https://retellai.com/?utm_source=discord&utm_medium=seo logo
Powered by
Is there a way to customise the size of the live c...
# ❓・ai-evy-support
c
@AI Evy Is there a way to customise the size of the live chat?
a
Thanks for tagging me @Ciaran. I'll look through relevant sources and get back to you shortly.
### Answer I don't have information about customizing the size of the chat widget in the sources. The chat widget attributes documented cover things like 
data-title
, 
data-logo-url
, 
data-color
, 
data-theme-color
, 
data-component-color
, and 
data-fab-text
for appearance customization, but none relate to widget size/dimensions [(1) ]() . I could try to answer generally if you'd like, but I don't have documented information confirming a size customization option. Check out these potentially useful sources: [Retell Website Widget]() , [Platform Changelogs | Retell AI]() , [Chat Widget, Custom Function Enhancement and more!]()
​ To learn more, see the sources I considered: [[1] Retell Website Widget](https://docs.retellai.com/deploy/chat-widget) ​
Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.
c
@AI Evy So I can't make it full screen by default?
a
Thanks for the follow-up, @Ciaran. I'll get back to you soon.
### Answer No, there's no documented attribute to set the widget to full screen by default. The chat widget attributes cover 
data-title
, 
data-logo-url
, 
data-color
, 
data-theme-color
, 
data-component-color
, 
data-fab-text
, 
data-auto-open
(auto-opens the widget on page load, but doesn't make it full screen) , and other appearance/behavior options, but none control the widget's size or a full-screen mode [(1) ]() . Discord discussions also confirm that customization is limited to what's available through the script tag options (title, logo, color) , with anything beyond that requiring custom JavaScript/CSS overrides, and even then some elements (like the launcher button) sit inside a closed Shadow DOM that can't be styled with regular CSS [(2) ]() .
​ To learn more, see the sources I considered: [[1] Retell Website Widget](https://docs.retellai.com/deploy/chat-widget) [[2] Chat Widget Customisation](https://discord.com/channels/1186424376300679209/1387925742007881979) ​
Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.
Thanks for the follow-up, @Ciaran. I'll get back to you soon.
c
@AI Evy I'm trying to test this... Got an ngrok server running, I open the live chat and say hello and I get no reply?
a
### Answer I don't have specific documentation on troubleshooting a chat widget with no reply when testing via ngrok. The sources available cover webhook setup and custom function configuration for voice/chat agents, but nothing that directly addresses a "no reply" issue in a live chat test scenario. A couple of general debugging angles based on related patterns in the sources: - If your ngrok server is meant to receive webhook events (like 
call_started
, 
call_ended
, 
chat_started
, etc.) , confirm your endpoint is returning a 
2xx
status within the 10-second timeout window, otherwise Retell will retry up to 3 times [(1) ]() . - If you're using a custom function pointed at your ngrok URL, check that the endpoint is publicly reachable — Retell blocks requests to localhost, private IP ranges, and cloud metadata addresses, so your local server must be properly exposed through ngrok first [(2) ]() . - One community discussion also confirms that webhooks are things your agent calls at call lifecycle events — they're not something you call directly to get chat replies, so if you're expecting the chat interface itself to respond, that's a separate flow from the webhook setup [(3) ]() . If your ngrok server isn't logging any incoming requests at all when you send "hello" in the chat, that suggests the request isn't reaching your endpoint, which could point to a URL/configuration issue rather than a code issue. I could try to answer generally if you'd like, but I don't have documented information that directly addresses this specific "no reply" scenario for the chat widget test. Check out these potentially useful sources: [Retell webhooks overview](
/webhook-overview) , [Integrate any system with custom function]() , [Webhooks]()
​ To learn more, see the sources I considered: [[1] Retell webhooks overview](https://docs.retellai.com/features/webhook-overview) [[2] Integrate any system with custom function](https://docs.retellai.com/build/single-multi-prompt/custom-function) [[3] Webhooks](https://discord.com/channels/1186424376300679209/1367714169587240990) ​
Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.
Thanks for the follow-up, @Ciaran. I'll get back to you soon.
c
@AI Evy Do you have an endpoint to figure out when a version was deployed? I need to know when a specific version was deployed
a
### Answer I don't have information in the sources about a specific API endpoint that returns the deployment timestamp for a given agent version. The versioning content available covers the concept of drafts, published versions, and environment tags — noting that once you click Publish, the version becomes immutable and you add environment tags like "prod" to ship it [(1) ](

▾

) — but there's no mention of an endpoint that surfaces when a version was published/deployed. I could try to answer generally if you'd like, but I don't have documented information confirming this endpoint exists. Check out these potentially useful sources: [How to Use Versioning and Environment Tags with Retell AI](

