Hi Byron,
AFAIK, two choices for the server side comes to mind: writing a CGI program and use it together with Apache or another web server, or using Harbour's Httpd and build the whole server by yourself.
Depending on the requirements, you should use one of another. Using CGI is easier, you don't have to deal with the issues related to the http protocol that is up to the webserver, and you just write the specific code to produce the JSON response.
On the other hand, Harbour's httpd may be a little bit more complex but it will be faster, and you can have full control of the whole thing.
I have no examples for JSON in Harbour, i've used it only in php apps.
From the client's POV, you can 'consume' any kind of services including XML, JSON or whatever, using the TIP library from Harbour's project, or using libCurl. I'm currently using libCurl with great success, connecting to AEAT services (the spanish IRS).
JSON processing should be easy, there are function already written in the Harbour Project to deal with it.
May be you can tell a little bit more about your project so having a closer idea may help to be more specific in the answer.