We are entering an exciting time in structural biology where artificial intelligence can be used to predict protein structures with greater accuracy than ever before. Extending this level of accuracy to the predictions of disulfide-rich peptide structures is likely to be more challenging, at least in the short term, given the tight packing of cysteine residues and the numerous ways that the disulfide-bonds can potentially be linked. In many cases it has been shown that several disulfide bond connectivities can be accommodated by a single set of NMR-derived structural data without significant violations. Disulfide-rich peptides are prevalent throughout nature, but arguably the most well-known are those present in venoms from creatures such as cone snails. Here we have determined the first three-dimensional structure of a U superfamily cone snail venom peptide (TxVIIB), and despite AlphaFold predicting the peptide to adopt a mini-granulin fold with a granulin disulfide connectivity, our experimental studies with NMR spectroscopy and orthogonal protection of cysteine residues indicate it adopts a mini-granulin fold with the inhibitor cystine knot (ICK) connectivity. TxVIIB contains the VI/VII cysteine framework, which is most often associated with the ICK fold, but our results point to the mini-granulin fold being more wide-spread than originally thought. Our findings also provide structural insight into the underlying features that govern formation of the mini-granulin fold rather than the ICK fold and will provide fundamental information for prediction algorithms, as the subtle complexity of disulfide isomers appears lost in the current prediction algorithms.