The first half of this paper is largely expository, wherein we present a systematic combinatorial approach to the theory of polynomial (semi)invariants and multilinear invariants of several vectors and covectors, for the classical groups. This culminates in a graphical description of graded linear bases. By applying well-known results of lattice path combinatorics to Weyl's fundamental theorems of classical invariant theory, we write down Stanley decompositions and Hilbert-Poincare series in terms of families of non-intersecting lattice paths, enumerated with respect to certain corners. In the second half of the paper, we revisit the (semi)invariants in the first half as a special case of a much broader phenomenon. On one hand, polynomial invariants of a group $H$ can be generalized to modules of covariants, i.e., $H$-equivariant polynomial functions between $H$-modules. On the other hand, from the perspective of Roger Howe's theory of dual pairs, these modules of covariants can be viewed as infinite-dimensional simple $(\mathfrak{g}, K)$-modules. This suggests an expanded program in which our goal is to apply combinatorial techniques involving lattice paths in order to write down Hilbert series for arbitrary unitarizable highest-weight $(\mathfrak{g},K)$-modules. As a preview of future work in this program, we present examples showing how modules of covariants -- even those which are not Cohen-Macaulay, and therefore which we would not expect to be combinatorially nice -- can be decomposed in terms of lattice paths. We also extend these methods beyond the classical groups.