Two datasets are provided. They were developed to simulate and assess sustainability and resilience of scenarios of exchanges between arable and livestock farms, i.e. scenarios of territorial crop-livestock systems. The data are a result of a research collaboration between the scientist from INRAE, agricultural advisers from Chamber of Agriculture of Pays de la Loire (CRAPL) and a collective of five arable and two livestock farmers located in the district of Pays de Pouzauges (Vendée department, western France). All participants formed part of the DiverIMPACTS project (https://www.diverimpacts.net/) that was built on existing and newly developed initiatives related to crop diversification in Europe. The first dataset includes data prepared to be used as inputs of MAELIA (http://maelia-platform.inra.fr/), a spatial agent-based integrated modelling framework. The second dataset contains the raw outputs of MAELIA simulations and the respective indicators at the farm, group (AF and LF) and territory level (all farms). The data comprise multiple shape and csv files characterizing the 7 considered farms in terms of edaphic-climatic heterogeneity, cropping systems, farmers’ decision management rules and general information (e.g. crops, agricultural equipment, average crop yields). Data is reported for the baseline situation (current structure and functioning of the 7 farms) and three scenarios of legumes exchanges between AF and LF containing different cropping systems co-designed by scientists, agricultural advisers and the farmers.