hosted on G2.com.

It can be slow at times, but I’m not sure that’s Maestro’s fault. That said, it does seem to be improving. At first I had issues with filling in forms and with scrolling, but it’s been getting better. Review collected by and hosted on G2.com.