Abstract: Aiming at the complex challenges prevalent in the geographical indication agricultural product supply chain - such as insufficient motivation for quality upgrading, difficulties in realizing brand value, and low market coordination efficiency - particularly the decision-making barriers caused by information asymmetry, this study constructs a multi-agent dynamic decision-making model based on the trapezoidal fuzzy non-cooperative-cooperative biform games. By quantifying uncertainties within the supply chain, this model optimizes the decision-making process, enabling supply chain members to make more reliable decisions. Simultaneously, it designs a profit distribution mechanism that balances efficiency and fairness, incentivizes cooperation among supply chain members, and enhances supply chain resilience, thereby establishing a sustainable cycle of "quality-value-benefit" improvement. Case results demonstrate that the model increases the profits of supply chain members and promotes the development of new quality productive forces in agriculture by coordinating multi-stakeholder decision-making and benefit distribution within the agricultural product supply chain.