Hey,

I’m currently building a greenfield app with Elixir. I’ve really enjoyed using GitHub - elixir-wallaby/wallaby: Concurrent browser tests for your Elixir web apps. · GitHub for tests so far.

When using claude code I’ve noticed that sometimes it makes a mistake which causes stuff like:

• The page to overflow from the viewport
• Certain image is broken
• Javascript errors
• CSS not loading
• Syntax errors in the html (which might cause different browsers to render it bit differently than chrome)

These issues might not be so bad that the actual integration tests would fail but usually they are a signal that the user experience in the page is degrading.

I’m wondering if I could collect that kind of data while wallaby is using chromedriver and visiting the pages in the tests?

This way I would not need to run duplicate browser tests so it would be faster and I would still be able to learn about potential issues (and failing the test runs and asking claude code to fix them).

Has anyone done something similiar? What are the current best practices here?