Safety analysis for nuclear reactors is essential to licensing and operational assurance, yet it remains resource-intensive and dependent on multidisciplinary expertise. These analyses require the integration of diverse engineering documents, structured logical modelling, and domain-specific reasoning under strict regulatory constraints. To support advanced reactor safety workflows, an agentic architecture leveraging generative artificial intelligence (AI) is under development at Idaho National Laboratory. The architecture decomposes safety analysis processes into specialized agents, each assigned a defined analytical role and operating under task-specific reasoning constraints. This multiagentic structure enables the separation of data interpretation, structured safety analysis, and logical model construction into coordinated but distinct components. Within this framework, this paper presents a schema-guided structured reasoning approach for fault tree generation. Rather than generating fault trees directly from system descriptions, the proposed approach constructs a safety analysis schema that captures functions, configurations, systems, components, and associated failure modes as an intermediate representation of engineering knowledge. The generated schema can be reviewed and refined by analysts through interactive interfaces that support evidence tracing and knowledge validation prior to fault tree construction. The reviewed schema is subsequently transformed into fault tree artifacts through structured logic conversion rules. To support model validation, a minimal cut set (MCS)-based evaluation methodology is introduced to compare generated and reference fault trees based on logical equivalence rather than structural similarity. The proposed approach provides a transparent and human-reviewable workflow for AI-assisted fault tree generation and validation.