Kant's project is often misunderstood, especially by contemporary readers. We tend to think of the Grundlegung as a rival (for example) to J. S. Mill's Utilitarianism or On Liberty. These books -- and there are many others -- do provide us with primary and secondary principles for distinguishing right from wrong conduct. (For an example of a contemporary Kantian who does this, see Alan Donagan's The Theory of Morality (1977).)
While Kant believes that the categorical imperative in its various incarnations does indeed determine whether a certain kind of action is morally permissible, his central focus lies elsewhere. Because moral controversies are so widespread and deep in our own day, however, we tend to read Kant (or Aristotle) for solutions to pressing contemporary moral problems from abortion to genetic engineering to unilateral nuclear disarmament, and are disappointed when we don't find them.
Kant isn't particularly concerned to resolve pressing moral controversies because he thinks the moral judgments of ordinary, reflective people are largely correct -- that there is a body of "ordinary rational knowledge" which everyone possesses. Everyone, for instance, "knows" that we shouldn't make lying promises, or turn our backs on those in distress, commit suicide, or waste our talents. Kant may have been mistaken about this, but his belief that reflective, sincere people know right from wrong means that he is less concerned to figure out what is morally justified than to determine how to ground these judgments.
Perhaps an analogy will make this point clearer. Everyone knows that hummingbirds fly, but what laws of flight enable them to do so? This is a different inquiry than one that would be undertaken if there was a scientific controversy whether hummingbirds fly -- or simply float on air currents. Someone who says "Given that hummingbirds fly, how is this possible?" would seem uninformative -- even question-begging -- if the question were "Do hummingbirds fly?" If, however, humingbirds do indeed fly, then the laws of flight should explain how. This analogy, of course, cannot be pressed very hard: Kantian moral laws describe the way the world ought to be, not how it is. Still, the anaology suggests that we shouldn't suppose Kant's primary purpose is to remove concrete moral perplexity so much as provide the grounding or foundations of morality.
We, on the other hand, are often baffled as to what is morally permissible concerning (e.g.) abortion, allocation of scarce life saving resources, physician assisted suicide, genetic screening, affirmative action, surrogate motherhood, and many, many other moral matters. (Notice, by the way, how many contemporary moral problems implicate technology in ways that would be unthinkable in the late eighteenth century, when Kant writes.)
So much by way of what Kant isn't trying to do, with a hint regarding what he hopes to accomplish. What more can be said about his aims in the Grundlegung? First, as he makes clear in the Preface , Kant thinks there is a need for "pure" ethics; i.e., ethics free from any empirical element. (See Comment). This "pure" or a priori element will be wholly rational, uncontaminated by anything empirical. But second, Kant makes explicit that what he aims to provide will not be a complete metaphysics of morals. That, he says, he will reserve for a later time. (In fact, his Metaphysics of Morals doesn't even attempt to do what he says here he will do; but that is another story.) Here he resolves to discover and establish the foundation or fundamental grounding of the metaphysics of morals -- the supreme principle of morality. Or, as T. K. Abbott translates Kant: "The present treatise is . . . nothing more than the investigation and establishment of the supreme principle of morality."
To employ another analogy, imagine someone writing a short book entitled "The Foundations of Formal Logic" in which the author attempts to establish the indispensability of the principle of Non-Contradiction for rational thought. Even if successful, the articulation and defense of the principle would not tell us everything about formal logic; e.g., about other important valid principles of reasoning such as modus tollens or the law of the undistributed middle. Nonetheless, if successuful, something significant would have been established, to be augmented by a more complete treatise on formal logic. Yet even if supplemented by other logical principles, a logical system cannot -- without interpretation -- tell us whether a particular concrete argument is either valid or sound.
Like "word problems" in mathematics where we need concrete background knowledge to know that a dime is twice the value of a nickle, so too in logic. And a complete metaphysics of morals would need concrete background knowledge before it could be appropriately applied. Suppose, for instance, we hold that it even in war time it is wrong deliberately to target the innocent in order to save lives. Are munitions workers "innocent?" Their spouses? What about those who make munitions factories? Those who supply the building materials? Those who provide them with health care? Further, what does it mean to be "innocent"? A rabid dog is "innocent" in one clear sense yet an "aggressor" (and hence not innocent) in another sense, as it attacks.
This is not a frivilous example. Those who regard typical abortions as "murder" often appeal to the "innocence" of the fetus. (Or is it an "unborn child"? Does it matter? Why?) The point is simply that no general principle or precept can be either self-interpreting or self-applying, so we cannot escape the responsibility of interpreting and applying principles in logic, mathematics, or morality.
Limited though Kant's aim is, he believes that it must be made out if morality isn't to prove chimerical. David Hume had said that morality is "more felt than judged" -- that there may be intersubjective agreement, but there morality cannot be objective. Kant thinks that morality requires greater authority than mere agreement -- however widespread -- if we are to explain how morality necessitates the will.