I finally got around to testing my drag knife out. This particular unit uses a very small blade that can be extended down past the end of the holder. You adjust the blade up and down to match the thickness of the paper you are cutting, ideally so you just barely make it through the material.
The blade has a small offset to it so it trails behind the movement of the holder. That helps the blade self orient in the direction of motion. However when you change directions the blade will lag behind and follow a distorted path. If the change in direction is large enough then the blade can bind up and cause a small tear in the material.
To compensate I'm using a software tool that distorts the path of the housing to correct the path of the blade. For this small of a blade this correction is not critical, but it does seem to help compared to an uncorrected cut.
The end results were not very good, all the corners are soft, no matter how much compensation I used. And things were very inconsistent. Watching this for a while I saw two issues. First my original tool holder was weak and flexed easily and it did not hold the tool very well. And second I have a hard time getting a consistent height above the material.
I have not solved the height problem, I need to experiment more with that. But I did print up a new tool holder that is much stronger and that firmly clamps the tool. It still wiggles a bit but it feels much more solid. I need to test it out to see how much it helps but I have a feeling it will work well. Part of my struggle is because there is such a large distance between my Z carriage and the work surface. I need to extend everything out really far to make it all fit. If this is not strong enough I may look into raising the surface of the bed up 6" or so.
Hopefully tomorrow I can find the time to test this out and then we will know for sure.