A large fraction of known terrestrial-size exoplanets located in the Habitable Zone of M-dwarfs are expected to be tidally-locked. Numerous efforts have been conducted to study the climate of such planets, using in particular 3-D Global Climate Models (GCM). One of the biggest challenges in simulating such an extreme environment is to properly represent the effects of sub-grid convection. Most GCMs use either a simplistic convective-adjustment parametrization or sophisticated (e.g., mass flux scheme) Earth-tuned parametrizations. One way to improve the representation of convection is to study convection using Convection Resolving numerical Models (CRMs), with an fine spatial resolution . In this study, we developed a CRM coupling the non-hydrostatic dynamical core WRF with the radiative transfer and cloud/precipitation models of the LMD-Generic climate model to study convection and clouds on tidally-locked planets, with a focus on Proxima b. Simulations were performed for a set of 3 surface temperatures (corresponding to three different incident fluxes) and 2 rotation rates, assuming an Earth-like atmosphere. The main result of our study is that while we recover the prediction of GCMs that (low-altitude) cloud albedo increases with increasing stellar flux, the cloud feedback is much weaker due to transient aggregation of convection leading to low partial cloud cover.