From QAright
How this product answers the problems every browser suite eventually runs into. The reasoning holds whatever you build with.
These are ours, and they are on a marketing site, so read them with the appropriate suspicion. They are written to explain a mechanism rather than to sell one, and each names what the approach will not do, which is usually the more useful half.
The reason they are worth reading even if you never open the product is that the problems are not ours. Every browser suite eventually has to decide what to do when a locator stops matching, how much extra time a slow page has earned, and whether an automatic fix should ever be applied without a person seeing it. We have had to answer those in code, which forces a specificity that an opinion piece does not.
They are not product documentation and they are not a tutorial. Each one explains a mechanism and the rules it obeys, in the order those rules matter. Start with How it works if the others read as though they begin in the middle, because they do.
The guides
7 entries-
How it works
OursRecording, playback, healing and reporting, end to end, in the order they happen. The overview to read first if any of the others look like they start in the middle.
qaright.com -
Why the approach earned a bad name in the first place — single brittle locators, no waits, no structure — and what has to be true of a recorder before it deserves a second look.
qaright.com -
The locator cascade, the things it deliberately refuses to do, and the rule that a healed locator only becomes permanent after proving itself over several runs. The refusals are the interesting part: a heal that silently rewrites a test is worse than a failure.
qaright.com -
Finding the change that will break a test before the test runs into it, by comparing the live application against what was recorded.
qaright.com -
Separating “not there yet” from “gone” from “nothing happened”, and why retry-until-green destroys the one piece of information a failure carried.
qaright.com -
What recording on top of Playwright adds — ranked locators, repair at run time, measured waits — and the cases where writing Playwright yourself is still the better choice.
qaright.com -
What an assistant such as Claude Code or Cursor can do with a test suite it is connected to, and the limits the API holds it to whatever the model decides.
qaright.com
Open source
2 entriesNeither of these needs an account, or us.
-
SpecReport
OursHTML reporting for Reqnroll suites: living documentation generated from an actual run, so what it claims the software does is what the software did. Open source, no account, nothing to do with the rest of this site.
github.com -
SpecHygiene
OursStatic hygiene checks for Reqnroll feature files, distributed on NuGet: unused step definitions, duplicated scenarios, and the other quiet rot that sets in once a suite outgrows what one person can hold in their head.
github.com
Reading is the slow way to learn how a suite behaves. Record one against your own app, free, and watch what the report says.