I noticed there don't seem to be any tests currently in this repo, so if that's the preference, I can rebase out that addition. On the other hand, I would happily add a test step as a github action, and even add a few more tests to this repo if that would be helpful!

Awesome stuff! Thanks for your contribution! Tests are always welcome, but it's not like having a test would prevent this particular bug. If I forgot to update the listDir path when adding the .git ignore logic, I'd also forget to write a test for it. Having more of an e2e test using something like bats might be more helpful - then the test wouldn't make any assumptions about the actual implementation of the push command (e.g., separation on listFiles and listDirs) and only validate that no there is no .git/** entries in the list of watched/pushed files. But such e2e tests are pretty expensive. So, it's always a trade-off.

Yeah, fair point. I mostly added the test to prove that it was failing without the code change and then just left it in, but you're right about e2e testing.

At any rate, my small 14 inch laptop monitor is very happy about this change, since I use the watch mode very frequently!