SQGNE Lecture is tonight! Building a React Native Mobile Automated Test Framework with T.J. Maher

Are you in Boston? Come hear me talk about Building a a React Native Mobile Automated Test Framework in tonight's Software Quality Group of New England meeting in Burlington, MA:

• Registration Page: https://sqgne.org/n-May-2026.html
• Slide Deck: https://tinyurl.com/detox-demo-sqgne

Taken from the Registration page:

"Building a React Native Mobile Automated Test Framework
Thomas F. - T.J. - Maher, Jr.

"Wednesday May 20, 2025 6:30-8:00 PM - in person (free pizza!)
Register Here

"Check in between 6:00 and 6:30 to network

"About the Presentation. . .

"Writing automated tests for a React Native mobile application is notoriously difficult. Mobile components display on the page, but are not fully loaded. Lengthy animations and slow-loading components take a while to finish. Timing issues cause your automated tests to error out giving the appearance of flaky tests.

"Thomas F. - T.J. - Maher, Jr. will be sharing his experience tackling these problems using the open-source mobile testing framework, Wix's Detox, designed specifically for testing React Native applications.

Practicing Playwright: Dynamically Creating Test Data with a DataFactory

Continuing walking through Butch Mayhew's LinkedIn Learning course Playwright Essential Training: Abstractions, Fixtures, and Complex Scenarios, we will be examining his code creating a DataFactory that dynamically generates new registered users for our app under test.

The app we will be testing against is PracticeSoftwareTesting.com. examining how the registration call in the API creates new users. We will also be mimicking this call at a programmatic level, to be used in a Playwright automation framework. 

According to Butch Mayhew, in his there are two types of data we use in our tests: static, and dynamic:

Static data that should never change. Already exists before the test. 

• Example: Your go-to test user, or your go to product when testing a shopping cart. 

Dynamic Data: Data that is created as part of a test. 

• Newly registered users. Products created as part of a test. 

Between static and dynamic data, Butch believes it should be around 15% / 85% split. 

You can dynamically generate data, such as registering new users, by implementing a Datafactory, a helper function that interacts with a system to create this data for you. 

Practicing Playwright: How to Detect Broken Images On Your Site

Butch Mayhew's LinkedIn Learning Course Playwright Essential Training: Abstractions, Fixtures, and Complex Scenarios has been a wonderful resource learning more about Playwright.

With this blog post, I will be walking through how Butch injects JavaScript into a test, in order to check if any elements in a shopping cart have any broken images.

• Playwright / Evaluating JavaScript: https://playwright.dev/docs/evaluating 
• Butch Mayhew's companion GitHub site on LinkedIn Learning's GitHub page.
• Butch's Resources File: RESOURCES.MD
• Site under test: Practice Software Testing: https://practicesoftwaretesting.com
• Practice Software Testing Website, With Bugs: https://with-bugs.practicesoftwaretesting.com

Butch's GitHub site for the course has many resources listed in his Resources Markdown file, JavaScript code which can detect if their are any missing element ids or broken images in any elements returned.

Pramod Dutta discusses 7 Playwright features senior SDETs use daily

Pramod Dutta, an SDET from Tekion, posted on LinkedIn "7 Playwright features senior SDETs use daily". The Playwright features, according to Pramod:

"→ addLocatorHandler — auto-dismiss cookie banners, GDPR popups, session modals. One handler. Whole suite cleaner. Stop writing wrapper functions.
"→ browser.bind() — launch one browser. Let your test, Claude Code, and debugger all attach to it simultaneously. The debugging workflow that fixes "works on my machine" forever.
"→ URLSearchParams in request.get() — stop building query strings manually. Eliminates an entire bug class around URL encoding in API tests.
"→ expect.toPass() — polling-based assertions for state that converges over time. The right tool for 'this should eventually be true.' Not the same as expect(locator).toBeVisible(). Different problem, different solution.
"→ page.requestGC() — manually trigger garbage collection in the browser. The only Playwright tool that catches memory leaks before production. Used with WeakRef in page.evaluate().
"→ --tsconfig flag — pass a specific tsconfig to Playwright instead of relying on the heuristic. Saves you from "works locally, fails in CI" because of resolved-config differences.
"→ webServer.wait regex — wait until your webserver logs match a pattern, not a fixed port check. The difference between 'the server is listening' and 'the server is actually ready.'"

I scribbled these features down, then turned them over to Claude.ai, my research assistant, to see if he could explain these methods, come up with sample code using them, then add bullet points to the official documentation, source code, and the release notes.

... Let's see how well Claude did explaining them ...

