Since this came up before I put the code up for review: I deliberately want to keep some input sanitization here. The format I've picked is in line with the general format we use for class and id attributes, so it feels appropriate to me.

I'd still prefer actual JS tests instead of or in addition to these Capybara tests. That said, I agree with Jack Cody that this was likely the easiest testing to set up.

I'm not sure actual JS tests are beneficial when the only output is DOM manipulation. If you had a function like getSanitizedHash, you could test that, but as-is this tests everything you can.

I do agree that the coverage here is good.

Compared to some JS testing I've done in the past, doing it this way does feels to me like wearing thick, knitted gloves on my fingers when I'm not used to wearing them.

My main concern is that it feels too easy to accidentally break something here in a way that would generate a JS error that this test won't notice. Detecting a JS error like that in a headless browser with Capybara is tricky and fragile. I'm particularly concerned with breaking unrelated behavior on a page where this JS is included but the page doesn't have an accordion (or doesn't have one anymore).