Golden Gate Assembly depends on the accurate ligation of overhangs at fragment fusion sites to generate full-length products with all parts in the desired order. Traditionally, fusion-site sequences are selected by using validated sets of overhang sequences or by applying a handful of semi-empirical rules to guide overhang choice. While these approaches allow dependable assembly of 6–8 fragments in one pot, recent work has demonstrated that comprehensive measurement of ligase fidelity allows prediction of high-fidelity junction sets that enable much more complex assemblies of 12, 24, or even 36+ fragments in a single reaction that will join with high accuracy and efficiency. In this chapter, we outline the application of a set of online tools that apply these comprehensive datasets to the analysis of existing junction sets, the de novo selection of new high-fidelity overhang sets, the modification and expansion of existing sets, and the principles for dividing known sequences at an arbitrary number of high-fidelity breakpoints.