Artifacts not only combine data and process into the same semantic entities that evolve according to precise state-based lifecycles, but they also provide casual users with a simple and intuitive framework for modeling, exe- cuting, and managing business processes. Despite the surge in interest, artifacts still lack effective languages and tools to define, manipulate and interrogate their instances. Most existing artifact languages and notations focuses only on modeling and executing artifact-based processes in which data have discrete values and fail to handle artifacts in which their data include continuous data streams. In this paper, we present the Continuous Artifact Query Language (CAQL) for modeling stream-based artifacts and semantically manipulating and interrogating their instances.