I've often wondered how CRISPR is actually useful, because at its core it's just a nuclease. It allows to you precisely cut DNA where you want to cut it. But for that to be useful you still have to put the DNA back together afterwards. I found this article useful for understanding the different ways that happens in practice:
https://www.addgene.org/CRISPR/guide/
Turns out you get some small mutations in the most common mechanism (non-homologous end joining or NHEJ). This could be a problem for use in therapies, but definitely not a huge issue for most research-based applications where you can control which mutants you allow to proceed to the next experiment.