The relevance stems from the complexity of managing the weakly structured socio-economic systems of municipalities, where a multitude of divergent interests must be taken into account. Existing models are unable to reflect this complexity fully, creating a need for new methodologies. The scientific problem lies in the methodological gap between the need for a comprehensive territorial analysis and the lack of a tool that formalises the interrelationship between the system's key components, including resources, rules, values, and multi-agent interactions. The research methods are based on cognitive modelling. The work includes conceptualisation, construction of structural-functional diagrams and integration of the model with AI tools for hypothesis verification and predictive calculations.  Research results: a concept of a cognitive model was developed, the core of which is a structural-functional scheme of five indicator groups: resources, actors, mechanisms, values, rules; a system of six principles for structuring the model (e.g., management logic, Pareto principle) was defined, enabling its decomposition into 17 spheres of local government activity; the dynamic logic of managed development through the “As Is – What Is Being Done – To Be” cycle is described; practical significance is confirmed by the model's implementation as the core of a specialized SaaS platform for strategic decision support, where it is undergoing testing. Conclusions: the research proposes a holistic methodology for the analysis and predictive modeling of municipal development. The developed model integrates objective data and socio-cultural context, and its implementation in an industrial platform confirms its practical value for enhancing the substantiation of managerial decisions.