I'm not sure how closely the concepts I'm familiar with might match what you're aiming for, but I do have some familiarity with the development of Proof Theory, and your search for terms seems to line up with some ideas we've explored in that field.
In proof theory, particularly in discussions around Natural Deduction, we sometimes talk about a proof or argument being in Normal Form. A Normal Form argument is one that has been written in "the most basic way", which is to say we've formally looked at all and only the necessary premises of the argument, broken them down into component syntactic parts (via "elimination rules"), then reassembled them to structure the desired conclusions (via "introduction rules").
Not all formal arguments, or even all validly constructed formal natural deduction proofs, are in Normal Form. However, many formal systems aim to show something like a Normalization Theorem, to the effect that when any non-minimal use of our logical rules is invoked, we could without loss of generality rewrite the argument to eliminate it. One of the main proponents of this kind of work was Dag Prawitz, whose thesis on the Proof-theoretic analysis of natural deduction helped inform a lot of the philosophical writing around proof, inference and computation that would follow.
A valuable concept Prawitz introduces in his work is the notion of an "Argument Skeleton". (see his On the Idea of a General Proof Theory for a more accessible overview). This is a generalization of the tree structures involved in formal natural deduction arguments or proofs, in that we allow not only that we are working from logical axioms as premises to conclusions (which we call a Closed argument), but also that we can allow unproved antecedents that lead to consequents via the same kind of logical rules of inference - these "open argument" structures are also Argument Skeletons.
(Natural deduction often tries to do without Axioms altogether in its structures, rather deferring everything that is "purely logical" to the application of structural inference rules.)
So perhaps some useful turns of phrase might be these: your "weaker" formal arguments are Open Arguments, and their "proofs" are Argument Skeletons, since they hint at a structure of proof that could potentially be further developed. Your "stronger" arguments are Closed Arguments, in that their skeletons do not leave extra-logical assumptions dangling, and the most syntactically minimal version of such an argument (ideally suited for machine processing) would be its Normal Form.
There are alternative interpretations of this kind of work in other forms of Proof theory. Where Prawitz makes use of Argument Skeletons to support his Natural Deduction system, the more common Sequent Calculus technology developed from Hilbert's system by Gerhard Gentzen allows us to capture transformation rules for inferences, collapsing the distinction between open and closed arguments. However, understanding that distinction can help in getting a grasp for what the Sequent Calculus is doing differently, and how we can put the principles of consistency and soundness-preserving argument transformations to use in mechanically manipulating proof strings.