In order to run in client mode you need the address of someone who is set up to run as a server. The address can be in the form of a DNS name (e.g. server.domain.com) or as an IP address (e.g. 222.222.111.111). You will also need to know the port to use (if they're not using the default port). Then choose a blocksize (I'd recommend sticking to the standard size but if you know what type of errors you've got then a different blocksize could well be faster).
Once started, the client will connect through to the server on the given port and start the repair process. You will be kept informed of what is being done by both a progress bar and status information. If you wish to stop at any point then just click on the STOP button.
There's a special case when passive mode is enabled. Here, after connecting to the server, the client switches roles to act as the source for the repair. This allows you to repair files even if you are unable to receive a remote connection (e.g. behind a firewall or on a NATed address).