We propose the development of a multigrid open-source operational forecast system, with the oceanic region off Rio de Janeiro state as our first case study. The system is composed by three nested ROMS domains, employing Mercator Ocean at the lateral open boundaries of the outermost grid, and GFS (Global Forecast System) at the surface open boundary of all grids. Two schemes were built for the operational forecast workflow, free (without Data Assimilation) and assimilated (with data assimilation). For evaluation, we performed hindcast simulations using Mercator and ERA5 as input data for different parts of the Southeastern Brazilian oceanic region, covering various regimes of the unique Brazil Current system. The modelled data was compared against velocity components from three oceanic buoys, SST from MUR (Ultra-high Resolution Sea Surface Temperature), and SSH from AVISO. The comparison against the buoys showed a good representation of the velocity field by ROMS, performing better than Mercator for most scenarios. ROMS also presented good performances for SST and SSH. For the three-nested forecast system, the velocity fields presented comparable magnitudes against Mercator, without the presence of jet-like instabilities, indicating a stable system. We also implemented the I4D-Var method in a 2-day analysis for the simulation of 7-day forecasts, finding an improvement in the modelled forecasts initialized from the assimilated analysis for both velocity components, with an overall reduction in the RMSE and bias, and an increase in correlation.