The JSON file generated by FE does not retain the object type information contained in the originating model, such that a round-trip (model->JSON->model, or JSON->model->JSON) would result in the loss of information.

The type of a JSON object is normally identified by the url or filename.  However in your implementation you have chosen to generate a single JSON file for the entire model, even though a model can be generated from multiple JSON files in RE.  Other similar implementations representing multiple object types include a field to identify the type (e.g. Google Calendar uses a field called "kind").

Can I propose that you include options to :

• Generate entities to separate JSON files in FE
• Specify a field to be used as an object type indicator in both FE and RE processes