The SARS-CoV-2 pandemic has highlighted the need for improved technologies to help control the spread of contagious pathogens. While rapid point-of-need testing plays a key role in strategies to rapidly identify and isolate infectious patients, a cornerstone for any disease-control strategy, current test approaches have significant shortcomings related to assay limitations and sample type. Direct quantification of viral shedding in exhaled particles may offer a better rapid testing approach, since SARS-CoV-2 is believed to spread mainly by aerosols. It potentially measures contagiousness directly, the sample is easy to obtain, its production can be standardized between patients, and the limited sample volume lends itself to a fast and sensitive analysis. In view of these benefits, we developed and tested an approach where exhaled particles are efficiently sampled using inertial impaction in a micromachined silicon chip, followed by an in-situ RT-qPCR molecular assay to detect SARS-CoV-2 shedding. We demonstrate that sampling subjects using a one-minute breathing protocol, yields sufficient viral RNA to detect infections with a sensitivity comparable to standard sampling methods. A longitudinal study revealed clear differences in the temporal dynamics of viral load for nasopharyngeal swab, saliva, breath, and antigen tests. Overall, after an infection, the breath-based test is the first to consistently report a negative result, putatively signaling the end of contagiousness and further emphasizing the potential of this tool to help manage the spread of airborne respiratory infections.