The presence of GFP at flow sort does not necessarily mean successful editing. The polyclonal population might contain unedited wild-type cells. I tend to use single-cell KO clones, but that carries a risk of clonal heterogeneity, so would need testing multiple clones. Some papers derive single cell clones, confirm the presence of a KO, and then pool 4-5 of these clones to use as a polyclonal population in-vivo.
I use non-edited clones as controls, but you can also make a separate non-targeting sgRNA control.
It is good to check for any GFP expression after the clones are expanded, just to confirm no genomic integration of the plasmid has taken place. I haven’t seen that given there is no selection involved in the process, and the transiently-nucleofected plasmid is lost over time.