Welcome to the 304th issue!

Big news! Meta announces a new testing method — Just-in‑Time Tests (JiTTests).

It's a new way of unit testing at scale thanks to the LLM-powered mutation technique.

The idea is simple — let AI generate lots of unit tests to catch possible failures, and then delete them.

Here's a 17-page research paper describing in detail why and how it works.

Also, Ayyaz Zafar summarises it: Meta's JiTTests: How AI Writes 22,000 Tests, Finds 4x More Bugs, Then Deletes Them All.

Happy (just-in-time) testing! 🙂

Dawid Dylowicz

Want to create a culture of quality in your team? Stuart Day shares how to go beyond mere testing activities and build a strong quality mindset and practice.

Similarly, Joshua Bihun describes what it takes to move towards Quality Assistance: Reframing QA From Gatekeeper to Coach.

Hürkan Tuna raises a valid concern — it's becoming increasingly easier to build software faster, but proper verification and tracking are harder to scale. You can read more about it in the second part and the third part.

And, as Arik Aharoni points out, this creates a need for a new role — Test Critic: Navigating the AI Coverage Tsunami.

Testing isn't just code — it's a mirror of your assumptions. Masoud Bahrami explores how doubt, curiosity, and bias influence every test you write.

On that note, Matthew Sullivan outlines why Software is treacherous and how we can reclaim purpose and influence in testing.

Curious where AI actually supports daily QA work? Todd Conner breaks down how he uses it across planning, scripting and debugging, and where human judgment is still best.

At the same time, Stuart Thomas advises to Forget coverage; focus on risk.

That's a million-dollar question! And for a good reason, as the answer depends on context. Ady Stokes makes a great attempt at providing possible answers from various people and sources.

Similarly, Stuart Day emphasises Why the Language We Use About Quality Matters More Than We Think.

In this series of five articles, Vivek Munde brings up several points on how to correctly identify web UI elements for test automation. You can also read parts two, three, four and five.

If you're looking for a detailed description of how to build an agentic AI solution for auto-generating test automation, Shrey Vats has something for you. Find out more in the second, third and fourth parts.

Additionally, Vera Marczenko explains how to effectively Let AI Write Your Assertions — Then Validate Them With Your Own Patterns.

A bold statement from Pramod Dutta, but there are some arguments supporting it, including how the Screenplay and App Action patterns can replace it.

On that note, I'd also like to highlight A Field Guide to Not Overengineering Test Automation by Irfan Mujagić.

Serhii Smetanskyi lists common pitfalls when trying to follow the test pyramid and explains what to do at each layer to make it work.

Moreover, Okan Yurt explains why Coverage ≠ Quality: Rethinking Unit Tests in the Age of AI.

Testing web app flows with authentication is often challenging. But there's a way to make it much easier, and Sajith Dilshan shows how to achieve that with Playwright.

Furthermore, Faizan Shaikh provides a few tips on proper Playwright Error Handling Explained: Common Errors, Strict Mode, and Best Practices.

Bart Vanherck demonstrates a helpful solution to make Cypress API tests much more concise and easier to write.

Similarly, Mohammad Faisal Khatri wrote a helpful guide on How to test POST API Requests with Playwright TypeScript.

Wondering which browser MCP fits your automation tasks? Andrii Cheparskyi ran a hands-on comparison across Playwright MCP, Chrome DevTools MCP and Agent Browser, measuring how they behave under test.

Raymond Saba and Breno Bittencourt walk through their shift from a legacy setup to a modular Appium 2/3 solution with plugins, parallel threads and shared ownership.

What's more, Anthony Dawson demonstrates how Appium MCP Unlocks New Methodologies in Mobile Automation.

Curious which test tools were the most trending on GitHub in 2025? Each year, there's a slight change, and there should be no surprise at the new contenders given the rise of AI-first tools. By Courtney Zhan.

In another great example, Karthik KK shows how to pair Playwright MCP with LM Studio to set up a local test agent without paying a dollar.

95% pass rate... 😂

Thanks for reading!

If you enjoy this newsletter and find it helpful in becoming a better tester, please consider sharing it with others.

Dawid Dylowicz