Cyclic peptides (CPs) are a promising class of drugs because of their high biological activity and specificity. However, the design of CP remains challenging due to their conformational flexibility and difficulties in designing stable binding conformation. Herein, we present a high-throughput MD screening (HTMDS) process for the iterative design of stable CP binders with a combinatorial CP library composed of canonical and non-canonical amino acids. As a proof of concept, we apply our methods to design CP inhibitors for the bromodomain (BrD) of ATAD2B. 698,800 CP candidates with a total of 25,570 ns MD simulations were performed to study the protein-ligand binding interactions. The binding free energies (ΔGbind) estimated by MM/PBSA approach for eight lead CP designs were found to be low. CP-1st.43 was the best CP candidate with an estimated ΔGbind of −28.48 kcal/mol when compared to the standard inhibitor C-38 which has been experimentally validated and shown to exhibit ΔGbind of −17.11 kcal/mol. The major contribution of binding sites for BrD of ATAD2B involved the hydrogen-bonding anchor within the Aly-binding pocket, salt bridging, and hydrogen-bonding mediated stabilization of the ZA loop and BC loop, and the complementary Van der Waals attraction. Our methods demonstrate encouraging results by yielding conformationally stable and high-potential CP binders that should have potential applicability in future CP drug development. Communicated by Ramaswamy H. Sarma