I think you are asking what is the hidden-variable interpretation/description of unsharply defined observables when, in the standard interpretation of QM, these observables are incompatible and the quantum state is not eigenstate of both variables.
For systems quantistically defined in finite dimensional Hilbert spaces, existence of non commuting elementary observables already implies contextuality in realistic hidden variable interpretations of QM, in view of the Kochen Specker theorem.
That theorem assumes hypotheses on the (non-quantum) valuation functions
$v_\lambda(A)$ only for pairs of compatible observables $A,B$ and proves that the theory must be contextual. That is the powerfulness of the theorem.
How the quantum phenomenology of uncompatible observables is described in hidden variable theories is matter of each concrete hidden variable theory (independently from the fact that all these theories must be contextual).
Contextuality alone is by no means able to provide an intepretation of the quantum phenomenology of incompatibke variables, it is only a necessary constraint.
Presumably, a description should be related to the epistemic (vs ontic) stochasticity of the hidden-variable state in a hidden variable theory. In other words an explanation should be given in terms of the probability distribution which describes our knowledge of the hidden variable $\lambda$.
A problem is the the (standard version of the) KS theorem applies to finite dimensional Hilbert spaces and bounded observables. Therefore it cannot be directly applied to, e.g., a particle on the real line (where the Heisenberg principle applies!) without stronger hypoteses on the set of observables and on the valuation function.
However, you can focus on hidden variabe theories explicitely constructed to deal with this case. I know only one of them: the Bohm theory. It is contextual and it includes an explanation of the Hesinberg principle