There is an outline of the design here:
Each probe weighs 16 grams and consist of three 98 mm diameter aluminum sheets at 90 degrees to each other, effectively forming a sphere. The intent is to be lightweight and have a constant cross section, independent of orientation to the velocity direction so that atmospheric drag can be measured in-situ. RF modeling predicts that the atmospheric probes will have a radar cross section equal to 1U CubeSats, which have been tracked on-orbit many times.
The designers describe them as:
...a set of 28 individually releasable atmospheric probes. These lightweight circular probes, similar in size to CDs, spring open into spherical objects. Due to the probes’ large surface areas being exposed to the atmosphere, they lose altitude quickly and burn up in a matter of weeks.
The second link has a helpful picture - they look like the X/Y/Z planes crossing the a sphere, and it makes sense that this would mean a constant circular aerodynamic cross-section regardless of how it tumbles around. This shape also makes them a corner reflector, which means that they will be easily trackable by radar.
How do 28 of these fit?
The cubesat carrying these is only 1.5 U. If they were real spheres each with a volume of $\frac{4}{3} \pi r^3$ with a close packed volume fraction of $\frac{\pi}{3 \sqrt{2}}$ they would need 18.6 U.
However, the probes are stored flat and then pop open when deployed. This explains how they can all fit in - 98mm diameter would fit nicely inside a standard cubesat of 1.5U size. Each of the three sheets is probably less than 0.5mm thick, judging by the weight, so there is plenty of space for 29 of them to be stored one alongside another and still leave room for the rest of the workings of the satellite.