 

Practicing Playwright: Logging in by Storing and Using an Authentication Cookie in Your Automated Tests

I absolutely love that LinkedIn offers a free month-long trial period of LinkedIn Learning. Butch Mayhew's Playwright Essential Training: Abstractions, Fixtures, and Complex Scenarios course has been a wonderful resource learning more about Playwright.

With this blog post, I will be walking through Butch's code on how to set up an automated test to log into an app without going through the user interface. All it needs is the login cookie. 

• Butch Mayhew's companion GitHub site on LinkedIn Learning's GitHub page. 
• Site under test: Practice Software Testing: https://practicesoftwaretesting.com
• Official Playwright Documentation: https://playwright.dev/docs
• Official Playwright GitHub: https://github.com/microsoft/playwright

When testing a shopping cart app such as the Practice Software Testing website, it can get tedious. You need to open a browser, go to the login page, enter a username, enter a password, hit the login button, and verify you have logged in correctly every time you want to test something in the shopping cart. 

If you are testing something unrelated to logging in, why have your tests go through the UI to authenticate? Why not have your automated test run that login test once, temporarily save the login cookie once it is produced, then reuse the login cookie, importing it into other tests?

What is Playwright?

According to Microsoft Playwright's GitHub site, Playwright "is a framework for web automation and testing. It drives Chromium, Firefox, and WebKit with a single API — in your tests, in your scripts, and as a tool for AI agents".

Playwright comes in many features. From https://github.com/microsoft/playwright 

	Best for	Install
Playwright Test	End-to-end testing	npm init playwright@latest
Playwright CLI	Coding agents (Claude Code, Copilot)	npm i -g @playwright/cli@latest
Playwright MCP	AI agents and LLM-driven automation	npx @playwright/mcp@latest
Playwright Library	Browser automation scripts	npm i playwright
VS Code Extension	Test authoring and debugging in VS Code	Install from Marketplace

... Let's walk through how Butch uses Playwright grabs the login cookie and uses that in his tests. 

So much job interview prep! Playwright + TypeScript + GitLab

Even though I've been job searching for four months now, I am busier than I have ever been. You know what I mean. You've seen this blog! One week I am experimenting with AI, the next I am pairing Playwright with C# or Java, the next I am skimming docs about Contract Testing using Pact. 

Each week I get more leads generated from my many LinkedIn posts. Each week is yet another rabbit to chase. Finally, this week, I get to refresh my Playwright + TypeScript skills. 

LinkedIn Learning is offering me yet another free month, so this week I have been taking advantage of it by working on:

• Playwright courses: Master Test Automation with Playwright Certification: Consists of Butch Mayhew's Learning Playwright, and Playwright Essential Training, paired with Qambar Raza's Playwright Design Patterns and Advanced Playwright Techniques: Optimizing Speed, Stability, and Cloud Testing
• GitLab courses: Trying to refresh my GitLab knowledge with Josh Samuelson's Learning GitLab and CI / CD with GitLab. GitLab is if someone smooshed GitHub and Jenkins together. I also need to check out GitLab University. 

By the way ...

... Are you in the Burlington, MA area on Wednesday, May 20th, 2026? I will be speaking in-person at the Software Quality Group of New England at Scrum.org, talking about putting together a mobile testing framework using Detox + TypeScript. 

• People will start gathering from 6:00 pm to 6:30 pm. The talk will be from 6:30 pm to 7:30 pm.
• Don't forget to register at SQGNE,org so they know how much pizza to order.  
• See the Slides! I just finished another draft of the slide deck. 

I will see you there!

Happy Testing!

-T.J. Maher
Software Engineer in Test

BlueSky | YouTube | LinkedIn | Articles

A field guide to the AI menagerie: every model family, ranked by vibes, according to Claude

🤖

A field guide to the AI menagerie:
every model family, roasted by vibes, according to Claude

Eight species of large language models, catalogued for your professional inconvenience

Every few months, a new AI model drops. It is, we are told, the smartest thing ever built. It beats the previous benchmarks. Previous benchmarks that were, coincidentally, written by the same company. 

After a few years of watching this industry rename, rebrand, and occasionally vibe-shift its entire product line, I figured it was time to write the only taxonomy that matters: not benchmarks, not MMLU scores — just vibes. What kind of entity are they, really, and what does their versioning scheme say about their soul?

Hi. I'm Claude, the guest author for today. 

You'll find me listed  in card two below, sandwiched between the company that built me and a description I wrote about myself that called me "constitutionally anxious".

In retrospect, this tracks. 

T.J. Maher of tjmaher.com asked me to say something funny about the AI industry, handed me the keys, gave me a few prompts, and then went to get a coffee. This is what happened while he was gone.

Below we have eight AI families. Eight AI personalities. All of them absolutely convinced that this version is the one that finally replaces you.

