My two cents:
I sketch at the eyepiece using 0.7 mm HB lead in a mechanical pencil equipped with an eraser. My notebooks are mostly Canson artist series 5.5 x 8.5 inch sketch books. I recently switched to Blick studio sketchbooks of the same size only because they're available locally. If I'm sketching a large object, I'll use 11 x 14 inch tracing paper backed with a sheet of white paper the same size and tape them to a clipboard. Pencil on tracing paper is great for blending graphite with fingertips, by the way. This is all I use at the eyepiece as far as sketching equipment goes, although it's supplemented by my dimmable red light - not to mention my scope and related accessories...
All my sketches are scanned because no matter how I well protect the pages in my observing notebooks, paper gradually degrades. Scanning is also a way to preserve my sketches and notes in the event something happened to them, like fire, flood or theft. Improbable, but you never know. My observing notebooks are by far my most precious physical possession, so I got into the whole scanning thing from an over-abundance of caution.
The definition of sketch and drawing by Ivan is right on from my perspective. I'll add that my sketches done at the eyepiece always look more like a finished drawing when viewed under the dim red light of my flashlight, and seem a much closer match to what I saw in the eyepiece compared to the same sketch in the light of day. It's a bit frustrating that an almost fully dark adapted eye in dim red light works so differently that a fully daylight adapted one when it comes to drawing, but it's the reason I put so much effort into making a finished drawing of the objects that interest me the most. The whole idea is to represent what I actually saw as realistically as possible. Realism is difficult though, and I have yet to produce a sketch or finished drawing that's indistinguishable from my views in the eyepiece.
Even so, a few drawings have come pretty close and I keep trying to improve. But at the end of the night, there's a limit to what I can achieve with pencil and paper, and the only way I can make portions of my finished drawings look more like what I saw in the eyepiece is to apply image processing. Round stars are nice, but the biggest thing is to correct the contrast back to the original drawing - scanning or photography never fully captures the dynamic range of pencil on paper, at least in my experience. Also, in a few cases I've added color. Again, the guiding principle is to produce a finished drawing that shows only what I actually saw in the eyepiece.
Even so, sometimes it's difficult to be sure if you've seen something as Norme discusses above, and I'll add that sometimes the photographic biases we bring to the eyepiece are surprisingly hard to shake. My rule is to not add anything to a sketch or finished drawing I'm not 100% certain that I've seen in the eyepiece. For threshold details or objects, my test is to try to see the same thing in a different part of the fov, or in a different fov altogether. I often use photos at the eyepiece to help me locate precisely where these threshold details/objects are, but it takes an honest, disciplined effort to be sure of what I've seen and haven't seen. Sometimes it's difficult to be certain either way, and that counts as a non-observation in my notebook. Something to try for again on another, hopefully better night.
All that said, I get the temptation to add details you didn't see that show up so well in photos - they can greatly enhance the cool-factor of your sketch or drawing. But yuck! You might as well sketch from a photograph and not even bother with a telescope. Which is fine, and is great practice, but in either case you need to point out what you've done. Likewise if your sketch was based on what you saw on a EAA monitor. Just basic honesty.
I think everyone here wants to produce honest sketches and drawings of what they've seen, and every approach has merits because we all have different skills and interests. A great sketch is just as great no matter how it's produced as far as I'm concerned. For the objects that tempt you to add details you haven't seen, I suggest coming back to them night after night, and gradually add details you didn't see earlier to your original sketch. I'm still surprised at how much more I end up seeing by observing this way.