Relaxin family peptide receptors (RXFPs) are the potential therapeutic targets for neurological, cardiovascular, and metabolic indications. Among them, RXFP3 and RXFP4 (formerly known as GPR100 or GPCR142) are homologous class A G protein-coupled receptors with short N-terminal domain. Ligands of RXFP3 or RXFP4 are only limited to endogenous peptides and their analogues, and no natural product or synthetic agonists have been reported to date except for a scaffold of indole-containing derivatives as dual agonists of RXFP3 and RXFP4. In this study, a new scaffold of tricyclic derivatives represented by compound 7a was disclosed as a selective RXFP4 agonist after a high-throughput screening campaign against a diverse library of 52,000 synthetic and natural compounds. Two rounds of structural modification around this scaffold were performed focusing on three parts: 2-chlorophenyl group, 4-hydroxylphenyl group and its skeleton including cyclohexane-1,3-dione and 1,2,4-triazole group. Compound 14b with a new skeleton of 7,9-dihydro-4H-thiopyrano[3,4-d][1,2,4]triazolo[1,5-a]pyrimidin-8(5H)-one was thus obtained. The enantiomers of 7a and 14b were also resolved with their 9-(S)-conformer favoring RXFP4 agonism. Compared with 7a, compound 9-(S)-14b exhibited 2.3-fold higher efficacy and better selectivity for RXFP4 (selective ratio of RXFP4 vs. RXFP3 for 9-(S)-14b and 7a were 26.9 and 13.9, respectively).