SARL is a versatile general-purpose agent-oriented programming language designed to offer fundamental abstractions for agent-oriented programming (AOP) while remaining agnostic to specific agent architectures. The chapter details SARL’s core features, including dynamic agent reconfiguration, domain-oriented social environments, and support for collective agents through holonic structures. The metamodel underlying SARL is explored to provide essential abstractions for AOP. Extensions and ecosystem developments are discussed, highlighting significant enhancements such as goal reasoning agents, explainable agents, and simulator architectures that expand SARL’s applicability across various domains. A brief comparison with other agent frameworks is provided, evaluating SARL against established criteria for multi-agent systems. The chapter also addresses future perspectives, outlining challenges and proposed solutions for integrating AI models, enhancing explainability, and supporting meta-programming and protocol-based interactions. The SARL roadmap is presented, emphasizing continuous improvement and community participation to drive the evolution of AOP with SARL.