the cluvfy is either executed by the grid control agent as you expected, but with 188.8.131.52. the grid infrastructure is also executing cluvfy commands.
(See crsctl stat res -t and you will find the cluster resource ora.cvu, which is the clufvy tool called by the grid infrastructure).
I have done the same and get the run around from the support tech when I ask for clarification. Even asked duty manager for assistance and was sent right back to the support tech that I could not get conclusive answers to my question from the first time. Seems he is telling me that he opened a new bug up for my answer but the description of the problem is not the same so how can I believe him. He wants me to apply patch 12 but I thought the later patches set are supposed to contain previous patches, so why not apply patch 14 or newest. Seems none of the patches show that they solve the Windows Heap Issue for 184.108.40.206. Even the bug note says it will be fixed in 220.127.116.11. Just one more reason we are planning (although against my better judgement) to move away from oracle.