AI Testing Tools, WireMock GUI, and a Free Event Near You

Are you still writing Playwright tests by hand, while AI can now generate them straight from user intent? Is your AI testing adoption stalling because you've skipped the most critical step most teams get wrong? And could a free in person testing event be coming to a city near you this week? Find out on this episode of the Test Guild News show for the week of February 22nd. So grab your favorite cup of coffee, tea and let's do this. Let's kick things off with some big news from the Test Guild itself, because if you're anywhere near Miami, you're going to want to hear this. Test Guild has taken its community off screen and into the real world. Following a successful debut in Chicago last September, Test Guild IRL is now expanding to cities across the country. Our first stop on the 2026 tour is Miami. On Thursday, February 26, from 4 to 8pm, Test Guild is hosting a free AI and software testing event at FIU's College of Engineering and Computing. Seats are capped at 100 and the speaker lineup includes AI and autonomous testing pioneer Tyreek King. I'll be speaking, as well as testing industry veteran Jim Trendoo, and we'll also be joined by Madassa as well. Registration is now open using that link down below. Register now. Hope to see you there. Big special thanks to Test Mu AI for sponsoring this tour and Jim from Meridius for making this all happen. Speaking of test skill, this brings up our Webinar of the week. So in this week's webinar, we'll be tackling a problem familiar to most automation teams, UI test frameworks that still require manual work at the selector level, fragile weights and consistent scaffolding just to get tests off the ground. If that's you, you want to join us tomorrow at Blinkio, who's presenting a session on AI Powered Vibe Testing, a workflow designed to capture real user intent and translate it directly into Playwright native test code that this session will cover how actions, assertions and checks are inferred automatically during execution, how to generate a test plug into an existing project structure, and how the approach aims to reduce maintenance overhead without producing opaque black box code. And as always, the agenda includes a live demo walkthrough of the full flow from a tent automated scenario to execution, plus a Q and a segment. Registration for this is also now open. I highly recommend you register down below even if you can't make it, or if you see this after the event so you can get an instant replay. Hope to see you there. Do you use Wiremark? If so, you want to hear this next announcement I found on LinkedIn. Once again, it's from Roy who just released Wiremark GUI which is an open source web based interface for managing Wiremark directly from its admin API. This was just released on GitHub and it eliminates the need to hand edit JSON files by providing a visual layer on top of Wire Mock's existing admin endpoints, and the GUI covers the full range of wiremox core features. Testers can create, edit, import, export, search and persist stub mapping to both a form editor and a raw JSON editor. The request journal lets you browse all logged requests, filter for unmatched ones and clear the logs. Scenario states can also be viewed and reset, and the tool supports starting and stopping recordings against a target URL. One practical standout is a multi server support. The GUI can connect to any Wire Mock instance, whether it's running locally, on a remote host or inside a container by entering a configurable URL in the connection bar. Another cool solution from Roy that you definitely should check out as well. All right, another article that caught my attention on LinkedIn was from Jay, who created a blog post that walks through a process of building a playwright automation framework for Sala X. I think that's how you pronounce it, which is an enterprise secure messaging web client developed by SSH Communication Security, which is a Finland based cybersecurity company. I didn't know much about it, but it turns out the application is built on the Matrix Open standard in the Element Communications platform and is designed for organizations in government, defense and regulated industries that require end to end encrypted communications and this post focuses on specific testing challenges that come up with automating a secure messaging application. Like Multi step authentication flows post login security prompts, real time messaging, synchronization between simultaneous users, and complete session isolation. A key test scenario in the framework validates bi directional messaging between two users logged in simultaneously, verifying that messages are delivered and are visible in real time. The post also documents an important behavioral edge case when one user is offline during a message exchange, the receiver user sees those messages marked as unencrypted when they come back online, a real world encryption behavior that has direct implications for test assertions. She also clearly talks about where automation stops. Video and audio calls, screen sharing and full encryption verification are flagged as areas that still require minimal testing as they involve hardware access, OS level controls and security flows outside the scope of browser based automation. And this full framework is available