Skip to content

v0.1.37

Choose a tag to compare

@omnibox-bot omnibox-bot released this 07 Jul 09:30
0974370

v0.1.37 (July 7, 2026)

https://github.com/import-ai/omnibox/releases/tag/v0.1.37

Web

  1. Added context usage display on conversation pages, including used percentage, remaining percentage, and token usage
  2. Improved OmniBox AI chat history display by collapsing intermediate steps by default and adding left-side navigation for quickly revisiting conversations
  3. Added support for showing each message's creation time and answer time during conversations, and generation duration for merged messages
  4. Added built-in default questions to the chat input when there is no visible chat history
  5. Improved the Agent 2.1 approval experience, supporting three approval modes: manual, auto-approve, and auto-reject
  6. Fixed some known issues

OmniBox AI

  1. Agent 2.1 now supports viewing resource content line counts and child resource counts
  2. Agent 2.1 added OmniBox product Q&A capabilities
  3. Agent 2.1 added long-context capabilities, so long conversations no longer error directly when approaching the model context limit
  4. Improved Agent 2.1 memory saving and fixed an issue where Let OmniBox remember did not work in some scenarios

Sharing & Clipping

  1. Fixed abnormal saving when sharing content from the Twitter app
  2. Fixed an issue where sharing Zhihu content to OmniBox could fail without an error message when the login state was abnormal

Mobile APP

  1. Added text descriptions for Web Search and Deep Thinking in the Mobile APP OmniBox AI chat box, aligning with the web experience
  2. Added a confirmation dialog when tapping WeChat login in a mobile browser, then redirects to the mini program after confirmation
  3. Aligned resource tool call parameters and resource link display in answers with the web experience
  4. Resource links in answers now support jumping to the corresponding detail page, and Smart Folders support jumping to the Smart Folder detail page
  5. Improved approval mode button display, keeping it on one line in English and truncating overflow text

WeChat/QQ and Other Assistants

  1. Fixed an issue where WeChat and QQ message processing could be interrupted when API key permissions included Tag / Search

What's Changed

Full Changelog: v0.1.36...v0.1.37