Metal-organic frameworks (MOFs) consist of metal nodes that are connected by organic linkers. They are thus highly chemically tunable materials given the broad range of potential linkers and nodes that can be chosen for their synthesis. Their tunability has recently sparked interest for the development of new MOF photo-catalysts for energy-related applications such as hydrogen (H2) evolution and CO2 reduction. The sheer amount of potentially synthesizable MOFs requires to define descriptors that allow to predict their performance with this aim. Herein we propose a systematic computational protocol to determine two energy-based descriptors that are directly related to the performance of a MOF as a photocatalyst. These descriptors assess the UV-vis light absorption capability and the band energy alignment with respect to redox processes and/or co-catalysts energy levels. High-throughput screening based on cost-effective computations of these features is envisioned to aid the discovery of new promising photoactive systems.