Complex multi-agent negotiation systems with high security, efficiency, and trust requirements have been developed in response to the rapid development of e-commerce platforms. Strong security measures for ensuring data integrity and confidentiality in distributed environments are absent in conventional negotiation models. This paper attempts to use graph theory, game theory, and cryptographic protocols in a blockchain environment to develop a discrete mathematical framework for secure multi-agent negotiation systems. Edges in a weighted graph are used to represent negotiation interactions with utility functions, while agents are represented as vertices in the graph. For optimal decision-making by multiple conflicting agents, a game theory approach using Nash equilibrium is used. Hash function validation of blockchain is used for ensuring security, and numerical analysis is used to test a new secure multi-agent negotiation algorithm (SMANA).