The test runner uses Babashka, since each solution is run as its own script it made sense to trade feature-completeness for responsivity, although we're open to running things differently if folks would prefer. I've personally been brainstorming adding a SCI interpreter to the mix to offer some kind of inline evaluation experience, and give the option to execute code with regular Clojure. In any case we probably ought to at least present this info somewhere to the user, since it trips people up occasionally.

You can add Clojure spec to the mix at least for babashka but use the babashka fork. About let*: just file an issue in the SCI repo and I'll add it.