In reading the literature on Optimality Theory, e.g. Kager (1999), every reference to well-formedness constraints refers to prosodic structure, especially syllable structure. It seems to me that well-formedness could also refer to segmental structure, e.g. phonotactic or co-occurrence restrictions, e.g *tl or *VN. Do people use well-formedness constraints to apply to structures other than prosodic structure? I suppose the Obligatory Contour Principle might be one case, especially when extended to feature co-occurrence. Also, isn't well-formedness and markedness the same thing?
The OT construct "well-formedness" is indeed the same as "markedness", which is to say, any constraint penalizing a particular arrangement of phonological units. (Recall that constraints fall into two taxa, "markedness" and "faithfulness" i.e. Correspondence -- I am leaving out Prec to simplify life, since not everybody has signed on to OT-CC). The OT literature has plenty of markedness constraints that don't refer to any prosodic entities, such as the myriad OCP constraints as well as general sequencing constraints like *NC, *VN etc. Plus... a huge number of segment-markedness constraints. In fact, I conjecture that one can find more posited markedness constraints in the literature that make no mention of prosodic objects than one can find of constraints that do mention them. The crucial theoretical connection between prosody and markedness / faithfulness is that, ex hypothesi, prosodic structure is not present in inputs (due apologies to Chris Golston, I'm just reporting the belief), so IO faithfulness constraints referring to input prosody wouldn't exist, except insofar as "prosody" is a kind of loose term that could also refer to tone and moraic structure which are present in underlying forms.