Checkouts can fail, so let's have the ability to retry them. Yes, they are typically at the start of a job and you can just re run it, but that's a pain. It's even more of a pain when you are running a job split across 25 runners and 1/25 fails to checkout, it's a big waste of credits and money. I'd have thought this would be an easy one to add :-) It's only when I read the small print (https://circleci.com/docs/guides/orchestrate/automatic-reruns/) I realised this wasn't supported - we've actually added this to our config.yml and the circleci config validate script didn't say a word about it! It was only after it failing to retry we discovered this.