I notice two things. Other folks here might have other ideas too.
(1) the RA tracking errors in both screenshots are larger than the declination error, and thus they determine the total RMS error -- recall that the total RMS error is the square root of the sum of the squares of each RA and declination RMS error if the noise for both axes are statistically independent. To get the total RMS number down, you would need to concentrate on the usual problems with the RA axis. The Declination error is very good, so the polar alignment is very accurate and you need not bother with that as a potential problem. Check if something in the RA axis has changed, like a not fully tightened RA clutch, imbalance in the RA axis (unless you are using a harmonic drive mount), or the differential flexure that I mention below.
If you don't have a problem with some direction but have problems with other directions, then check for "third axis balance." This YouTube video illustrates the issue quite well:
https://www.youtube.com/watch?v=zmla32ksH_I
(2) More importantly though, your screenshot shows much larger star trailing than a 0.75" type error should produce. What is your plate scale? If it is larger than 2" per pixel, the star trailing should not be so visible. That could mean that the guiding is tracking OK although not great but you still see trail in the main OTA. 0.75" is plenty good enough for 700mm type focal lengths, and certainly way more than enough produce round stars for a 300mm type focal length, unless the camera that you use have really tiny pixels (like 1 µm x 1 µm).
That seems to indicate that there might be differential flexure involved. I.e., the guide scope's optical axis is moving relative to the main scope's optical axis. Check how the guide scope is mounted to the main OTA. If you are using a dual saddle plate, I would focus on the flexure of the saddle plate, for example.
Unless you are using an OAG, make sure the guide scope is securely bolted to the main OTA. If you are using an OAG, check to make sure the optical train is completely tightened.
In both cases, make sure the optical train (from the telescope's metal back to the main camera) are fully tight. Check the guide scope too, although typical weights of the instrumentation attached to the guide scope are usually lightweight; but it is worth checking too since that too can show up as differential flexure.
So, try checking for those two things: why RA shows more error than Declination, and where the differential flexure can be coming from, since a 0.75" RMS error at the guide scope should not show up in an image on the main scope with a 300mm focal length.
By the way, if you look closely, your first enlarged image shows two distinct bright centers for that one star, and the bright centers are more than 1 pixel apart. With a 3" per pixel plate scale, for example, that would be a slippage of 6 or more arc seconds). Much more than the errors shown on your guide graph.
That points to a sudden slippage instead of a slow progressive slippage of the guide scope. Small, on the order of a few arc seconds, but that is the magnitude scale that we are dealing with).
Be sure to check the dither options (I think they are in the guide scope window) to make sure that you have set a small enough settling distance. If you have set the distance to 6 or more arc seconds, that could explain the 6 arc second type errors that we see -- i.e., you'd started taking an image even before the guiding error has damped down after a dither.
Good luck,
Chen