Take this example:

extension E on String {
  void m() {}
}

Today, you can refer to m in a doc comment reference by the extension name (/// [E.m]), but not by the extended type's name (/// [String.m]).

I think it should be perfectly logical to refer to an extension member by the extended type's name. But there may be some quirks. E.g. given two extensions, E on List<int> and E on List<bool>, each with a member named m. What does /// [List.m] refer to? There are rules about resolving extension members and "specificity," which can probably apply without modification to comment references. We just have to start with a rule about instantiating the type to bounds. And probably make it non-nullable, so that it can refer to an extension E on String?.