The scenario that prompted this approach was that I was trying to reproduce a performance issue with verbose tracing enabled but the traces had such an impact on the system I couldn't generate the necessary load. I needed a focused way to generate load on a particular Service and I didn't want to have to jump through all the hoops necessary to get a Test Harness deployed to the Performance Test Environment.

A more common scenario might be that you're struggling to diagnose what seems to be an environment-specific issue (on an off-line Process Server) and you want to get some visibility into what a specific Service is returning.

Anyway, if you follow these instructions you can call a Service using client-side Javascript...

The bulk of the Javascript is a (only slightly) modified copy of an out-the-box function and the last few lines are the ones that specify what Service to call and which Parameters to pass. You'll need to adapt this to suit your requirements...