Even though boron clusters are quite significant, bare boron clusters as ligands in chemical compounds are still unknown. Triangular B-3 is a key constituent of all-boron clusters and widely applied in the boron compounds. As a basic step toward understanding the assembly and stabilization of bare all-boron clusters and the possibility of their fusion during the cluster-assembly process, we made the first attempt to assemble the smallest bare all-boron unit B-3(-). Both the "homo-decked sandwich" and "hetero-decked sandwich" schemes were applied to the assembly of B3- at the B3LYP/6-311++G(d, p) level. For all the considered alkali- and alkaline earth metals, B-3(-) can only be assembled in "hetero-decked sandwich" scheme (e.g., CpMB3q-) so as to avoid cluster fusion, whereas it cannot be assembled in the traditional "homo-decked sandwich" scheme (B3MB3q-) because of thermodynamic and kinetic instability. Various assembled species in extended frameworks are designed. In particular, the dimerization of the hetero-decked sandwich-like CpMB3q- could lead to a new type of antiaromatic triple-decker sandwich-like complexes CpMB6MCP(2q-) that contain the all-boron antiaromatic unit B-6(2-). Our work supports the experimental identification of the B-6(2-) anion (with M+ counterions) in a photoelectron spectroscopy study. Additionally, the electronic and structural properties of B-3(-) are well conserved during cluster-assembly, characteristic of a "superatom" feature. Our results are expected to be helpful for understanding the assembly and stabilization of bare all-boron cluster chemistry. Also, our work should give insight toward designing and understanding bare boron clusters as potential new ligands for coordination chemistry and as new building blocks for materials science. Interestingly, our results should provide hints to embellish, functionalize, isolate, and protect bare all-boron clusters.