关注

发布于 更新于

AI总结: 本文介绍了多个软件更新的变更说明,包括Cherry Studio、DeepChat、anx-reader、subs-check、new-api、xpipe、alist、mihomo、ublacklist、Ghosten-Player、higress、termora、TVBoxOSC和fit-framework等。每个软件的更新内容涵盖了新功能的添加、性能优化、bug修复以及其他改进,旨在提升用户体验和软件稳定性。 优化建议: 1. 对每个软件的更新内容进行分类,明确标识出新功能、修复的bug和性能优化,以便用户快速找到所需信息。 2. 增加版本号和发布日期的清晰标注,方便用户了解更新的时效性。 3. 提供用户反馈渠道,鼓励用户报告问题和提出建议,进一步完善软件。

🍒 Cherry Studio is a desktop client that supports for multiple LLM providers.

变更说明

⚠️ 仅供内部测试使用,普通用户请勿更新

⚠️ 更新务必备份数据,如果没有备份数据,升级后无法降级

v1.3.0 测试版本讨论区:

What's Changed

  • refactor(TopicsTab): Use onContextMenu instead of onMouseEnter
  • Refactor/message block structure
  • feat: add painting aihubmix provider
  • feat: 添加 messageBlock、messageThunk 和 useMessageOperations 使用指南文档
  • refactor(MainTextBlock): enhance content processing by ignoring tooluse
  • fix: message do not use knowledge
  • refactor: 重构网络搜索模块,支持快捷菜单快速切换
  • perf: <tool_use> display
  • fix(GeminiProvider): relocate grounding metadata handling
  • fix: The Error display of the failed mcp call shows that the Error type cannot be displayed
  • feat: optimize extract logic
  • feat(messageThunk): integrate autoRenameTopic functionality to update topic names based on assistant responses
  • fix: determining thinking process using block status
  • fix: update webSearch type and results structure in upgrade logic
  • style: optimize mcp arg name display layout when tool or prompt descr…
  • Fix/message block structure
  • feat(WindowService): add main window monitor for renderer process events
  • fix: prevent overriding block status on aborting
  • feat: support streaming for model health check
  • Fix error
  • fix: refresh thinking state on streaming
  • fix(messageThunk): ensure usage is only estimated when not already present in response
  • fix: reset hasReasoningContent
  • feat(OpenAIProvider): support image edit
  • perf: improve inputbar button memorization
  • fix(mcp-tools): improve error message formatting and response handlin…
  • refactor: enhance image block rendering and styling
  • refactor: remove unused selection window
  • fix: 调整图片高度,修复gemini画图
  • refactor: Seperate MiniWindow loading from MainWindow
  • Feat/qwen3 support
  • feat: implement store synchronization across windows
  • perf(inputbar): improve long text paste
  • refactor: simplify custom CSS functionality by store sync
  • fix(Anthropic): add base64 file handling to IPC and file management
  • fix: resolve scrolling issue on files page
  • Remove redundant condition check in MCP service args handling
  • fix(i18n): add missing translation for Ru
  • fix(messageThunk): enhance reset message logic to include model for s…
  • feat(messageOperations): add editMessageBlocks functionality and upda…
  • fix(OpenAIProvider): enhance token budget calculation in OpenRouter
  • refactor(WindowService): improve context menu setup
  • fix the issue of webdav local backup file storage
  • refactor(WindowService, ThemeProvider): streamline title bar style and theme toggling logic
  • fix(models): normalize model IDs to lowercase for consistency
  • chore(dependencies): update electron-updater to version 6.6.4 and remove patches
  • chore(dependencies): update electron-builder to 26.0.15
  • fix: remove redundant local variables
  • feat: 支持自定义助手地址
  • Fix/image height
  • fix: conditionally show loading more spinner
  • feat: popup question editor support translation assistant
  • fix: infinite scroller layout
  • refactor: add Linux support for margin adjustments in MinappPopupConainer and McpSettingsNavbar
  • Fix: 'Web Search' and 'Clear Context' don't work
  • feat(OpenAI): new Responses API support
  • fix: openaiprovider timeout is not an integer
  • fix: include thinking_millsec in message block creation
  • fix: memorize assistant with model
  • feat(miniapps): add Grok / X
  • feat: emit event on message complete
