When you send a command to the agent from the console this goes over port 4155 by default, but this is configurable within the General Settings of your agent policy as well.
You might also want to make sure that the agent appears to be working on its own, and that it has successfully registered with the Protect console system.