[GH-ISSUE #5320] Few things more which can be improved #5022

Closed
opened 2026-06-05 14:51:33 -04:00 by yindo · 1 comment
Owner

Originally created by @CreativeCoder0 on GitHub (Apr 1, 2026).
Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/5320

Here are few more space for improvements i found. Btw I am using linux desktop version of anythingllm with ollama.

  1. side scrollbar to scroll upwards, scrolling upwards is a painful thing when u have a long chat.
  2. i used wrong prompt. now i don't want to start from beginning. we should have option to edit the prompt and continue from there, basically you are doing undo. yes i know we can edit the first prompt in that thread, but can we do so in every prompt so that we can undo a step from middle (we are continuing from there, after chats will disapper) or at least the last step.
  3. stop thinking mode - few models are capable of thinking but it is sometimes irritating. so please give us option to stop it.

a personal request

Think and decide with your team on other improvements but please bring enable/disable of thinking feature as soon as possible, if possible today - one of my project really need it, this is just a toggle button. you can add this for linux desktop version and only for ollama, so it will not take much time and will effect less people if you are not sure to add this. please please please...

P.S. I replied you back in last issue after your reply to the things i raised, but you did not reply, may be github doesn't show notifications of closed issues or you did not get time or forgot i don't know... i felt you are interested getting feedback, so ignoring the probablity of ignoring my replies https://github.com/Mintplex-Labs/anything-llm/issues/5292

Originally created by @CreativeCoder0 on GitHub (Apr 1, 2026). Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/5320 Here are few more space for improvements i found. Btw I am using linux desktop version of anythingllm with ollama. 1. side scrollbar to scroll upwards, scrolling upwards is a painful thing when u have a long chat. 2. i used wrong prompt. now i don't want to start from beginning. we should have option to edit the prompt and continue from there, basically you are doing undo. yes i know we can edit the first prompt in that thread, but can we do so in every prompt so that we can undo a step from middle (we are continuing from there, after chats will disapper) or at least the last step. 3. stop thinking mode - few models are capable of thinking but it is sometimes irritating. so please give us option to stop it. ## a personal request Think and decide with your team on other improvements but please bring enable/disable of thinking feature as soon as possible, if possible today - one of my project really need it, this is just a toggle button. you can add this for linux desktop version and only for ollama, so it will not take much time and will effect less people if you are not sure to add this. please please please... P.S. I replied you back in last issue after your reply to the things i raised, but you did not reply, may be github doesn't show notifications of closed issues or you did not get time or forgot i don't know... i felt you are interested getting feedback, so ignoring the probablity of ignoring my replies https://github.com/Mintplex-Labs/anything-llm/issues/5292
yindo closed this issue 2026-06-05 14:51:33 -04:00
Author
Owner

@timothycarambat commented on GitHub (Apr 2, 2026):

  1. side scrollbar to scroll upwards, scrolling upwards is a painful thing when u have a long chat.

Good callout, we'll look into improving scroll behavior for long conversations. Does it not show though once you start? It does on most systems

  1. we should have option to edit the prompt and continue from there, basically you are doing undo.

You can do this though. I think however there is a bug where the Edit icon is missing when a chat is fresh. If you leave a thread and come back - do you see the edit (pencil) icon? That is how to edit. You can edit and save and edit and re-submit.

  1. stop thinking mode - few models are capable of thinking but it is sometimes irritating. so please give us option to stop it.

this is just a toggle button. please please please... you can add this for linux desktop version and only for ollama so it will not take much time.

I wish this were true, it's genuinely not that simple. Even scoped to a single provider, thinking mode isn't a flag we can just flip. It touches how responses are streamed and rendered, and doing it properly without introducing regressions takes real work. It's on our list but I can't commit to a timeline on it. Not to mention that passing that flag to a provider for non-thinking models will also throw an error for some providers. Its just not that simple and we support nearly 30 providers who have various support for that flag.

P.S. I replied you back in last issue after your reply... I felt you are interested in getting feedback, so ignoring the probability of ignoring my replies #5292

Apologies for missing that. For context, across GitHub, Discord, etc I'm looking at somewhere around 200-350 messages/notifications a day. I do my best but things slip through. If something is urgent, opening a new issue is always the most reliable way to get visibility.

<!-- gh-comment-id:4180555705 --> @timothycarambat commented on GitHub (Apr 2, 2026): > 1. side scrollbar to scroll upwards, scrolling upwards is a painful thing when u have a long chat. Good callout, we'll look into improving scroll behavior for long conversations. Does it not show though once you start? It does on most systems > 2. we should have option to edit the prompt and continue from there, basically you are doing undo. You can do this though. I think however there is a bug where the Edit icon is missing when a chat is fresh. If you leave a thread and come back - do you see the edit (pencil) icon? That is how to edit. You can edit and save and edit and re-submit. > 3. stop thinking mode - few models are capable of thinking but it is sometimes irritating. so please give us option to stop it. > this is just a toggle button. please please please... you can add this for linux desktop version and only for ollama so it will not take much time. I wish this were true, it's genuinely not that simple. Even scoped to a single provider, thinking mode isn't a flag we can just flip. It touches how responses are streamed and rendered, and doing it properly without introducing regressions takes real work. It's on our list but I can't commit to a timeline on it. Not to mention that passing that flag to a provider for non-thinking models will also throw an error for some providers. Its just not that simple and we support nearly 30 providers who have various support for that flag. > P.S. I replied you back in last issue after your reply... I felt you are interested in getting feedback, so ignoring the probability of ignoring my replies #5292 Apologies for missing that. For context, across GitHub, Discord, etc I'm looking at somewhere around 200-350 messages/notifications a **day**. I do my best but things slip through. If something is urgent, opening a new issue is always the most reliable way to get visibility.
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: Mintplex-Labs/anything-llm#5022