  • fix: update thinkingbutton model checks
  • feat(WebSearchButton): add 'Disable Web Search' option
  • fix(StoreSyncService): set flag bug when register IPC handler
  • 优化消息块渲染性能,使用 useMemo 缓存图像块组;移除调试日志以清理代码
  • feat: support dashscope reranker
  • fix: chat message translate
  • feat: change citation list style
  • feat: optimize knowledge recognize
  • fix: knowledge base url error
  • refactor(middleware): add extractReasoningMiddleware for enhanced text processing
  • fix: failed to connect OAuth MCP server
  • fix: citation list loading
  • feat: Add OpenDAL based Remote Storage class
  • feat: add zoom factor setting and localization support
  • chore: remove unuseful doc
  • feat[MCP]: implement login shell environment retrieval
  • feat(MCP): update auto-install server name and add server registratio…
  • fix(theme): citation list dark theme display & url repeatedly requesting
  • feat: enhance search functionality with optional HTTP options
  • Feat: change Gemini Safety Settings to OFF; upgrade package
  • fix(MCP): Revise the server type judgment logic to ensure the correct…
  • Feat/claude websearch support
  • feat: Custom mini app
  • refactor(ModelList): improve model list style and grouping
  • feat(minapps): add n8n mini app with logo and URL.
  • fix: display image from mcp response
  • feat: tooluse result display style optimization
  • fix: remove useless border and border radius for code block
  • perf: virtual list for quick panel and SelectModelPopup
  • feat: add the "Export Agent" feature
  • fix: user message usage
  • feat: ParateraAI 添加支持
  • fix: SelectModelPopup sticky header
  • feat: support both function call and system prompt for MCP tools
  • fix: update input schema reference in MessageTools for accurate tool …
  • feat: add inspect option to context menu with localization support
  • feat(Qwen3): Add Qwen3 Model Thinking Mode Switch
  • feat: combine to general reranker
  • fix: SelectModelPopup scrolling behaviour
  • fix: qwen3 check
  • feat(webSearch): add Bocha web search provider integration
  • docs(README): add Architecture overview document
  • fix: update selectedTypes logic in ModelEditContent for better handling
  • feat: add prompt display control
  • style: remove small size prop from Switch component in AssistantModel…
  • feat: Agent Regular Phrases
  • feat: regularPhrases可以随Agent一起导出。
  • fix: jina embedding error
  • feat(i18n): 更新机器翻译多语言,使用Qwen3 236b进行机器翻译。
  • fix: check isComposing on keydown
  • fix: clean up code
  • revert: customtag tooltip delay
  • Revert "fix: update selectedTypes logic in ModelEditContent for better handling"
  • fix: conditionally aquire focus for inputbar
  • refactor(checkAPI): check api or model with stream enabled first
  • feat: add message translate copy & close
  • fix(Inputbar): do not reset selection on focus
  • feat: add motion library for animations and enhance Spinner and Messa…
  • fix: next release bugs

New Contributors

  • made their first contribution in
  • made their first contribution in
  • made their first contribution in
  • made their first contribution in
  • made their first contribution in
  • made their first contribution in
  • made their first contribution in
  • made their first contribution in
  • made their first contribution in
  • made their first contribution in
  • made their first contribution in

Full Changelog:

🐬DeepChat - A smart assistant that connects powerful AI to your personal world

变更说明

🚀 DeepChat 0.1.1 正式发布 | 重新定义你的 AI 对话体验!

—— 不再是简单的 ChatBot,而是你的自然语言 Agent 工具🌟

✨ 本次主要更新内容 ✨

  • 优化了打包尺寸,最终包减数百兆
  • [issue] 自定义代理不支持认证/Custom proxy doesn't support authentication
  • fix: 水印拼接逻辑修改
  • feat: 知识库开关放到到每个具体知识库
  • 默认MCP设置可以开到30
  • fix 知识库开关展示逻辑 & 快捷键描述
  • 修复了一些Bug
  • Release 0.1.1

New Contributors

  • made their first contribution in

Full Changelog:

🔥 为什么选择 DeepChat?

