I would agree...up to a point. This would be a good course of action if MG didn't exist, and/or if dealing with a very limited build overall. However, given that we DO have this site as both a reference and a testbuild sandbox, there needs to be some prefatory points to that list...
1) Make your ModularGrid build bigger than you know it should be. This can then be pared down in size over time prior to making any move toward purchasing physical devices. Once you have that under control, then start considering cases, preferably ones that offer room to expand from your build's basic elements...because you're going to want to do that, eventually.
2) Two VCOs are always better, even if one is a simpler device and the other is more complex. The ability to detune, mix different waveforms, have one modulated and the other not, and the like results in a far richer palette of sounds. This doesn't necessarily require a separate mixer, also, as there are VCFs that have dual audio inputs available.
3) Some VCFs are not necessarily the ones you would start with. Your best bet is a state-variable VCF, which allows for low, high, and bandpass (and sometimes notching) in the same module, sometimes with separate outputs for each available simultaneously. Keep in mind that the VCF is the "timbral heart" of any synth, and that you will be playing it in of itself just as much as any other controller-connected device such as VCOs, etc.
4) When looking for envelope generators, there are actually two basic types: ADSR and AR (or AD, depending on how they treat incoming gates). Many of the latter have the ability to loop their envelope settings, also. It's also important that there are at least two EGs in a synth...one for applying to the filter for timbral modulation, and another for use with the main audio VCA for dynamic modulation. Technically, you can use one for both, but the results are far better with two separate ones under the same gate/trigger control.
5) VCAs: there are two types of these. Linear VCAs tend to be more useful for modulating the levels of CVs, modulating signals, and so on, and these also tend to be DC-coupled so that they can pass sub-audio signals. Exponential VCAs, however, are what's needed for best performance with audio, since human hearing responds to changes in apparent loudness as an exponential curve.
6) Modulation circuits come in more flavors than just simple LFOs. One can use looping AD envelope generators, which allows the user to define the rise and fall rates of the modulation curve. Then there are voltage-controlled slope generators, which are like those but which allow CV control over the rise and fall rates; these circuits are the backbone of such Eurorack staples as the Maths, for example. LFOs can also have delays, which allow the user to define the time needed for the LFO to reach full amplitude; this is useful to allow variable vibrato and/or tremolo effects depending on note length. This is also a situation where more than one is much better than just one, hence multiple-circuit modules such as the Maths, Batumi, et al.
7) Multiples are and AREN'T necessary...it depends on the build size. If you're only dealing with sending your pitch CV to two or three VCOs, you won't need a buffered mult. Those are useful when you have several VCOs (or other CV destinations) for the same CV and you'll need to regenerate that CV to prevent tuning issues due to voltage sags from the exponential converters in those modules. Also, if you have a small build and need every last hp for primary functionality, consider using inline multiples, stackcables, and so on instead of dedicated mult modules. But you'll invariably find that your build needs at least one multiple module, preferably something with at least a pair of 1 - 3 mults. And remember, you can't use multiples as mixers without a certain level of risk to any module that doesn't have diode protection on the output; mults are only designed to split outgoing signals, not to combine incoming ones.
8) You will need something at the end of your audio chain to attenuate your signal level. Synthesizer levels often exceed 5V peak-to-peak, and this is usually too hot for the front ends of many mixers. Either add a passive attenuator at that final point, or an output module that's designed for stepping the level down to proper 1.2 or .775V p-2-p line levels.
9) Colored patch cables are...pretty. And that's about it. As long as YOU know what you're doing in YOUR patch, it doesn't matter what color the patchcables are. This really only applies to systems such as Buchlas, where you have separate paths for audio and control signals...and in those cases, the cables are terminated differently anyway. Buy plenty of cables that you can afford, but don't splurge on decorative notions.
This is all based on some 40-ish years of experience of poking around with these things, across many years of their development...where some things have definitely changed, but many of these basics still remain the same.