This article presents the Problem-Solution-Territory (PST) method. This reproducible hybrid approach combines LLM-mediated semantic assistance, expert-defined rules, functional agents, and deterministic document generation to transform territorial evidence into context-specific municipal climate adaptation portfolios. The method comprises four phases: (1) territorial knowledge modeling and local transition diagnosis; (2) expert modeling and adaptation solution archetypes; (3) territorial knowledge and technology-supported integration infrastructure; and (4) participatory validation, consolidation, learning, scaling up, and methodological refinement. Applied in 125 municipalities of Antioquia, Colombia, PST harmonized 60 generalized climate-related problems and 70 transferable solution archetypes. Each phase is defined by explicit inputs, processing procedures, validation criteria, and outputs that support traceability throughout the workflow. The implementation integrates a JSON-based knowledge repository, geospatial extract-load-transform processes, expert rules, bounded LLM-mediated assistance, script-based automation, and Jinja templates for deterministic document rendering. The resulting workflow supports semantic normalization, controlled data integration, source traceability, and layered technical, institutional, and community validation. The method is designed as an expert-governed automation procedure rather than an autonomous decision-making system. It can be adapted to other territories by recalibrating its taxonomies, applicability rules, data sources, and validation instruments.