This is likely perfectly fine. The GPLv2 calls this case “mere aggregation”: you can distribute a GPL-covered program alongside a proprietary program without issues, provided you comply with the GPL for the GPL-covered program.
It is not always clear whether two executables form two separate programs, or whether they are so intertwined that they effectively form a single program. The GPL FAQ suggests that two programs are separate when they only communicate “at arms length”. Since your programs communicate via simple command line arguments and JSON documents, there's a very good chance that they are indeed separate.
Personally, I think a good test is to consider the actual goals of the GPL: software freedom. Is an end user actually able to inspect and modify the GPL-covered program, and your proprietary system will be able to use modified versions? If so, very good.
Relevant entries from the GPL FAQ: