Testing the Click, Not the Render - AEM automation testing with Playwright

I’ve been experimenting with Playwright for testing the AEM authoring experience — not just the rendered pages.

A few interesting issues surfaced that component/rendering tests simply wouldn’t catch:

• Dialog validation that didn’t match the field description
• A field that silently accepted an invalid configuration
• Rich-text input losing characters
• Test cleanup failing because of browser/API session differences
• Shared/included dialog fields that were easy to miss in component-based test coverage

These bugs lived in the authoring interaction — the clicks, typing, dialog state and sequence of actions an author actually performs.

I wrote up the lessons learned, including some of the Playwright patterns that worked (and the ones that didn't 😄):

👉 Testing the Click, Not the Render

Curious how others are approaching automated testing of the AEM authoring experience. Are you testing the editor UI today, or mainly the rendered output?