you're (ab)using their ability wrong. just have them try over and over without spending any karma at all.
so far as i can tell, the traps created (or modified) are permanent. and the power can be used as often as they want. so they can just create a trap, and if they didn't roll well, they just modify the same trap over and over until they do roll well. without using karma, one in 64 attempts will score double 8, which will on average give them mid 20s to work with... and while 1 in 64 would be terrible for someone that takes hours or days to make a trap, that's not a big deal for a creature that can make or change a trap every few seconds. and now we're looking at almost 6 of such traps per hour... a single scurrier that has been hanging out somewhere for a day (assuming they don't need sleep) could have laid or repaired over 100 such traps easily. a mechanical trap where they aim for "typical" ratings, for example, could be detection 5, disarm 7, initiative 8, and damage 6, and would be well within their grasp assuming they retry until they get a good one.
also note that approximately every 4100 attempts, they'll come up with a rating 32 + 2d8 trap (again, without spending karma). actually, more often than that, because there's a chance their bonus die will roll an 8 that i didn't account for... that would be something like 4 such traps per day of solid work (this requires that we presume the scurrier is obsessed with making traps, but that sounds plausible to me
)
also, they explicitly are making traps that aren't so much deadly as painful, so if the trap is somewhat lacking in the ability to deal lots of damage in order to have decent values in other areas, i'd say that fits rather well
so really, i think you'd be *easily* justified in having a very large number of traps (over 100) at around 26 total rating, a large number (20+) at 34 total rating, and a few (7ish?) at around 42 total rating, even for an area that has only had a single scurrier preparing for one day. for an area that has had multiple scurriers, or a single scurrier for longer, you would shift more of the traps to the high end or multiply the number of traps appropriately. i don't know about you, but that sounds more than adequate for a circle 3 enemy to me.
you'd want to grab someone who is better at probability math than me to get more exact numbers, but basically i think scurriers will be doing just fine. particularly when you remember that they can tip the odds in their favour by hanging around and cursing your luck.