The total free energy of some solutions to EternaFold puzzles seems to be about 0.2 to 0.3 kcal higher than the threshold normally needed for the shape to fold at all. See my comment on [EFOLD]Break Test for further explanation and an example. I’ve observed that each such solution includes a CG-UA quad, whereas sequences with similar total free energy but no CG-UA quad misfold to the unfolded shape as expected. This leads me to wonder if this particular quad is being assigned a wrong energy value in the relevant calculation, even though the puzzle on the screen shows the correct value of -1.