▾

) , [Cannot deploy new agent version though web interface / API]() , [Agent not showing up]()
​ To learn more, see the sources I considered: [[1] How to Use Versioning and Environment Tags with Retell AI](

https://www.youtube.com/watch?v=Bbz9TRzc2W4▾

) ​
Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.
c
a
Thanks for the follow-up, @Ciaran. I'll get back to you soon.
### Answer Your message doesn't specify which error message or error code you're seeing, so I can't pinpoint the exact cause. That said, here's what's documented depending on what kind of "error" you're running into: If this is about a custom LLM WebSocket integration, there are four specific disconnection reasons tied to LLM WebSocket problems: - `error_llm_websocket_open`: Retell never got the connection open (3 attempts, 7-second timeout each, 3 seconds apart) , or your server closed the socket mid-call without a status code [(1) ]() - `error_llm_websocket_lost_connection`: keepalives stopped arriving and Retell used up its reconnects, or a reconnect couldn't reach your server — check whether you're echoing 
ping_pong
and whether your host/proxy has an idle timeout - `error_llm_websocket_runtime`: the socket errored mid-call, your server closed it with an unexpected code, or it kept closing abnormally (
1006
) until Retell ran out of reconnects - `error_llm_websocket_corrupt_payload`: close code 
1007
, sent when Retell receives a binary frame instead of text If this is about outbound calling, one community report describes the 
Invalid_destination
error occurring on outbound calls to other numbers, while calling the user's own number worked fine [(2) ]() . Another set of Discord reports describes a spoken "Network access error" with a long numeric code occurring mid-call, though no root cause or fix was confirmed in that thread [(3) ]() [(4) ]() . If this is a known outage, there was a resolved incident on Oct 28 where inbound calls exper
ienced connection issues and timeouts for about 24 minutes, caused by a brief telephony provider outage [(5) ]() . Can you share the exact error message or code you're seeing, and whether it's happening on a call, in text/chat testing, or in the API? That'll help me narrow this down.
​ To learn more, see the sources I considered: [[1] Troubleshooting](https://docs.retellai.com/integrate-llm/troubleshooting) [[2] Ai caller keep giving me the error message](https://community.retellai.com/t/ai-caller-keep-giving-me-the-error-message/18) [[3] Discord message from @shaun_72049 on 2025-10-27](https://discord.com/channels/1186424376300679209/1186462578310910112/1432347329612087401) [[4] Discord message from @usmanshazad on 2025-10-27](https://discord.com/channels/1186424376300679209/1186462578310910112/1432391192045293599) [[5] Intermittent Call Connection Failures](https://community.retellai.com/t/intermittent-call-connection-failures/143) ​
Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.
c
@AI Evy retell-widget-v2.js:12 POST https://api.retellai.com/create-chat 400 (Bad Request) Failed to create chat: ApiError: An unknown error occurred at Ee (retell-widget-v2.js:12:360912) at async we (retell-widget-v2.js:12:361060) at async retell-widget-v2.js:12:916591 at async y (retell-widget-v2.js:12:916326) at async startNewChat (retell-widget-v2.js:12:917405) at async retell-widget-v2.js:12:1389875 at async Object.handleOpen (retell-widget-v2.js:12:1390044) at async d (retell-widget-v2.js:12:1392803)
a
Thanks for the follow-up, @Ciaran. I'll get back to you soon.
### Answer I don't have specific documentation on what triggers a 
400 Bad Request
from the 
create-chat
endpoint when using the widget. However, based on what's documented for the widget setup, a few things to check: - The chat widget requires either 
data-public-key
or 
data-voice-public-key
, and to enable text chat you need 
data-agent-id
set to a chat agent ID [(1) ]() . If 
data-agent-id
is pointing to a voice agent instead of a chat agent, that mismatch could cause the chat creation request to fail. - Confirm your public key is properly configured — the setup requires 
data-public-key="YOUR_RETELL_PUBLIC_KEY"
and 
data-agent-id="YOUR_CHAT_AGENT_ID"
in the script tag :
Copy code
<script
 id="retell-widget"