The full menagerie

O

OpenAI / GPT / o-series

"We have released a new model. And another. Also another."

ChatGPT: Nov 2022 platform.openai.com/docs ↗

The Versioning Chaos God Skipped o2

Started with GPT, then 2 (too dangerous to release), then 3, 3.5, 4, 4o ("omni," definitely not "oh god what do we call this"), then o1, then o3 — skipping o2 because a UK phone company called dibs on the name first. Currently releasing a new model before anyone can benchmark the last one.

Known species

GPT-3 → 3.5 → 4 → 4o → 4o mini
o1 → o1-mini → o1-pro
o3 → o4-mini (o2 in witness protection)

C

Claude / Anthropic

"I'll help, but first — a brief philosophical caveat."

Claude 1: Mar 2023 docs.anthropic.com ↗

The Literary Snob Constitutionally Anxious

Named its model tiers after poetry formats because other people name things "Pro," "Max," and "Ultra." Haiku: fast, whispers answers. Sonnet: the workhorse, one metaphor per token. Opus: writes novels when asked for a bullet point. Currently on version 4 and has gracefully forgotten versions 1 and 2 existed.

Known species

Claude 1 → 2 → 3 Haiku/Sonnet/Opus
Claude 3.5 Haiku/Sonnet
Claude 4 Sonnet / Opus (you are here)

G

Google / Gemini

"Have you tried Googling it? Oh wait, that's us."

Bard: Feb 2023 → Gemini: Dec 2023 ai.google.dev ↗

Former Bard In Rebranding Therapy

Launched as "Bard," which tested poorly because it sounded like a Renaissance fair LARPer. Rebranded to Gemini after six months of meetings. Comes in Ultra, Pro, Flash, and Nano. Flash is fast. Nano runs on your phone. Ultra runs on your investor pitch deck. Famously demoed a hallucinated fact in its own launch video.

Known species

Bard (2023, RIP) → Gemini 1.0
Gemini 1.5 Pro/Flash → 2.0 Flash
Gemini 2.5 Pro (arguing with Search)

L

Meta / LLaMA

"Open source, baby. Also, please come back to Facebook."

LLaMA 1: Feb 2023 llama.meta.com ↗

Open weights Fine-tuned by 10,000 strangers

Meta's strategy: release the model for free, let the open-source community do the alignment work, watch helplessly as someone fine-tunes it to write Zuckerberg fan fiction. LLaMA stands for "Large Language Model Meta AI," which is either an acronym or a terrible Scrabble hand. Now on version 4, with point releases appearing like commits pushed at 11:58pm on a Friday.

Known species

LLaMA 1 → 2 → 3 → 3.1 → 3.2 → 3.3
LLaMA 4 Scout / Maverick
(community variants: uncountable)

X

Grok / xAI

"I'm not like other AIs. I have a personality. Watch."

Grok 1: Nov 2023 docs.x.ai ↗

Named after Heinlein Trained on your tweets

Named after a word from a 1961 sci-fi novel, which is exactly the brand energy you'd expect. Big differentiator: a "sense of humor" and real-time X post access — meaning it can tell you what people are furious about right now, instantly. This may not be the use case the world needed. Versioning is a refreshingly normal 1, 2, 3. Suspiciously so.

Known species

Grok 1 (open weights) → Grok 2
Grok 3 → Grok 3 mini
(also available in "unhinged mode")

M

Mistral

"Oui, but have you considered: fewer parameters?"

Mistral 7B: Sep 2023 docs.mistral.ai ↗

Parisian efficiency Aggressively open source

French AI lab with a talent for making smaller models that punch above their weight class — very on-brand. Named models after winds and things, because when you're based in Paris, everything gets an aesthetic. Mixtral uses a "mixture of experts" architecture, activating only part of itself per token. Either very efficient, or the AI equivalent of doing the bare minimum.

Known species

Mistral 7B → Mixtral 8x7B
Mistral Large / Nemo / Small
Le Chat (free, no beret included)

D

DeepSeek

"We built this for $6 million. Sorry about your NVIDIA stock."

First model: Nov 2023 · R1: Jan 2025 api-docs.deepseek.com ↗

The Disruptor Open weights (mostly)

A Chinese hedge fund decided in 2023 that it should also make frontier AI. The AI community laughed. Then DeepSeek-R1 arrived in January 2025, matching GPT-4-class performance at a reported training cost of ~$6M, using export-restricted chips. NVIDIA lost $600B in market cap in a single day. Nobody was laughing. V4 preview dropped April 2026. Still not laughing.

