In response to the challenges of multi-source heterogeneous data fusion, inadequate adaptability in dynamic comprehensive assessment, and insufficient integration of domain-specific knowledge in transmission line hazard analysis, this paper proposes a large model framework for hazard comprehensive assessment and analysis based on a transmission line visualization platform. The framework establishes a multi-agent collaborative system: the Hazard Comprehensive Assessment Agent achieves a panoramic perception of hazard characteristics through dynamic correlation analysis of hazard data reported by monitoring devices and related multi-dimensional data; the Visualization Platform Operation Data Analysis Agent periodically acquires platform operation and equipment status information to evaluate and analyze the overall status; the Knowledge Enhancement Agent employs a hybrid semantic-keyword retrieval mechanism and a parent-child block context recall model to accurately retrieve relevant professional knowledge, significantly enhancing the large model’s understanding of domain-specific knowledge; the Visualization Platform Device Ledger Management Agent precisely obtains detailed device information through device ledgers and hazard tags, enabling control operations on devices. This paper designs this collaborative framework, demonstrating outstanding performance in hazard comprehensive assessment, visualization platform data analysis, and hazard analysis and management. The research outcomes provide effective technical support for the intelligent comprehensive assessment and analysis of transmission line hazards, significantly enhancing the intelligent level of transmission operation and maintenance in power grids.