Statistics question: regression diagnostics necessary for SEM?


I'm doing a mediation analysis with the "sem"-function in lavaan. Since the process is regression-based i thought that regression diagnostics (e.g. test of normal distribution of residuals, outliers etc.) would be necessary. But functions that are normally used for this kind of diagnostics from the "car"-package don't work in Lavaan. I've looked into this, but found no information relating to regression diagnostics in SEM - which led me to believe that it might not be necessary. Does anybody know for sure?

Any help would be much appreciated.