Correct: nothing more is needed after installing grpc, really.
I've grabbed my Macbook to run some tests, and yes, I can see something's broken. It's not tied to running the protoc command however, it's tied to try to compile a check. There's two ways the protobuf 3.0.0 message can appear on make verify-install: either the protoc command doesn't return something that looks like 3.0.0, or it failed to compile a test software. You can see that by doing make run_dep_checks. I'll spend some time troubleshooting that one, but it looks like it's tied to my usage of /dev/null in the tests. Now that I think about it, I've seen that before, but I didn't think it was causing the whole test to fail however. The solution might be as simple as replacing /dev/null by `mktemp` or something. It's currently preventing grpc from using the system openssl and protobuf, spending useless CPU cycles to compile them from the third_party directory. It's not critical, but it's probably annoying enough.
I've also run the tests there, and I can confirm that there's currently a problem with the mac port of grpc. Some of the tests are failing, and I don't link the output I am seeing; it's looking like there's some kind of overflow. But that's just me glancing at the error output. But this is definitely something to investigate. I'll file a proper issue in our github project.
As for the hello world example, after some poking around, I was able to reproduce what you're experiencing. I need to investigate more, but a quick workaround I've come up with is to delete the dynamic versions of the libraries. Basically, delete the *.dylib versions of grpc, grpc++ and gpr from your /usr/local/lib directory. What I'm saying is that I managed to run the helloworld example under mac while linking with grpc statically instead of dynamically. Which is going to be a fun one to debug.
Thanks for your reports and efforts! We definitely need to make the MacOS experience better.