商业友好:基于原版 开源,无任何协议外的额外约束,面向开源。 ✅ 开箱即用:极简配置,即刻开启你的智能对话之旅。 ✅ 极致灵活:自由切换模型,自定义模型源,满足你多样化的对话和探索需求。 ✅ 体验绝佳:LaTeX 公式渲染、代码高亮、Markdown 支持,模型对话从未如此顺畅。 ✅ 持续进化:我们倾听用户反馈,不断迭代更新,为你带来更卓越的 AI 对话体验。

📥 立即体验未来

💬 反馈有礼:欢迎提交你的宝贵建议,加入 VIP 用户社群,与我们一同塑造 DeepChat 的未来! <img width="400px" src="

Featuring powerful AI capabilities and supporting various e-book formats, it makes reading smarter and more focused.

变更说明

1.5.1

  • Fix: Can't open book note list in some cases
  • Fix: WebDAV sync show Not Found
  • Fix: Context menu is difficult to distinguish in e-ink mode
  • L10n: Optimized Arabic translation
  • 修复:某些特殊情况下笔记列表无法显示
  • 修复:某些情况下 WebDAV 同步时显示 Not Found 的问题
  • 修复:E-ink 模式下上下文菜单难以辨认
  • L10n: 优化阿拉伯语部分翻译

订阅转换合并,节点可用性,测速,重命名,导出为clash.meta/clash/base64/qx等等所有格式的订阅工具

变更说明

Changelog

  • 5dd4994d9b7b2e6fb0605777c0cb7964f33ebada feat: 添加测速带宽限制

AI模型接口管理与分发系统,支持将多种大模型转为统一格式调用,支持OpenAI、Claude等格式,可供个人或者企业内部管理与分发渠道使用,本项目基于One API二次开发。🍥 The next-generation LLM gateway and AI asset management system supports multiple languages.

变更说明

What's Changed

  • feat: Group rate limiting feature
  • feat: update OpenRouter balance
  • fix: ensure SSE ping packets are sent before upstream response
  • fix: correct formatting string in PriceData.ToSetting to handle Image…
  • feat: add option to allow worker HTTP image requests

New Contributors

  • made their first contribution in

Full Changelog:

Access your entire server infrastructure from your local desktop

变更说明

SSH

  • The shell profile loading on Linux has been reworked with the goal of better handling the SSH agent detection
  • The SSH identity none selection will now always disable using default keys like id_rsa
  • There is now a new dialog to add a host key type to a connection if none of available ones are supported by the client
  • The support for FIDO2 SSH keys on Linux and macOS has been improved:
    • PINs are now cached by default so they only have to be entered once. Caching can be disabled in the security settings
    • User presence confirmation requests are now shown with an icon notification in XPipe
    • Fix FIDO2 keys not working with xpipe if they were already added to the SSH agent
  • You can now create all types of tunnels to VMs as well
  • The tunnel addresses now support IPv6 addresses
  • Add some more automatic recognition for some Mikrotik devices

Other

  • There's a new option to automatically close terminals on successful exit without having to press any key
  • The custom icons now dynamically check whether the colored icon variant has enough contrast in dark mode. This will make icons more colorful instead of mostly using black-and-white icons in dark mode. You have to refresh the icons in the icon settings for this to apply.
  • There is now a confirmation prompt when erasing file contents in the file browser
  • Add support for Enpass password manager
  • The arch AUR package now supports automatic updates via makepkg from within XPipe
  • There is now an automatic check and a new button for merge conflicts when another user/instance pushed to a remote git sync repo
  • Add support for Void editor
  • The docker parent entry no longer has a refresh button to prevent confusion with multiple refresh buttons
  • The actual file names of shell session script entries are now shown on hover
  • Add ability to close connection creation dialog quickly with ESC if no changes were made
  • Improve category styling

Fixes

  • Fix terminal logging being broken on Linux and macOS
  • Fix some file existence checks failing on Windows 10 since the latest Windows updates
  • Fix zsh errors due to unquoted glob pattern
  • Fix slow macOS startup due to mdfind being slow
  • Fix zellij multiplexer not properly opening first tab sometimes due to timing issues
  • Fix tmux window immediately closing if connection failed
  • Fix choco updater showing available updates before they are available in choco
  • Fix choco updater not launching as admin
  • Fix updater terminal window not closing automatically on success
  • Fix script group naming in creation dialog
  • Fix various NullPointers

Downloads

You can find all downloadable artifacts below attached to this release. For installation instructions, see the .

