The digital ecosystem’s steady state is characterized by sophisticated operational dynamics, interrelationships, and disruption scenarios which need to be employed. The complexity and fluidity of an organization’s operations have increased over time leading to the need for developed enterprise systems architectures. This study presents a multi-agent system (MAS) architecture framework for modeling, stress-testing and measuring the stability of distributed enterprise architectures. MAS aids in providing realistic architectural views by allowing mapping of services, data processes and control elements to self-forming software units called agents. The framework captures emerging from local failures, communication delays or breakdowns. It facilitates the realization of multiple architectural styles including microservices, service oriented and event driven architectures or topologies and allows them to be compared under various levels of disruptions. Architectural resilience is quantified by measuring performance indicators like Mean Time To Recover (MTTR), cascading failure probability, service availability, and SLA violations. The analyses reveal important design compromise between modularity, redundancy as well as coordination overhead. The simulations fulfill requirements for modern adaptive systems, where enterprise architects can analyze architectural strength, inform data-driven transformation planning, and proactively design stability-increasing interventions. The results establish a basis for incorporating AI based adaptation and predicting oriented enterprise architecture management.