on GitHub as well, you definitely check out down below and if you attended Automation Guild a few weeks ago, you know about this new company called Thunders. Well, they just published a guide introducing Dunders mcp, which is a model context protocol integration that connects AI assistants directly to a testing workspace, allowing them to author, execute and analyze tests, plain English commands with no code required. And the Thunders MCP is compatible with Claude Desktop, ChatGPT, Desktop, Cursa, Devin and Windsurf. Setup takes three steps. Add the server URL, authenticate via OAuth and verify the connection. No SDKs are involved. Once connected, testers can use natural language to accomplish tasks that would otherwise require manual tool switching, and this article outlines five specific use cases like converting existing Playwrights, Selenium and Cypress task suites into Thunders test cases without manually rewriting authoring new test cases from Flows requirements FIGMA designs running tests under defined Personas such as QA Engineer or Accessibility Tester, triggering test runs against specific environments and reviewing failures, coverage gaps and trends across recent runs, all explained back in natural language rather than raw logs. And this article frames this as removing the context switching that typically moves testers between project management tools, a QA tool, a CI tool and an issue tracker. So if you miss Automation Guild, you want to learn more about Thunders and the MCP server? Definitely. Check it out down below. Let me know your thoughts. Speaking of Claude code, I just came across this official announcement. Anthropic announced on February 20 that its Claude Code Desktop tool now supports three capabilities that they announced on their official Claude account on X. First, they have a Live app Preview feature which lets developers inspect UI state of running applications while they're actually writing code without switching to a separate browser or emulator. Second, which I think you'll be more interested in, it is automated code reviews surfaces diffs alongside suggested fix directly in the tool and third, Cloud Code can now handle CI failures and pull requests in the background, reading logs, processing patches or opening or updating periods without requiring the developer to manually triage each failure. And this announcement specifically highlights use cases for large repositories and teams looking for standardized review processes to reduce review backlog and handle continuous integration pipelines without adding headcount. Also on LinkedIn, I just noticed one of our Automation Guild speakers, Elias, just released a new blog post and the core problem Alias sets out to solve is one many testers are running into traditional pass fail. Automation doesn't work for non deterministic AI systems, so to answer questions like did this model change break anything? Or is my new chunking strategy actually better. He built a framework using deep evalve for running evaluations and ML Flow for tracking results and enabling side by side comparisons of runs across different configurations. Three failure modes emerge from the work that every tester should know. First one is app failures, the expected kind, including hallucinations and ungrounded responses where the app fails answers correctly but for the wrong reasons. Second is eval failures, especially wrong assertions in your test setup caused by ambiguous metric definitions. And third is a combination failures where both the app and the eval fail simultaneously, making the issue nearly invisible. He also recommends meta test test that tests the tests to catch eval level problems before they mask real application bugs. And as practical advice, start with a small number of v vals, define your metrics carefully before writing any code, and gather your framework requirements early. The side by side comparison view he built late in the process turned out to be the one of the most valuable parts of the system and he wish he built it sooner. And the full project is now available on GitHub as well. So check out the article down below and try it out for yourself. Last up is new research from leapwork. In this article they put some hard numbers on a tension most testing teams are already feeling. Enthusiasm for AI and testing is high, but trust has still earned one reliable test at a time. 88% said AI is now a priority for the testing strategy, and 80% expect a positive impact over the next two years. But actual deep adoption tells a different story. Only 12.6% currently apply AI across key test workflows, while 65% are still in the early exploration mode. What's holding teams back? According to the study, 54% point to concerns about accuracy and test stability, and they also have a quote from leapworks CEO Kenneth, who puts it plainly in the report. The question is no longer whether teams will use Agenta capabilities, but how confidently and predictably can they rely on them. All right, for links of everything of value we covered in this news episode, head over to the links in the comment down below. So that's it for this episode of the Test Skill News Show. I'm Joe. My mission is to help you succeed in creating end to end full stack pipeline automation awesomeness. Now, as always, test everything and keep the good Cheers.