When applying model-driven engineering in an agile environment, new requirements continuously expand the domain scope and trigger an extension of the concepts covered by the Domain-Specific Language (DSL). While programming languages streamline code extension and reuse through libraries, similar approaches for DSLs are more complex or target a specific language. We present and discuss an approach for designing DSLs with a self-extension mechanism to enable model library creation and seamless reuse. We use the self-extension mechanism to introduce concepts in models, gather reusable models into a library, and provide an infrastructure for its usage. We explain our language-specific realization of the self-extension mechanism using a DSL and discuss model libraries of the DSL with a use case from practice. The approach provides more flexibility for agile model-driven engineering. It enables application modelers to introduce and reuse concepts via models without changing the DSL, reducing the communication overhead within the development team.