In this work, we consider the Golumbic, Kaplan, and Shamir decision sandwich problem for a property Π: given two graphs G 1 = (V, E 1 ) and G 2 = (V, E 2 ), the question is: Is there a graph G = (V, E) such that E 1 ⊆ E ⊆ E 2 and G satisfies Π? The graph G is called sandwich graph. Note that what matters here is just the "filling" of the sandwich. Our proposal is to try different kinds of "bread" for each chosen special sandwich filling. In other words, we focus on the complexity of sandwich problems when, beforehand, it is known that G i satisfies a property Π i , i = 1, 2. Let (Π 1 , Π, Π 2 )-sp denote the sandwich problem for property Π when G i satisfies Π i , called sandwich problem with boundary conditions. When G i is not required to satisfy any special property, Π i is denoted by * . A graph G is (k, ℓ) if there is a partition of V (G) into k independent sets and ℓ cliques. It is known that ( * , (k, ℓ), * )-sp is NP-complete, for all k + ℓ greater than 2. In order to motivate this new work proposal, in this paper we describe polynomial-time algorithms for three sandwich problems with boundary conditions: (perfect, (k, ℓ), poly-