Realizing more sustainable food, feed, and bioenergy systems will require interventions such as increased recycling of nutrients and coordination of biomass flows among farms. Innovative tools to explore the co-benefits and trade-offs of improving flow circularity in agro-food systems at different scales are needed to better understand the efficacy of these sustainability solutions. Here, we applied the FAN ("Flows in Agro-food Networks") agent-based model to simulate contrasting scenarios of material flows locally in a small farming region of France. These scenarios aim to enhance: (1) best management practices at the farm scale; (2) organic material recycling and biogas production collectively across the agricultural landscape; and (3) system redesign toward complete local circularity through crop and livestock symbiosis, fewer livestock, and elimination of external inputs. Scenario simulation outcomes are assessed in terms of their degree of circularity and food production. We find that best management practices at the farm scale and collective solutions for recycling (organic fertilization and anaerobic digestion) substantially improved the degree of circularity by tightening the local nitrogen (N) cycle without affecting food production. Among other co-benefits, changes in farm rotations to feed livestock locally increased the degree of circularity without appreciably impacting food production. The maximum circularity scenario showed considerable potential to mitigate greenhouse gas (GHG) emissions, however, they involved large trade-offs with food production that were even more pronounced with fewer livestock animals. Although regulating livestock numbers combined with eliminating chemical fertilizers was the most effective at mitigating GHG emissions, when applied simultaneously it substantially impacted food and bioenergy production. Such trade-offs for soil fertility demonstrate the importance of coupling crops and livestock for reaching self-sufficient circular systems. Our study illustrates how the FAN agent-based model can be applied to account for multiple types of interactions involved in transitions toward circularity in local agro-food systems, including the potential for co-benefits, and unintended consequences of interventions.