CCPv1: Log out agents automatically when they close their CCP

Important

This topic only applies to customers who use CCPv1. The URL for CCPv1 ends with /ccp#.

When using the default Amazon Connect CCPv1, closing the CCP window doesn't change an agent's status from Available to Offline. An agent must change their status manually to Offline and then log out.

To change this behavior, you can do one of the following:

• Use CCPv2. For instructions on upgrading to CCPv2, see My CCP URL ends with /ccp#.

• Create a custom CCP. See Amazon Connect Streams API and the Agent API

• Use the following steps in this topic to update your CCP so it switches agents to Offline and logs out agents automatically when they close the CCP window.

Step 1: Set up the Streams API

For instructions, see the Amazon Connect Streams Documentation.

Step 2: Update your application code to change the agent state

Integrate the following Streams API calls into your web application:

1. Use connect.agent() to subscribe to agent events and retrieve agent objects.

   let mAgent;
   
   connect.agent(function(agent) {
     mAgent = agent;
   });    

2. Call agent.setState() in the onbeforeunload event handler to change the agent state.

   window.addEventListener("beforeunload", function(event) {
       if (mAgent != null) {
           let states = mAgent.getAgentStates();
           // "states" is an array of changeable states. You can filter the desired state to change by name.
           let offlineState = states.filter(state => state.name === "Offline")[0];
                         
           // Change agent state
           mAgent.setState(offlineState, {
           success: function() {
               console.log("SetState succeeded");
           },
           failure: function() {
           console.log("SetState failed");
           }
       });
     }
   });      

Step 3: Design for errors

If an API call fails to execute the first time and a contact takes the error branch of your contact flow, there's a chance that an agent's state won't change as expected. Be sure to include logic to account for this possibility. For example, you could delay the page unload while the API call is tried again. Or, you could pop a "Call failed" warning message in a modal dialog before the page unload.