We analyse the evolution of the red sequence in a sample of galaxy clusters at redshifts $0.8<z<1.5$ taken from the HAWK-I Cluster Survey (HCS). The comparison with the low-redshift ($0.04<z<0.08$) sample of the WIde-field Nearby Galaxy-cluster Survey (WINGS) and other literature results shows that the slope and intrinsic scatter of the cluster red sequence have undergone little evolution since $z=1.5$. We find that the luminous-to-faint ratio and the slope of the faint end of the luminosity distribution of the HCS red sequence are consistent with those measured in WINGS, implying that there is no deficit of red galaxies at magnitudes fainter than $M_V^*$ at high redshifts. We find that the most massive HCS clusters host a population of bright red sequence galaxies at $M_V < -22.0$ mag, which are not observed in low-mass clusters. Interestingly, we also note the presence of a population of very bright ($M_V < -23.0$ mag) and massive (${\log(M_*/M_\odot)} > 11.5$) red sequence galaxies in the WINGS clusters, which do not include only the brightest cluster galaxies and which are not present in the HCS clusters, suggesting that they formed at epochs later than $z=0.8$. The comparison with the luminosity distribution of a sample of passive red sequence galaxies drawn from the COSMOS/UltraVISTA field in the photometric redshift range $0.8<z_{phot}<1.5$ shows that the red sequence in clusters is more developed at the faint end, suggesting that halo mass plays an important role in setting the time-scales for the build-up of the red sequence.