OMate 1.9.2 (b216): Stream output adds smart scrolling, supports mouse wheel and dragging

OMate
1

Modified the dialogue loading method to allow streaming output to be draggable, at the cost of a loading process when entering the chat interface.

  • Upgraded the screenshot tool to optimize the overlap issue between buttons and the status bar on some devices
  • Backtracking supports manually specifying titles
  • Added a floating button for quick scrolling to the bottom on the chat page
  • Streaming output adds intelligent scrolling, supporting mouse wheel and drag
  • When the user does not provide modification suggestions and memory is not used as a reference, no rewrite prompt is added (some models may produce repeated output)
  • Notepad automatically loads prompt words supporting {{char}} and {{user}} variables
  • Fixed the issue where the speaker’s name in the message bubble was incomplete when the content was too short
  • Adjusted the default values for permanent memory-related settings, changing the context number from 10 to 3
  • Fixed the issue where the calendar fetching tool could not obtain the id on Android devices due to compile compression
  • Fixed the error caused by a deleted knowledge base

b218 Update

  • Fixed the issue where after mounting a knowledge base txt in the role settings, editing and re-vectorizing the knowledge base txt required remounting it in the role settings
  • Automatically syncs and updates the corresponding options in role settings and event settings when renaming the knowledge base
  • Fixed the issue where the loading mask on the chat page did not disappear when the message count was 0

b219 Update

  • Fixed the issue where an error occurred when summarizing the reasoning log after creating a new topic

b220 Update

  • Memory extractor prompt template (memoryExtractorPrompt) supports the {{knowledge_content}} variable, allowing the content extracted from the knowledge base in this conversation to be passed in

b221 Update

  • Optimized tool definitions to be compatible with more interface parameters

b222 Update

  • Optimized support for Gemini 3 inference mode

b223 Update

  • Fixed the issue where changes to role knowledge base settings in dialogue management did not take effect immediately when returning to the dialogue interface
  • Fixed the permanent memory error that might be caused by legacy data
  • Optimized the scrolling of the chat bubble menu and when the keyboard is launched
  • Long-term memory allows setting 1 as the minimum value
  • When historical records exist but are pushed out due to length limitations, a prompt is displayed

b224 Update

  • Fixed the issue where deleting content might reappear when the keyboard is retracted
  • Optimized the height of the scrollbar appearance on the settings page for API/masks/roles (changed to screen ratio)

b225 Update

  • Regular expressions support replacement with data in the notepad
  • Optimized the position of the add and edit buttons for the scrollbar on the dialogue management page for API/masks/roles
  • Fixed the issue where edits to API/masks/roles on the dialogue management page did not take effect in real-time on the dialogue interface
  • Fixed the error that occurred when exiting immediately after entering the dialogue page

b226 Update

  • Adjusted the margin and transparency of the scrollbar on the dialogue management page for API/masks/roles
  • Added the gemini-flash-3-preview model to the list
  • Optimized the performance of the chat interface to avoid lag

b227 Update

  • Added an undo menu for the last message, allowing the last message to be undone and rolling back long-term memory and role notepad data
  • Fixed the issue where entering API edit-replica configuration on the dialogue management page, turning to the replica configuration, caused multiple identical key configurations to be saved due to lag

b228 Update

  • Optimized the issue of the latest message being blocked when the keyboard is raised
  • De-duplicated knowledge base hit results

Issue Description

Explanation of the issue with the dialogue management interface - API list lag and loss:

  • The list is not stuck, but an error occurred because there was an API config with the same id
  • These API configs with the same id are mostly due to clicking - edit - replicate - turn to new configuration copy - save on the dialogue management page
  • How to solve: The part of the API list that is not displayed in the dialogue management interface is the API configuration with duplicate ids (in chronological order). Open the corresponding API from the settings page, replicate it again, and then delete the original.

b227 will handle the issue of saving duplicates when turning to a new configuration copy, but legacy data still needs to be cleaned manually.

Download link: https://download.omate.net/omate-pro-1.9.2-b228.apk

4

It seems that this new conversation loading method is causing an infinite loading of empty conversations…

I imported a new card without an opening line, and as soon as the conversation starts, it keeps loading indefinitely, making it impossible to operate on the conversation page. Since it’s a new card, there are no historical topics to read, and going into settings to reset doesn’t work either.

Only by adding a random opening line to the card and then resetting can it be operated.

5

When a knowledge base txt is mounted in the character settings, after editing and re-vectorizing the knowledge base txt, it needs to be mounted again in the character settings.

6

@YeZip @Selvadin b218 fixed the reported issue, try again.

7

The issue with mounting the knowledge base update has been resolved, and the log errors have not reappeared.

In the custom prompt section, the “Role Information Extraction Prompt” can use the variable {{recent_chat_history}}, which is the text before the regex. My status bar and chain of thought have a small amount of text, which currently does not seem to affect the use of the summary model.

8

I encountered a strange bug. After updating this version, my Meituan longcat API stopped working. The test in the settings was successful, but I no longer receive replies in the conversation. It keeps spinning, but other models are fine.

9

LongCat previously had issues with streaming, and non-streaming needs to be used. Try non-streaming first. Additionally, there are some issues with the parameter definitions used in the tool, and I will update a version later to optimize it.

10

In the conversation settings—basic settings, click the pencil icon to edit the character settings. Setting the maximum number of recalls and the threshold for the knowledge base will not work; you must exit and edit the character settings from outside.

11

It seems that it didn’t take effect immediately. Did you try exiting the conversation and re-entering?

13

It indeed doesn’t take effect immediately; just exit the conversation and re-enter.

14

I have noted this issue and will fix it later.

16

Good evening, Teacher Fang Tang, waiting for the PC version.

17

In version b223, the message editing window does not automatically clear after sending, and sometimes even when manually cleared, a character may remain and automatically reappear.

18

This issue has been documented and will be fixed in the next build.

19

IMG_20251217_194937

May I suggest moving this edit button to the top?

20

Teacher, the input box issue has been resolved. However, I can’t change the API name or key from within the conversation management. Even after changing the settings from outside and returning to the conversation management, it remains unchanged…

21

Oh, there’s a problem again. I have 22 APIs in the conversation management, but there are 24 in the settings. I should have 24…

22

Is it possible that the list can be scrolled up and down?

23

I know this, but it is indeed missing. Also, the API information cannot be changed in the dialogue management.

1 Like