<![CDATA[The transformation of 21st century education demands assessments that not only measure basic cognitive learning outcomes but also encourage deep engagement of higher-order thinking. Computational Thinking (CT), as a systematic and logical thinking approach to problem solving, offers great potential to be integrated in learning assessment design. Thus, this study aims to explore the development of CT-based assessments in promoting deep learning through analyzing relevant scientific literature. The method used is a descriptive qualitative literature review, by analyzing articles from various scientific sources that discuss CT theory, components, and implementation in the context of education and assessment. Data were collected from articles indexed in Google Scholar, ScienceDirect, and other academic sources, then analyzed thematically and comparatively. The results of the discussion show that CT-based assessments, which involve the components of decomposition, abstraction, pattern recognition, algorithms, and generalization, are able to reveal students' thinking processes in depth. This assessment not only measures knowledge, but also encourages reflective, creative, and transdisciplinary skills. In the context of project-based learning, such as the design of an AI-based automatic air purifier system, CT assessments can serve as a tool that encourages students to contextually understand, design, and evaluate solutions. In conclusion, CT-based assessments contribute significantly to deep learning and need to be systematically developed in educational practice. This is in line with the need to form a generation of adaptive, critical, and solutive learners in facing the challenges of a complex and technology-based modern world. ]]>
