This paper proposes a novel generation system based on multi-agent collaboration, integrating techniques such as language style few-shot transfer, the Big Five personality model, and a God perspective + plot setting. The system achieves personalized settings and diverse automated novel generation by leveraging the collaborative work of multiple agents: a plot construction expert, a description generation expert, a God perspective narrator, and a character generation expert. Each agent communicates through API calls to efficiently generate text. First, the plot construction expert generates the main storyline; next, the description generation expert provides detailed scene and atmosphere descriptions; the God perspective narrator offers comprehensive background information and character inner activities; and the character generation expert, based on the Big Five personality model, creates behavior and dialogue that align with the character’s personality. The language style few-shot transfer enables the system to quickly adapt to specific language styles, the Big Five personality model provides a psychological basis for character creation, and the combination of the God perspective and plot setting enhances the story’s coherence and depth. Experimental results show that the system excels in generation quality, text diversity, and character vividness, offering new insights and methods for automated novel generation.