Blog
AI end-to-end testing, measured. Every number comes from a real run, and every post says what it doesn’t prove. Follow it with the RSS feed.
Playwright anti-patterns: 5 flaky code shapes we reject
Most flaky Playwright tests wait on time instead of on the app. These are the code shapes TestWatch rejects in every generated test, with the exact message the check returns and the rewrite it accepts.
6-minute read
Why Playwright tests pass locally but fail in CI
The test is the same on both machines; the machine isn’t. Each difference between your laptop and a CI runner, how it breaks a test, and the playwright.config.ts line that pins it.
6-minute read
How to evaluate AI testing tools: count the false greens
A green check from the tool that wrote the test is the tool grading itself. Here is how to score a trial so a test that passes on a broken app gets counted, with our own eval as the worked example.
7-minute read