Custom User Input Failing to Render

Using User Input Block.

Select Custom Interface.

Use chat to create even the most basic interface.

Preview or Render - Result: “Empty Interface” message on a coloured background with Mindstudio logo.

Even in one case where I previously had a working form, it has stopped working and I have to revert to standard interface for agent to be able to function.

I’ve tried creating new agent, new blocks etc - all the same result.

Link to a test agent showing the behaviour: https://app.mindstudio.ai/agents/c50899e5-7ee7-4646-b1db-850f5fe14989/edit

I’m experiencing the exact same issue. It looks like something may be broken with the custom user interfaces.

In addition, Live Preview is constantly logging repeated connect/disconnect events, for example:

[remy] Client connected.
[remy] Client disconnected.
[remy] Client connected.
[remy] Client disconnected.

This loop continues indefinitely. Has anyone identified the cause or a workaround?

Yes, that’s the same behaviour I’m seeing…

Hi @royden,

We’re currently investigating the root cause. In the meantime, we’ve pushed a temporary fix so you can keep working on your agents. I’ll get back to you once this is fully resolved.

Hi Alex. I can see your response is dated almost 3 weeks ago. I am running into this same issue so I assume it hasn’t been resolved yet. Is there a fix for this incoming soon? My membership just renewed but it’s seeming like I won’t be able to deploy my ai bot unless I can fix the UI. Thanks.

Hi @christinepoore,

Thanks for the post and welcome to the Community!

Our engineers are looking into this glitch, but it shouldn’t stop you from generating custom interfaces in the meantime.

Here’s a temporary workaround:

  1. Generate your code as usual
  2. Refresh the page
  3. Click Apply to save the generated code
  4. Click Compile

Let me know how it goes!

Hi @christinepoore,

Good news! This has been fixed, so you can skip the workaround above.

Now, when the code is generated, just click Apply, then Compile.

Let me know if there’s anything else!

Hey Alex,

Great thank you so much for getting back to me. I see it has been fixed, that’s perfect!

I am trying to deploy this ai chat bot I created on a website, however it’s not appearing the way I’d like. (I’ll reference what it does look like in photos vs. what I’d like it to look like) - and I am asking chatgpt (hehe) and they say it’s because I am showing the deployed agent rather than a website widget. It states I need to hit publish and look for (deploy/embed/website) to find that code - however I am not seeing that option. Could you point me in the right direction of where to look? I’d really appreciate it. :slight_smile:

Thanks!!


Hi @christinepoore,

You can find the iframe code in the Embedding settings of your agent:

Hope this helps!

Hey Alex,

You’re awesome, thank you. That worked perfect.

I have this working now, however it appears the “run history” hamburger menu is still viewable to the user. Is there a way to disable this? Seems as if we wouldn’t want users to be able to see the backend workflows. The chat experience also isn’t working like a true chat bot, keeping the experience in the chat screen, back and forth answers.

I was working on this previously: creating my own chat UI (a floating widget) and have it call my backend /chat endpoint that returns assistant text (the proxy approach), not embed the MindStudio run page in an iframe. (For context, this worked perfect for the problems I’m running into now, but I ran into a separate issue of it not holding chat memory and starting each turn like it’s a new chat, hence why I am more willing to go back to doing this via the mindstudio iframe embed).

I’m reading this may be the only way to have it be an actual chat bot experience and hide the hamburger menu UI? If that’s the case, I find that interesting that mindstudio wouldn’t already have a way to do this internally? I was referred to this platform by a friend because he said it was user-friendly but now it feels I need to be quite proficient in code if I want this set up properly. Let me know what you think lol.

I do appreciate your time and assistance! :slight_smile:



Hi @christinepoore,

Glad to hear the embedding is working!

Unfortunately, it’s not currently possible to remove the hamburger menu from embedded agents since it’s the only way users can access their previous chats. I’ve added your feedback to our feature request list, so our engineers will consider an alternative in future updates.

No backend workflows should be visible to users when they access their previous chats through the hamburger icon. If you’d like to remove the Workflow Preview animation while the agent runs through the steps before the Chat block, you can change it in the Startup Settings of your agent:

If you’d prefer full control over the chat UI to hide the menu icon the API route is the way to go. The key to getting chat-like behavior through the API is managing the conversation history on your end. Here’s how agents like this are typically set up:

  • Start block: Use Launch Variables to pass in the full message history and the latest user message
  • Generate Text block: Prompt it to answer the latest message based on the history, and save the response to a variable
  • End block: Return the generated response as a key-value pair (e.g. “response”: {{response}})

This would also solve the memory issue you ran into with your proxy approach.

Is the hamburger icon the main reason you’d want to switch to the API, or is there anything else you’re looking to customize?