This is just an FYI, with no expectation of any action (to be taken)!
Understood that ERA5 switches from one dataset (HadISST2) to OSTIA SST on Sep 1, 2007. This switch in the lower boundary condition seems to cause a jump in both SST (variable name: SSTK) and skin temperature SKT. Attached is a time-series through Sep 1, 2007 at [lat, lon]= [-1.75, 156.0].
For those who are interested in climate diagnostics (mon, ann, running mean), this is most likely immaterial! But for those interesting in process studies, you may want to treat this artificial jump cause by switching datasets carefully!