Stress is a major public health concern and a severe threat to everyone. Facilitated by their powerful sensing capabilities, mobile devices may assist individuals in coping with stress. Building on existing studies and mobile apps supporting stress coping, we propose the design of a mobile coping assistant that uses multimodal sensor data to reduce its user's stress. Based on sensor data, a mobile coping assistant (1) warns the user about elevated stress, (2) delivers a fundamental understanding of why they are currently stressed, (3) recommends targeted coping strategies to encourage and train effective coping behavior, and (4) executes automated actions to reduce stress exposure. The presented design comprises an architecture, good practices for designing the architectural components, and an algorithm for selecting adequate coping actions and recommendations. A prototypical instantiation indicates opportunities and challenges. Future research should evaluate the short-and long-term effectiveness of mobile coping assistants in the field.