However, to be more precise, my question was not whether the reviewer was perfectly correct in all respects (where I can see for myself that she's not), but whether the substance of her criticisms was sound based on having read Caplan (which I can't do unless I read Caplan, as you have).

You seem to have wilfully ignored the main point I made, which is this:

This implies that all non-heritable variance could be due to parenting effects, when in fact unshared environment is known be accountable for most of the remaining variance. That eliminates the review's main argument against the book.

See, she is arguing against the twin/adoption studies based on the idea that the remaining environmental influence could be parenting, when Caplan explicitly points out that this isn't the case. Unshared environment explains remaining variance. That was her sole substantive point (the rest being obviously false ad-homs and claims about some unspecified "other studies").

To restate my question: What content in Caplan's book can be mustered in response to a counter argument claiming that that heritability studies are weak for giving precise behavioral advice when clear thinking (and maybe even other studies) about the specific details of the advice show that some details matter?

Other studies cannot show that a parenting style detail matters without controlling for heritability. People frequently see studies that say (for example) kids who eat junk food are fat as adults, and assume this is due to childhood environment (controlled by parents). In fact, twin studies have shown that heritability for bmi is ~.80, with the remaining variance due mostly to unshared environment.

Now, if you want to dismiss the vast literature twin/adoption studies out of hand due to "confounds" (you haven't specified), then, go ahead I guess. But you should have very strong reasons, since twin and adoption studies are considered the gold standard by the relevant social scientists.

Does Caplan talk about details, or just twin studies?

Yes. It's a fairly long books, so he discusses may specific examples.

I should also address this, I guess:

Maybe he could also report that it doesn't help to teach your children to meditate or program computers or read books for pleasure. Maybe he could claim that it doesn't matter if they have violin lessons versus sitting in their bedroom listening to popular music that glorifies violence and the degradation of women. Maybe they shouldn't be encouraged to play sports and you shouldn't waste time ferrying them to soccer games when they're 11. Maybe it doesn't matter if you homeschool, versus sending kids to a school with metal detectors and class sizes of 70, versus sending them to a private Montessori school where they'll start being able to start building a social network that enables them to navigate the subcultures of the rich and powerful when they are in their 40's.

I think Caplan would say, you should do those things if you would consider it worth your time even if it doesn't contribute to adult success for your kids (it probably won't). If you enjoy classical music, there is a good chance you kid will as well. He's mostly against dragging your kid to "enriching" things or stressing out because your kid likes rap.