Not receiving `incoming_typing_indicator` push

[orinciog]

Hy!

Using Agent RTM I want to receive incoming_typing_indicator push.
I tried using both js libraries:

 "@livechat/chat-sdk": "^0.2.0",
 "@livechat/lc-sdk-js": "^5.4.0",

but is not working.

Code for @livechat/lc-sdk-js:

const { Agent } = require("@livechat/lc-sdk-js");
const { WebSocket } = require("ws");  

const config = {
   LIVECHAT_ACCESS_TOKEN:"",
   LIVECHAT_ORGANIZATIONID:""
}

async initLivechat(){
    const agentTokenGetter = () => ({
      accessToken: config.LIVECHAT_ACCESS_TOKEN,
      organizationID: config.LIVECHAT_ORGANIZATIONID,
      region: "dal",
      tokenType: "Bearer",
    });
    const agentRTM = new Agent.RTM(WebSocket, agentTokenGetter);

    try {
      await agentRTM.connect();
      await agentRTM.login();

      agentRTM.on("incoming_typing_indicator", (push) => {
        console.log("Incoming typing indicator", push);
      });

      agentRTM.on("incoming_customer", (push) => {
        console.log("incoming_customer ", push);
      });
    } catch (error) {
      throw new Error(`Unable to initialize Agent RTM connection: ${JSON.stringify(error)}`);
    }
  }

initLivechat();

Code for @livechat/chat-sdk:

const ChatSDK = require("@livechat/chat-sdk");

const config = {
   LIVECHAT_ACCESS_TOKEN:"",
   LIVECHAT_ORGANIZATIONID:""
}

async initLivechatOld(){
    const chat = new ChatSDK({ debug: true });
    chat.init({
      access_token: config.LIVECHAT_ACCESS_TOKEN
    });
    chat.on("incoming_typing_indicator", (event) => {
      console.log("incoming_typing_indicator event", event);
    });
    chat.on("incoming_customer", (event) => {
      console.log("incoming_customer event", event);
    });
  }
initLivechatOld();

Using both libs, I receive incoming_customer event, but I don't receive any incoming_typing_indicator. Is something that needs to be activated from your side per subscription?

[orinciog]

I debugged more push events:

• incoming_sneak_peek -> I don't receive
• incoming_multicast -> I'm receiving

The problem can be very easy reproduced using above code. (LIVECHAT_ACCESS_TOKEN used has all access scopes selected from App Authorization -> App scopes and API access)

[DamCzerniakiewicz]

Hello, I've investigated your report based on the code you provided, and I could not entirely reproduce your case. To be precise:

• incoming_sneak_peek push was sent properly when a customer starts writing
• incoming_typing_indicator push was sent properly when the Agent starts writing

The typing indicator / sneak peek currently works as follows:
customer starts typing -> agent get sneak peek
agent starts typing -> customer gets typing indicator
Let me know if anything is unclear about that behavior.

[orinciog]

Hy!

Sorry, but in my case is not working. I don't receive those two pushs when a customer starts typing. I can provide a clip to see that I don't receive those two push. Maybe it should be activated per subscription?

I also have a ticket in LC support to investigate this issue (XUVT2Y).

[DamCzerniakiewicz]

I am the person from LC who is investigating this problem :D When a customer starts typing, you should only receive incoming_sneak_peek. The typing indicator is reserved for Agent. Just to add - the subscription plan does not have any effect on it

[orinciog]

Here is the clip in which I don't receive any incoming_typing_indicator or incoming_sneak_peek push.

Screen.Recording.2025-05-08.at.18.12.55.mov

A little explanation:

• ESSBOTDEMO is a LC chatbot that transfers the conversation to an app: BitdefenderESSDevelBot.
• In test/rtm.js I used exactly the above code, with an access token for the BitdefenderESSDevelBot app.

When the ESSBOTDEMO transfers the conversation to the app, an incoming_customer is received. After that, when the customers starts writing the messages, I don't receive any incoming_typing_indicator or sneak_peek push.

Also, from what I understand from you, there is a little misunderstanding in docs.
You say this:

• incoming_typing_indicator sent only to Customer when Agent is writing ( But in docs: Informs that one of the chat users is currently typing a message. The message hasn't been sent yet. The push payload contains the typing indicator object. ) (This document contains a reference of pushes available in the LiveChat Agent Chat API.) So, from docs, incoming_typing_indicator is available to both customer or agent when the other user starts typing.

[DamCzerniakiewicz]

Could you tell me what code you used for your tests?
The one below with LC SDK works properly for me (I've just added incoming_sneak_peek push)

const { Agent } = require("@livechat/lc-sdk-js");
const { WebSocket } = require("ws");  

const config = {
   LIVECHAT_ACCESS_TOKEN:"<<your_access_token>>",
   LIVECHAT_ORGANIZATIONID:"<<your_organization_id>>",
}

async function initLivechat(){
    const agentTokenGetter = () => ({
      accessToken: config.LIVECHAT_ACCESS_TOKEN,
      organizationID: config.LIVECHAT_ORGANIZATIONID,
      region: "dal",
      tokenType: "Bearer",
    });
    const agentRTM = new Agent.RTM(WebSocket, agentTokenGetter);

    try {
      await agentRTM.connect();
      await agentRTM.login();

      agentRTM.subscribePush("incoming_typing_indicator");

      agentRTM.on("incoming_typing_indicator", (push) => {
        console.log("Incoming typing indicator", push);
      });

      agentRTM.on("incoming_customer", (push) => {
        console.log("incoming_customer ", push);
      });

      agentRTM.on("incoming_sneak_peek", (push) => {
        console.log("incoming_sneak_peek ", push);
      });

     console.log(agentRTM.subscribedPushes);

    } catch (error) {
      throw new Error(`Unable to initialize Agent RTM connection: ${JSON.stringify(error)}`);
    }
  }

initLivechat();

As for the docs - you are right. We've introduced a change after your report to make it more precise. The phrasing is now
Informs that one of the chat users (an Agent or a Bot Agent) is currently typing a message. The message hasn't been sent yet. The push payload contains the typing indicator object.