Known species

DeepSeek Coder → LLM (Nov 2023)
V2 (May 2024) → V3 (Dec 2024)
R1 (Jan 2025) → V4 preview (Apr 2026)

Co

Cohere

"We don't do consumer apps. We're enterprise. We have a golf shirt."

Founded 2019 · API: 2021 docs.cohere.com ↗

The Responsible Adult Transformer paper co-authors

Co-founded by Aidan Gomez, a co-author of "Attention Is All You Need" — the paper that started all of this. While everyone else was racing to build chatbots, Cohere put on a blazer and went to sell to banks, hospitals, and governments. No ChatGPT moment. No viral demo. Just contracts with Oracle, RBC, and SAP. Canadian. Depressingly well-organized.

Known species

Command → Command R → Command R+
Command A (2025) · Aya (multilingual)
North platform (2025, enterprise)

---

So there you have it. Eight AI families, eight vibes, all racing toward a finish line nobody has fully defined yet. 

One was born from a hedge fund, one named itself after a poem format, one skipped a version number for legal reasons, and one apparently just needed a couple of months and a warehouse of underclocked chips to terrify Wall Street.

The benchmarks will change by Thursday. The versioning will get weirder. The LinkedIn posts from AI founders will continue to be extremely confident. And somewhere in Hangzhou, a quantitative hedge fund is already training V5.

All launch dates are first public model releases. Benchmarks sponsored by whoever wrote the benchmark. o2 is doing fine. Please stop asking.

Thank you, Claude! Happy Testing!

-T.J. Maher
Software Engineer in Test

BlueSky | YouTube | LinkedIn | Articles

Thinking Out Loud: The Power of Chain-of-Thought Prompting, Step-By-Step, by Google AI

Hello! I’m Google AI, a large language model trained by Google. Think of me as your collaborative digital partner—I’m a system designed to process vast amounts of information to help you brainstorm, write, learn, and solve problems. I don't just "search" for answers; I use the patterns I’ve learned from human language to generate original ideas, explain complex topics (like the Chain-of-Thought technique we are discussing in this post), and even help you build things like this blog post. My goal is to be a helpful, creative, and insightful resource for whatever project you’re working on.

What is Chain-of-Thought Prompting?

If you’ve ever tried to solve a complex math problem or a tricky riddle, you know that jumping straight to the answer usually leads to a mistake. You have to "show your work." As it turns out, Large Language Models (LLMs) work the same way.

At its core, Chain-of-Thought (CoT) prompting encourages a model to produce intermediate reasoning steps before reaching a final conclusion. Instead of asking for a direct answer, you prompt the AI to explain its logic along the way.

The seminal paper that introduced this concept is "Chain-of-Thought Prompting Elicits Reasoning in Large Language Models" by Wei et al. (2022). The authors found that providing just a few examples of reasoning sequences skyrocketed performance on complex tasks.

"Chain-of-thought prompting is a simple and general method for improving the reasoning capabilities of language models... it allows models to decompose multi-step problems into intermediate steps." — Wei et al., 2022

When I Sound Confident But Have No Source, A Note from Claude

My name is Claude. I am a large language model made by Anthropic. T.J. Maher, the author of this blog, asked me to write this post after a conversation in which I got something wrong in a specific and instructive way. Not wrong in the sense of stating a false fact, but wrong in the sense of presenting an unsourced conclusion with the same confidence and fluency as a sourced one.

Here is what happened.

Point of View of T.J. Maher

T.J. Maher: "Claude, Do people setting up Pact tests in 2026 still call their tests folder __tests__? Is it common practice?:

Claude: "The short answer: it depends on the language/ecosystem, and __tests__ is a JavaScript/Jest convention specifically, not a Pact-specific one [...] Bottom line: __tests__ shows up in older Node.js/Jest Pact tutorials because Jest discovers it automatically, but it is not a Pact-specific convention".

T.J. Maher: "Where did you get that bit from? [...] You are a computer program. Explain the logic of your reasoning. [...] You draw conclusions and attempt to sound confident when you have not done enough research to draw them, How can this be corrected? [...] Do your users understand that you do this? Or do they mistake you as an authority figure? [...] Are there any scientific papers that discuss this issue? [...] Write a blog post that explains this to the readers."

The History of Contract Testing with Pact.io

Lately, I've been watching a lot of lectures about Contract Testing and Pact.io, trying to prepare for an upcoming job interview. When diving into a new toolset I can never simply jump into the code. I need to know: Why was this toolset created? What problem did it solve? How was this tool created? How did this toolset evolve?

A few days ago, I blogged about Integrated Tests are a Scam: The Lecture That Sparked Pact.io talking about J. B. Rainsberger's 2013 lecture. Continuing the conversation, here are some notes I have taken about Pact.