This is an excellent question, and I don't think Mattmark's answer is irrelevant at all. 'wherever' it is that humans live, our cultural-ecological niche is somewhere in the infinite regressions between the Actual and the Potential. I will return to this in a moment. In answer to the 'undiscovered skills' question, philosopher of technology Jaron Lanier came up with the interesting concept of homoncular flexibility to describe what he calls 'pre-adaptive biology'; that is, all the things we are 'wired' to do without knowing that we can do them. For example, all humans are physiologically capable of swimming, but those who live in an environment without bodies of water or knowledge of other humans who swim may never 'discover' that they can swim. Lanier first realized this while working with Jeremy Bailenson et al at the Stanford Virtual Human Interaction lab http://vhil.stanford.edu/, where 'mistakes' in Avatar environment experiments enabled them to experience skills and forms of embodiment that they didn't think were possible (like being 'turned into' a lobster claw, or a triangle).
In cognitive science, the modularists argue that we have an evolved 'language module', or a 'swimming module', or a 'cooperation module', etc. So from that perspective, we may not have fully-evolved modules for skills we haven't needed to practice yet. In this sense, it is a silly question to ask whether we posses a module for X, X being that-skill-which-no-human-alive-has-thought-about. But I think a cybernetic perspective is more interesting here, and we need to ask about, and experiment with, emergent properties of organisms and humans in different environments. From an object-oriented ontological perspectives, interacting with objects and tools also leads us to imagine and practice more skills - the kind of skills that 'exteriorize' some of our abilities (like storing our memories in written symbols, or increasing our movement abilities with bikes, cars, and planes), or require making objects an extension of our bodies. This is what Heidegger calls a 'readiness-to-hand' relationship with objects: when we drive, we are not consciously aware of how a car becomes an extension of ourselves; we are 'immersed' in or with it, or merged with the car. Archeologist Bjornar Olsen takes this notion farther and argues that 'objects remember'. He goes beyond the second-order phenomenological claim that memory is in the body (and not just in the mind-brain); like a body that 'remembers' how to ride a bike. He would say that here, memory is in the mind, the body and in the bike! (try riding a bike without a bike, he jokes, and you are left with the sound of one hand clapping).
All this to say that our potential for new skills is not just in our brains (and certainly not just, as Lanier suggest, in unknown, but a priori cognitive modules), but also in the tools, objects, and world around us. The Inka had toys with wheels, but they hadn't "yet" invented the chariot. Humans everywhere built sails and kites before they figured out flying machines.
Re: spoken language: inbuilt module or not, it is not something we are born with. Spoken language is a particular expression of our symbolic capacities, just like writing or painting, but it is a technology that is exterior to humans; the signs and symbols of spoken languages are stored in the memories of living humans (later tablets, papyri, books, and hard drives), but they die with humans, tablets, and hard disks! There are gesturalist theories (Michael Corballis) of the origin of language through hand-gestures ('sign-language is a pleonasm, since all language is the exchange of signs). Spoken language probably spread organically in some places, and diffusionally in others (like bows, arrows, and agriculture), and it is more than probably that some bands of paleo-luddites would have rejected it for a while (like people who intitally rejected telephones or facebook). Speaking with hand-gestures makes more sense in a predator-prey environment, in fact, and may be responsible for honing our symbolic skills for a very long time.
Now back to the question of 'potential'. As symbolic creatures, we mostly interact with what is not actually the case on the one hand, and what could or couldn't be the case on the other. When we make sense of the world through languages and mental categories, we do this through symbols, and not the thing itself ( a name is not the thing named; the map is not the territory, etc.). We also project our minds (or Theory of Mind) into anyone and anything, all the time. Consider our capacity for agent-detection, which is basically an extrapolated Theory of Mind. We cry in movies (by projecting our feelings into non-existent agents), we see faces in the clouds, 'detect' predators in the shadows, and believe in all kinds of supernatural agents. Cognitive anthropologist Pascal Boyer argues that we cannot not believe in and interact with 'supernatural agents', because we are always projecting our ToM, thereby creating imagined agents that are not actually there. I would argue that as humans, we are potential agents a lot more than actual ones.
So in terms of new skills, the sky is limit. My only concern is that the things we have built remember too much and do too much of the business of imagining at this point. Start with Plato's Phaedrus (how writing techology makes us forget), and read Tim Ingold on the history of lines and Michel Serres on the history of writing and memory to get a good sense how much potential we are also losing each time we outsource ourselves to things. Tim Ingold, unlike most other anthropologists, argues that we should think of knowledge and culture as skill: it is incorrect to say that 'culture is passed on from generation to generation', he proposes; instead, we should say that 'skills are regrown in each generation' through experimentation in the world. What happens to our ability to merge and experiment with the world when we leave most of it to objects? Consider language once more. You could ask if, at one point, we will develop a module for telepathic communication. In a sense, we already have. With non-spoken sign-language, our ancestors were already silently planting thoughts into each other's brains. Now with instant-messages, we also share symbols through entirely soundless communication. We do it for longer and we reach farther than our ancestors in that proverbial savanna, but our attention spans are much thinner, and we also rely on a chain of technology that none of us fully understands or could recreate on our own. In this sense, we have devolved, and our future generations will have less and less skills.