Let me know if you were able to make it work.

[orinciog]

@DamCzerniakiewicz Thank you for watching over this issue. I'm using exactly your code. I'm not receiving any push.

To be more clear: when a Customer starts writing, I don't receive incoming_typing_indicator nor incoming_sneak_peek pushes in the code logged with Agent credentials. (the console.log from the above code is not fired).

What I need is in the Agent code to be announced when a Customer starts or stops writing. It seems that incoming_typing_indicator is a proper push notification, but I don't receive any push when a Customer starts or stops writing. The scenario is exactly like in above provided clip: a Customer starts writing a message and the Agent does not receive the pushes.

Actually, there are other reports that this push is not working (see link)

Can you tell me how "The one below with LC SDK works properly for me" ? What does works properly for me mean?

[DamCzerniakiewicz]

Hey @orinciog, apologies for the delay. Here is how the code above works for me: https://share.livechat.com/m5vY52pl. When the customer starts writing, I get incoming_sneak_peek. When the Agent starts writing, I get incoming_typing_indicator. I am using the Agent bearer token to establish the WS connection. Just to be sure, is the token you are using to establish the WS connection linked to the user of the chat?

[orinciog]

@DamCzerniakiewicz Any news about this bug?

Is there any way that can I receive incoming_typing_indicator event in the agent ?

@wfabjanczuk Sorry, but this is getting annoying: Can you please find a way to receive incoming_typing_indicator event ? I provided you all the details, code, movie, etc. So, can you please find a way to fix it? (No, the code you provided it is not working )

@DamCzerniakiewicz Sorry, but you created your own case where it is working. You have not replicated 100% my case. Please replicate the same case which I already described to you: (Livechat chatbot (created from https://app.chatbot.com/) that transfers the talk to an chatbot that has an app attached (created from https://platform.text.com/console/apps) and see if this is working.

Also, your movie is not having any connection with my usecase:

• You have 3 entities: johnsmile1 -> Human Customer, SamChatBot -> ChatBot, Damian -> Human Agent,
• I only have 2 entities: johnsmile1 -> Human Customer, SamChatBot -> ChatBot. There is not human here.

I want to find a way that in SamChatBot to be alerted when johnsmile1 starts and finishes typing.

I repeat: all other notifications are working, only these two are not emitted: incoming_typing_indicator and sneak_peek

[DamCzerniakiewicz]

@DamCzerniakiewicz Any news about this bug?

Hey, just to be precise, there is no bug here. This is the intended way of working.

Is there any way that can I receive incoming_typing_indicator event in the agent ?

The incoming_typing_indicator and sneak_peek are only available for the Agent who is part of the chat. If you are not a part of the chat, you can use follow_chat or add yourself to the chat with add_user_to_chat. That way, you can get the:

• incoming_typing_indicator from Agent chatting with the customer

• incoming_sneak_peak from customer

I want to find a way that in SamChatBot to be alerted when johnsmile1 starts and finishes typing.

What you are trying to achieve is not possible. ChatBot is a separate service that has its own backend and you cannot subscribe to the pushes that the bot receives. Your case feels more like a custom bot (docs), but at the moment, bots cannot establish WS connection in LiveChat (that may change in the future)

I want to find a way that in SamChatBot to be alerted when johnsmile1 starts and finishes typing.

Could you share what exactly you are trying to do with that knowledge? Is it important to know that johnsmile1 was typing something but changed his mind and didn't send a message in the end?

[orinciog]

So, maybe I was not clear enough in my previous messages, so let me clarify:

I have a ChatBot that talks to a Human Customer (johnsmile1). After a series of interactions, the ChatBot transfers the talk to a BotAgent ( SamBotAgent). After this, SamBotAgent is acting like a human agent, and it can responses to question asked by the Human Customer (johnsmile1).

So, the conversation between the Human Customer (johnsmile1) and the BotAgent (SamBotAgent) is supposed to be like the following:

johnsmile1> Question1
SamBotAgent> Answer1
johnsmile1> Question2
SamBotAgent> Answer2

But unfortunately, the human customers are not respecting this format, and they are asking questions like: (We saw that this is the majority of the cases):

johnsmile1> first part of question1
johnsmile1> second part of question1
johnsmile1> third part of question1
SamBotAgent> Answer1

So I need a mechanism to detect when the human customer stops asking parts of questions and when the BotAgent should start answering. This is the reason why I need to be notified in BotAgent about incoming_typing_indicator or incoming_sneak_peek .

Right now I have a mitigation of this problem, which is to use a timer. If the human customer does not send a message for a number of seconds, then I assume that he has finished asking his question and BotAgent can start answering. But this is not a perfect solution, because sometimes the human customer may take more than that fixed number of seconds to ask his question or it induces a unnecessary delay in the BotAgent's response.

Regarding notifications, in SamBotAgent I am receiving all kinds of pushes, from https://platform.text.com/docs/messaging/agent-chat-api/rtm-pushes (for example, when ChatBot transfers to talk to SamBotAgent, I am receiving incoming_customer ), but I am not receiving incoming_typing_indicator or incoming_sneak_peek pushes when the Human Customer starts typing.

I hope this clarifies the issue. I am looking for a way to detect when the human customer is typing or has finished typing, so that the BotAgent can respond appropriately without relying solely on a timer.