It depends a lot on what you are doing. I think what you mean is that you are cutting a protein core out of a larger structure, and want to design loops between the extracted secondary structure elements. In that case, you want loop modeling. There are a couple of loop modeling tutorials. Briefly, you’ll need to modify your input PDB to have dummy coordinates for the N-CA-C-O backbone atoms for the “new” loop residues, and make sure to run loop modeling with the -build_initial flag (at least the first time, it won’t be necessary for later refinements). The demos/public/AnchoredDesign/loop_length_changing/loop_extend_in demo explains narrowly how to create loops where there were none.
FloppyTail is useful to suggest possible structures for multidomain proteins with extremely unstructured linkers, especially when you have constraints on the interactions between the known folded domains. In some cases ab initio folding is appropriate if the linkers are expected to be highly structured, even if you already know some portions of the structure.
.
