A series of SARS-CoV-2 main protease (SARS-CoV-2-M^pro) inhibitors were modeled using evolutive grammar algorithms. We have generated an automated program that finds the best candidate to inhibit the main protease, M^pro, of SARS-CoV-2. The candidates were constructed based on a pharmacophore model of the above-mentioned target; relevant moieties of such molecules were modified using data-basis sets with similar chemical behavior to the reference moieties. Additionally, we used the SMILES language to translate 3D chemical structures to 1D words; then, an evolutive grammar algorithm was used to explore the chemical space and obtain new candidates, which were evaluated via the binding energy of molecular coupling assays as an evaluation function. Finally, sixteen molecules were obtained in 3 runs of our program, three of which show promising binding properties as SARS-CoV-2-M^pro inhibitors. One of them, TTO, maintained its relevant binding properties during 100 ns molecular dynamics experiments. For this reason, TTO is the best candidate to inhibit SARS-CoV-2-M^pro. The software we developed for this contribution is available at the following URL: http://github.com/masotelof/GEMolecularDesign.