I’m adding this note here just in case it helps.

There is UaServer.getAddressSpace().getInstanceDeclarationHierarchyCache() (after you have loaded the type models). You can get the InstanceDeclarationHierarchy of the UaType. This is what is used internally when (NodeManagerUaNode) createInstance/NodeBuilder is used to make instances of the types. It uses Codegen classes when available, but technically that is separate from the instantiation algorithm (i.e. it will basically just use more specialized UaNode implementations if generated). However I must note that the API what the InstanceDeclarationHierarchy has is not that suitable for application-level uses (for example things can be overridden in subtypes etc. the full ruleset is quite complex).

IF you use createInstance/NodeBuilder to make your instances (i.e. the instances itself are not part of any NodeSets), you can make the TypeDefinitionBasedNodeBuilderConfiguration (this is used to e.g. configure which Optionals to instantiate) with a default InstantiationStrategy (the TypeDefinitionBasedNodeBuilderConfiguration.builder(default) parameter), possibly delegating to new DefaultInstantiationStrategy(MethodInstantiationLogic.SHARED) (this is the default for the overload that takes none). This should allow you to observe each UaNode instantiation when the instances are made. Then within the InstantiationParameters you will have the InstanceDeclaration and can getModellingRule and check is it ModellingRule.Mandatory.

I guess maybe some day we might have an easier way of doing this, but good if you already found out a way to check that.