There are three classes of information representation of skills and actions. The first is the representation of skills in DNA. The second is the representation of skills in the brain. The third is the application of a skill resulting in action.
For each species, there is a limit on the information that can be encoded in the DNA. Of that information, only a proportion is used for encoding innate skills. In theory we should be able to calculate how many bits are available to pack in all the innate skills for a given individual of the species. Even through the content may differ slightly across the population, the total available is probably fairly constant.
Counting the number of innate skills for any given species is an unsolved problem. The boundary between 'built-in' functions and innate skills is not clearly defined, so estimating the overall count of skills for a given species is difficult. It seems likely that species with a larger overall DNA also allocate more space to innate skills. On the other hand, some species with a longer pre-adult maturation period may have a smaller percentage of the DNA allocated to innate skills, and depend more on skills learned during the maturation period.
I would hypothesize that there is a trade-off between the total number of skills and the average complexity of skills.
I conjecture that the brain representation is equivalent to the DNA represention. The DNA representation may be slightly more compressed, as a trade-off to make the brain representation easier to apply to generate action. Skills are captured in instruction manuals and in textbooks. I hypothesize that there are three main components. The first deals with skeletal movements, and more specifically, with joint rotations. The second one deals with geometry, to integrate vision and to make the next motion dependent on the current position of the individual. The third deals with time. Motions of the joints have to be closely coordinated in time. I hypothesize an 'inner language' representation with an 'inner geometry' to choreograph and coordinate the capability for skilled action.
Action is captured in movies. Camera recordings capture the action in images, frame by frame, in a sequence over time. Pixelated 2D images correspond resonable well to retinal images. However, to support imitation and the internal geometry, I conjecture a stick-figure (vector graphics) representation.