I came across a tiny bug in scoring system.
The scenario is 'Ligand Debut' which is to design a protein to bind onto ligand.
I messed around with it, the target protein self folded and formed its own hydrogen bond.
The bond was strong enough that the score has overqualified the scenario even the ligand is far away and not getting any bond.
My suggestion on the programming side is to separate the scoring system rather than relying purely on folding of all protein structures indiscriminately.
The program is amazing, I am still learning my way, very fun to play with.
Thanks for all the efforts !