All artifacts are signed by

🗂️A file list/WebDAV program that supports multiple storages, powered by Gin and Solidjs. / 一个支持多存储的文件列表/WebDAV程序,使用 Gin 和 Solidjs。

变更说明

   🚀 Features

  • Add h2c for http server  -  by j2rong4cn in
  • Add UseLargeThumbnail for 139  -  by bigQY in
  • 115_open:
    • Implement rate limiting for API requests  -  by Andy Hsu
  • 139:
    • Add option ReportRealSize ( close )  -  by MadDogOwner and Copilot in and
  • azure_blob:
    • Implement GetRootId interface in Addition struct  -  by New Future in
  • cloudreve:
    • S3 policy support  -  by MadDogOwner in
  • doubao:
    • Support upload ( close )  -  by asdfghjkl and Copilot in and
    • Add get_download_info API and download_api option  -  by MadDogOwner in
  • doubao_share:
    • Support doubao_share link  -  by asdfghjkl and anobodys in
  • driver:
    • Add Azure Blob Storage driver  -  by New Future and Copilot in
  • url-tree:
    • Implement the Put interface to support adding links directly to the UrlTree on the web side  -  by Lee CQ and Copilot in

   🐞 Bug Fixes

  • Revert "refactor(net): pass request header"  -  by j2rong4cn in
  • Shebang of entrypoint.sh  -  by Mmx in
  • Remove auth middleware for authn login  -  by Mmx in
  • Prevent guest user from updating profile  -  by yoclo in
  • 115_open:
    • Add delay in MakeDir function to handle rate limiting  -  by Andy Hsu
  • 139:
    • Incorrect host  -  by Sam- Pan(潘绍森) and Copilot in
  • aliyundrive_open:
    • Resolve file duplication issues and improve path handling  -  by Yifan Gao and Copilot in
  • archive:
    • Unable to preview  -  by j2rong4cn in
  • baidu_netdisk:
    • Add another video crack api  -  by asdfghjkl and anobodys in
  • deps:
    • Update 115-sdk-go to v0.1.5  -  by Andy Hsu
  • docker_release:
    • Avoid duplicate occupation in docker image ( close )  -  by wxnq in and
  • doubao:
    • Update file size type to int64  -  by MadDogOwner in
  • fs:
    • Remove old target object from cache before updating  -  by Yifan Gao in
  • ipfs:
    • Fix problems  -  by jerry in
  • lanzou:
    • Remove JavaScript comments from response data  -  by MadDogOwner in
  • mega:
    • Use newest file for same filename ( close )  -  by gdm257 in and
  • net:
    • Unexpected write ( close )  -  by j2rong4cn in and
  • netease_music:
    • Change ListResp size fields from string to int64  -  by MadDogOwner in
  • pikpak&pikpak_share:
    • Update algorithms  -  by YangXu in
    • Fix WebPackageName  -  by Dgs in
  • thunder:
    • Fix login issue ( close )  -  by Dgs in and

   🏎 Performance

  • Optimize IO read/write usage  -  by j2rong4cn and MadDogOwner in
  • local: Avoid duplicate parsing of VideoThumbPos  -  by Lin Tianchuan in
  • quark_uc&quark_uc_tv: Native proxy multithreading  -  by xiaoQQya in
    

A simple Python Pydantic model for Honkai: Star Rail parsed data from the Mihomo API.

