If you are worried about off-target mutations, you sequence a cell and see if any showed up compared to the parents' genomes + intended edits, and if they did, you toss that embryo.
So grow the embryo further and sequence more of it, or draw upon the prior information about off-target rates to make the best use of 30% coverage and pick ones with inferred rates which are acceptably low. Or use even more CRISPR to get rid of existing mutations to maintain total mutation load at normal rates... Many options.
(Really, all these comments sound like people trying to find an excuse for why it doesn't work, rather than consider how it could work.)
