Before generating code from an abstract model of a data store, the model is first analyzed to detect normalization, rationalization, naming conventions, structure conventions, and other anomalies. The analysis is scored, and the score may be weighted. The analysis also suggests scripted solutions for resolving the discovered anomalies. A developer may then choose to implement one or more of the suggested solutions prior to code generation. The score may be compared to a threshold and result used to gate subsequent actions. For example, generation of code from the abstract model may be prevented until such time as anomalies are sufficiently addressed so that the score meets a minimum threshold score.