In Genetics and Reductionism, by Sahotra Sarkar, page 4, he considers what it means for a feature of a person to be genetic. He says that although many people assume it's simple and they understand it, actually it's a tricky problem and presents several unsatisfactory answers and explains their flaws. He doesn't offer the correct answer.
Here's the answer: A feature is genetic if the knowledge for the feature is the person's genes.
This is an example of the general purpose value of epistemology: epistemolgy is needed for solving problems in many other fields.
Let's consider an example. Jack is born and grows up to be a seven foot tall basketball superstar. Is his height genetic? Is his basketball playing genetic?
On both of these questions, some would get confused. They would say the basketball playing is a caused by a "gene-meme interaction" (or "gene-environment interaction"). That is true. Basketball coaches, and parents, are more encouraging to tall children. His cultural environment responds to tallness. And although many would overlook it, becoming tall is not purely genetic but requires environmental help, for example Jack must be fed regularly. The genes alone cannot make Jack tall without the meals.
Some would therefore conclude that both height and basketball skill, and pretty much everything else, are partially genetic.
But let's look at the knowledge. There is no knowledge about basketball in the genes. There is only knowledge about it in coaches, parents, other kids ... in Jack's culture. So basketball playing is not genetic, full stop. Note that this conclusion matches common sense, and also that it can give meaningful answers instead of just vaguely saying everything has multiple causes.
As to height, his parents don't know anything about making Jack tall, but his genes do have knowledge about how to construct a tall person. So height is genetic. Again the knowledge-based definition gets the common sense answer.
 There are some humans ideas about making people tall. You can stretch them. You can feed them "health food". You can bless them. To the extent the parents do those things — and they work — then the height is partially caused by knowledge in the parents.
Talking about genes 'knowing' this or that seems to be a very strained metaphor. Does a circle-shaped template 'know' how to draw a circle? Does a recipe for a cake 'know' how to make a cake?
In all of these cases one wants to say that somehow the information content of the end-product was already contained (albeit perhaps encoded) in the 'recipe'. (Is that basically what you mean by saying that the 'knowledge' of the feature is in the genes?)
But a recipe can be followed incorrectly, or under abnormal circumstances, and may yield something other than the 'intended' result, so perhaps some of the information characterising the end-product is lost unless one takes into account 'the typical environment'? But what exactly is that? And couldn't it be that an 'unintended' result happens often enough to itself be affected by natural selection? Hopefully you can see that these questions become pretty complicated pretty quickly.
I don't think that the metaphor of 'knowledge' helps very much here.
A gene knows X = a gene contains knowledge of X.
Templates and recipes do contain knowledge too, yes.
I do not see any problem here. Whether a person can use something has no relevance to whether it has knowledge. Knowledge isn't people centric.
Your criticism sounds to me similar to doubting books contain knowledge because they can be misread.
Have you read _FoR_, _OK_ and _C&R_?