In recent years, Computational Thinking (CT) made its way into K-12 as a key set of 21st-century skills. However, the issue of assessing student's learning of CT remains a challenging one: it is indeed not clear how to achieve a comprehensive evaluation that is able to assess not only the technical and domain-specific skills, but also the domain-independent meta-skills of CT. Finding a solution to this issue is crucial, because assessment not only determines whether or not educational goals are being met, but also supports the design of a curriculum. Multiple forms of assessments have been proposed within the main perspective of programming skills, neglecting the possible generalization of CT training on other skills. In this work, we first review the existing approaches to CT assessment. Then, we discuss what non-technical skills (relational skills and cognitive life-skills) should be included in a comprehensive CT assessment framework. Moreover, we provide an indication of possible solutions to capitalize on the existing experiences in order to achieve a feasible integration of different types of assessment.