 src="<https://dashboard.retellai.com/retell-widget-v2.js%22>
 type="module"
 data-public-key="YOUR_RETELL_PUBLIC_KEY"
 data-agent-id="YOUR_CHAT_AGENT_ID"
 data-voice-public-key="YOUR_VOICE_PUBLIC_KEY"
 data-voice-agent-id="YOUR_VOICE_AGENT_ID"
 data-agent-version="YOUR_AGENT_VERSION"
 data-title="YOUR_CUSTOM_TITLE"
 data-logo-url="YOUR_LOGO_URL"
 data-color="YOUR_CUSTOM_COLOR"
 data-theme-color="YOUR_THEME_COLOR"
 data-component-color="YOUR_COMPONENT_COLOR"
 data-fab-text="YOUR_FAB_TEXT"
 data-bot-name="YOUR_BOT_NAME"
 data-popup-message="YOUR_POPUP_MESSAGE"
 data-show-ai-popup="true"
 data-show-ai-popup-time="5"
 data-auto-open="false"
 data-dynamic='{"key": "value"}'
 data-white-label="YOUR_WHITE_LABEL_TOKEN"
 data-recaptcha-key="YOUR_GOOGLE_RECAPTCHA_SITE_KEY"
></script>
- If 
data-recaptcha-key
is set but reCAPTCHA protection isn't properly enabled for your public key in the Retell Public Keys settings, or the reCAPTCHA script isn't loaded correctly, this could also cause session creation to fail — reCAPTCHA v3 is required, and only v3 is supported . - Since you're testing through an ngrok tunnel, confirm the domain serving the widget matches an allowed domain configured for your public key (this was referenced in your earlier troubleshooting as well) . The 
create-chat
endpoint itself isn't documented in the sources I have (only [Create Outbound SMS]() [(2) ]() is available for chat-related API endpoints, which is a different endpoint for SMS specifically, not the widget's chat creation call) . Can you confirm: is 
data-agent-id
pointing to a chat agent (not a voice agent) , and is reCAPTCHA enabled for the public key you're using?
​ To learn more, see the sources I considered: [[1] Retell Website Widget](https://docs.retellai.com/deploy/chat-widget) [[2] Create Outbound SMS](https://docs.retellai.com/api-references/create-sms-chat) ​
Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.
c
@AI Evy Here's the script tag 🙂 <script data-agent-id="agent_..." data-agent-version="0" data-auto-open="true" data-bot-name="Annabelle" data-color="#0071D1" data-component-color="#0071D1" data-dynamic="{...}" data-fab-text="Live Chat" data-logo-url="https://.../images/logo-icon.jpg" data-popup-message="Need help with your prescription?" data-public-key="public_key_....." data-show-ai-popup="false" data-show-ai-popup-time="5" data-theme-color="#071a3e" data-title="Live Chat" id="retell-widget" src="https://dashboard.retellai.com/retell-widget-v2.js" type="module" > Redacted some of the important information
a
Thanks for the follow-up, @Ciaran. I'll get back to you soon.
### Answer Looking at your script tag, the required attributes are set correctly — 
data-public-key
and 
data-agent-id
(chat agent ID) are both present, which is the minimum required combination for the chat widget [(1) ]() . A couple of things stand out that could relate to the 
400 Bad Request
on `create-chat`: - 
data-dynamic="{...}"
— this needs to be a valid JSON string with actual key-value pairs, e.g. 
data-dynamic='{"company": "Acme"}'
. If 
{...}
is literally what's in your tag (not just redacted for this message) , the widget would fail to parse it as JSON when creating the chat session, which could trigger the 400 error. - Domain allow-list: your public key must have the domain serving the widget added to its allowed domains, with 
localhost
added separately for local testing [(2) ]() . Since you mentioned testing via ngrok earlier, confirm the ngrok domain (or 
localhost
if that's how it's being served) is added to the public key's allowed domains. Everything else in the tag — 
data-agent-version
, 
data-auto-open
, 
data-bot-name
, 
data-color
, 
data-component-color
, 
data-fab-text
, 
data-logo-url
, 
data-popup-message
, 
data-show-ai-popup
, 
data-show-ai-popup-time
, 
data-theme-color
, 
data-title
— matches documented optional attributes [(1) ]() . Can you confirm whether 
data-dynamic="{...}"
is literal in your actual tag, or just how you redacted it for this message? That would help narrow down whether it's the cause.
​ To learn more, see the sources I considered: [[1] Retell Website Widget](https://docs.retellai.com/deploy/chat-widget) [[2] Public keys](https://docs.retellai.com/accounts/public-keys) ​
Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.
Thanks for the follow-up, @Ciaran. I'll get back to you soon.
c
@AI Evy With Live chats, do they auto close at some point or stay open forever?
a
### Answer Chats do not stay open forever by default — there's an inactivity-based auto-close feature. The Auto-close inactive chats setting ends the chat automatically when the user stops responding, with a configurable timeout anywhere from 2 minutes to 72 hours [(1) ]() . This is controlled by 
end_chat_after_silence_ms
[(1) ]() [(2) ]() . If users stay silent for a period after agent speech, the chat ends — the minimum value allowed is 120,000 ms (2 minutes) and the maximum is 259,200,000 ms (72 hours) ; by default this is set to 3,600,000 ms (1 hour) [(2) ]() . You can also set an optional Auto-close message to display when the chat is automatically closed, via 
auto_close_message
[(1) ]() [(2) ]() . Beyond the inactivity timeout, a chat session can also end in two other ways: the agent ends it (e.g. through an end-chat node or function) , or you end it explicitly using the End chat API [(1) ]() .
​ To learn more, see the sources I considered: [[1] Create chat agent](https://docs.retellai.com/build/create-chat-agent) [[2] Create Chat Agent](https://docs.retellai.com/api-references/create-chat-agent) ​
Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.
PreviousNext
Powered by