I start with a basic raw conversion, just demosaic, white balance and colorspace defaults courtesy of libraw. Then:
- gamma 2.2 to stretch out the tones from the linear recording;
- black/white point to fill the histogram;
- tone curve for any other manipulations.
- rotate if straightening is needed
Save that as a TIFF as the basic image. From there, any of:
- saturation, not much, but sometimes I overapply for effect.
- crop, if the mood suits. This might produce a few versions, for different compositions.
- grayscale. Been into black and white lately; I’ll sometimes use the channel mixer to darken skies or bring out a particular color.
- for some websites, resize to 640xAspect, then apply the smallest sharpen to introduce acuity
These usually are saved as separate JPEG images, suitably named so I can keep track of them. My software stores the processing commands in the exif, so I can reconstitute any of the PP for a particular image.
If I need to do other correction or messing around, I’ll throw the TIFF or JPEG into GIMP.
I’m debating throwing in an input sharpening step, but it doesn’t seem to make enough difference to warrant the trouble. The output sharpening step for web resize makes a significant difference, however; reduction resizing loses resolution, and just the first sharpening step makes the reduced image look a lot crisper, for lack of a more technical term. Makes my D7000 look more like a D810… 
I may write a resize-sharpen script, but that’s about all the “automation” I’ll consider. I’m not into presets; I want to start as close to what the camera measured and go from there.