I do not fully understand the question, or the scale, also I do not get what is different between "uptime" and "cumulative uptime".
Did you want to report to them once a week saying that the update for device x is 26 days, or do you want to report every week saying that on Monday is was 26 day, on Tuesday is was 27 days...etc.
Does "cumulative" mean you want to add 2+ device uptimes together, eg router A and router B both have uptimes of 26 days, so lets report that "routers" has an uptime of 52 days?
So, if its for a small scale (eg, under 20) can you create a new "Sensor Factory" that contains all devices uptimes you want? You can then use this object in a report, and email from PRTG to the client directly. Or give the client access to a PRTG map that contains this data.
But if its for more than 20, or changes all the time, then you certainly want to look at something like PrtgAPI ( https://github.com/lordmilko/PrtgAPI ). In short it would be enumerating through all the sensors in a group (in the devices in that group) and returning their uptime, then dump it out somewhere. This will work for a single point in time, or if you want to report once a week on historical uptimes also, or if you want to group all the uptimes into one big number.