As a hobby (and for my personal edification), it seems like a fun challenge to transcribe a few Baroque music manuscripts into a more readable form, and maybe post the results to imslp.org or something. Is it okay to change the old Dorian key signatures for minor-key pieces into their modern equivalents, or is that generally frowned upon? Same question for old C and F-clefs. I can imagine today's violinists, for example, not really being too keen on reading the soprano C-clef. But in the interest of historical accuracy it would perhaps be better to retain it anyway?
Typesetting baroque music from primary sources is also something I do, and I can give the decisions that I've made on these questions. However, it is ultimately up to you.
As Laurence Payne says, it does depend on your purpose. In my case I am doing it partly for my own use and partly for general use (e.g. for IMSLP). Ideally I would like to play from the primary sources themselves, but sometimes they are too difficult to read, or they in clefs that are no-longer common and the people I play with would prefer modern clefs.
I use Lilypond to typeset and have set up a system so that I can generate parts (and scores) in both modern and original clefs. Having the original clefs is something that I use to help me gain fluency in reading these clefs. The original-clef versions are also easier to proofread against the original. Modern clefs are obviously more widely readable. I also generate parts in alternate clefs - for example, French music often has one violin part and three parts for different sized violas, and since there are more violin players available, I generate an additional copy of the highest viola part in treble clef so that it can be played by "second violins".
As far as key signatures goes, I almost always use the original. This is partly so that I have something closer to the original, but partly because accidental conventions of the time were not the same as the modern convention (retaining accidentals to the end of the bar). A common convention in baroque sources is that the accidental continues to apply if the next note is the same (even if the next note is in a new bar), but will not apply later in the bar if there are any intervening notes. Often it is not clear what convention is being used, and there's also the issue of errors in the sources. Because of this, it's not always obvious whether the note was intended to be altered or not. If you wish, you can become an editor and decide, but I prefer to let the performers make this sort of decision, and to do that they need to know what the original says. By trying to modernise the notation I quickly found I was making decisions for the performer.
You haven't asked about time signatures. I will generally also try to retain these where possible - a time signature of just "3" is very common, and you also see mensural time signatures.
There are other decisions that you'll have to make too: repeat marks follow different conventions and are often ambiguous; slurs can be somewhat haphazardly positioned; dotted notes may have their dot placed in the next bar; bar lines might be omitted where there is a hemiola. However, I have found transcribing from original sources to be a very rewarding hobby.
That's exactly what I use to do even when I only read a piece: Change the clefs and also the modes. This is probably not correct in a puristic historical sense but as long as you know what you're doing everything is allowed if it is fun, challenge or more comfortable to you.
The result will be: You learn reading the historical clefs and modes by this practice and you are really aware of this music inside out.