变更说明

Release created at Sun May 11 11:57:07 CST 2025 Synchronize Alpha branch code updates, keeping only the latest version

Blocks specific sites from appearing in Google search results

变更说明

(2025-05-07)

Bug Fixes

  • locales: update German translation

Features

  • add experimental SERPINFO mode
  • searx: add searxng instance
  • yandex: add support for yandex.ru

This release is also available on:

一个同时适配📱Android Phone和📺Android TV的视频播放器。整合影视资源,自动刮削媒体信息,支持IPTV播放,优雅打造私人影视库。界面简洁,功能丰富,开箱即用。

变更说明

如果项目对您有帮助,就请给颗吧。 请我喝咖啡(支持作者) ☕️ QQ群 1033449405

🐞Bug Fixed

  1. TV端播放失败后,切换播放列表无法恢复播放
  2. DLNA投屏问题以及页面展示问题
  3. 夸克网盘VIP速度过低

What's Changed

  • Releases/v2.0.3

Full Changelog:

🤖 AI Gateway AI Native API Gateway

变更说明

What's Changed

  • Update helm translated README.zh.md
  • fix: update golang filter README
  • fix : Add nacos username and password login option
  • feat: Support /v1/models API in ai-proxy
  • fix(ai-statistics): adjust requestBodyBufferLimit
  • fix : fix vs rewrite when mcp protocol is http
  • feat: Add SSE direct proxy support to mcp-session filter
  • fix param type error

Full Changelog:

Termora is a terminal emulator and SSH client for Windows, macOS and Linux.

变更说明

New features/Updates

  • Support RDP protocol
  • Highlighter keywords support regex
  • Support for compression algorithms
  • Support for xterm CBT
  • Scroll to the bottom after pressed any key

Bug fixes

  • Fix snippet \ characters
  • Fix snippet i18n

新功能/更新

  • 支持 RDP 协议
  • 高亮关键词支持正则
  • 支持压缩算法
  • 支持 xterm CBT
  • 任意键按下后滚动到底部

问题修复

  • 修复代码片段 \ 字符
  • 修复代码片段 i18n
变更说明

Credit: Commit: 117fa9593ba22b2e81c21b87303ec1981f830d89 Changelog:

解决部分盒子不能正常筛选分类的问题

FIT: 企业级AI开发框架,提供多语言函数引擎(FIT)、流式编排引擎(WaterFlow)及Java生态的LangChain替代方案(FEL)。原生/Spring双模运行,支持插件热插拔与智能聚散部署,无缝统一大模型与业务系统。

变更说明

这是 FIT Framework 社区 3.5.0 版本的第二个里程碑的发布!

FIT 函数平台

✨ 升级优化

  • 支持 mybatis 配置自动下划线转驼峰
  • 添加 fit-security-simple 插件
  • 捕获流式调用报错信息
  • 升级 python 依赖三方包的版本

✅ 问题修复

  • 修复校验泛型元素为空时的错误
  • 修复 http 进行流式下载时,status code 返回 0 导致链接中断的问题
  • 修复广播模式下如果 genericable 无返回值时的报错问题

Waterflow 流调度平台

🚀 功能特性

  • 支持 FIT for Java 插件自动扫描并生成工具元数据。

✨ 升级优化

  • 修改 ohscript 生成随机 uuid 方式提升性能

Elsa 图形编辑平台

🚀 功能特性

核心框架 (elsa)

  • 图形具备统一的抽象数据结构。

    • Graph、Page、Shape 结构通用于任意业务场景。
    • 全链路 JSON 序列化能力,兼容任何持久化存储方案。
  • 支持拖拽式流程编排。

    • 支持图形拖拽和画布拖拽,并提供节点整理和一键显示所有节点的能力。

前端应用 (elsa-react)

  • 集成 React 的能力。

    • 基于 Context 的上下文传递机制。
    • 节点渲染缓存:React.memo + 自定义 shouldComponentUpdate。
  • 集成 Ant Design 的能力。

    • 基于 Form 组件的实时校验提示系统。
    • 基于 Tree 组件封装的节点上下文观察者机制。

❤️ 贡献者

  • made their first contribution in
  • made their first contribution in

以及感谢所有为此次发布做出贡献的人: