v0.1.31

v0.1.31 (May 27, 2026)

https://github.com/import-ai/omnibox/releases/tag/v0.1.31

Web

1. Added Smart Folders
2. Improved the Notification Center refresh experience
   • Manual refresh now resets the 10-second auto-refresh timer
   • When there is no new content, refreshing no longer causes list text or the scrollbar position to jitter
3. Improved synchronization between the resource list and resource detail page
   • After actions such as creating a copy, renaming, or undoing deletion, the left resource list and right detail page stay in sync
   • Fixed incorrect resource list positioning in some cases
4. Improved action feedback for read-only permissions in team namespaces
   • Prompts are clearer when moving, uploading, or performing similar actions in a team namespace where you only have view permission
   • Removed unnecessary technical details such as 403 / 409 from some error messages
5. Improved some dark mode display issues
   • Fixed abnormal content title colors in the search dialog
   • Adjusted some underline, disabled state, and button styles
6. Improved namespace name validation prompts on Settings > General
   • When the input does not meet requirements, the prompt is shown below the input field
   • Validation messages are now "Namespace name must be at least 2 characters" and "Namespace name must be at most 64 characters"
7. Improved internationalization on Settings > API Keys
8. Fixed some known issues

OmniBox AI

1. Improved state persistence for OmniBox AI conversations on the web
   • Opening a historical conversation restores its search state
   • Tool selection state, such as web search and deep thinking, is preserved within the same conversation and after page refresh
   • Starting a new conversation clears the tool selection state automatically
2. Fixed an issue where the status stayed pending after a tool call failed or was rejected
3. Fixed an issue in mobile OmniBox AI where, after tapping a citation to open the source file, the title kept showing "Loading"

File Parsing

1. Improved the prompt shown when audio parsing fails
   • The original task error error is no longer shown directly
   • When no valid speech segment is detected, a friendlier prompt is shown: No valid speech content was detected. Please upload a longer, clearer audio file with human speech.
2. Improved the prompt shown when parsing fails because a file is too long
   • The technical message Content too long is no longer shown
   • Users are prompted that The current file content (xxxxx characters) exceeds the system processing limit (32768 characters). Please try splitting the document and uploading it again.
3. Fixed an issue where background tasks could still show as Completed after parsing failed because a file was too long
4. Improved tasks database implementation

Mobile (v0.1.30)

1. The Android Pricing page now supports promo codes
2. Fixed an issue where duplicated text input on mobile could fail to save
3. Mobile OmniBox AI is now implemented natively
4. Recording now supports continuing in the background, so switching apps or locking the screen will not automatically interrupt recording
5. Fixed an occasional issue where, during consecutive recordings, a short recording could be parsed as content from the previous longer recording
6. Improved the source-tracing return experience in mobile OmniBox AI: after tapping a source, you can return to the original Q&A page
7. Fixed extra top spacing on file detail and conversation pages on Xiaomi phones
8. Added Smart Folders
9. Fixed some known issues

What's Changed

• chore: replace MailHog with Mailpit by @omnibox-bot in #133
• chore: bump service images to v0.1.31 by @omnibox-bot in #134

Full Changelog: v0.1.30...v0.1.31