Nightly Build 2026-04-23

Nightly Build 2026-04-23 (20260423-ae1546f7)

⚠️ This is an unsigned nightly build for testing purposes.

🍎 macOS users: If you see "N.E.K.O is damaged and can't be opened", run this in Terminal:

xattr -cr /path-to/N.E.K.O.app

Desktop App (Electron + Python backend)

Platform	File
Windows x64	N.E.K.O.exe (Portable)
macOS Intel	N.E.K.O-*.dmg / .zip
macOS Apple Silicon	N.E.K.O-*-arm64.dmg / .zip
Linux x64	N.E.K.O-*.AppImage / .deb / .tar.gz

Python Backend Only (standalone, no Electron)

Platform	File
Windows x64	python-backend-win-*.zip
macOS Intel	python-backend-mac-x64-*.zip
macOS Apple Silicon	python-backend-mac-arm64-*.zip
Linux x64	python-backend-linux-*.zip

---

Built from 20260423-ae1546f7

[v0.7.4] Massive Performance & UI Upgrades

原本只计划了一次小更新，没想到最后变成了一次大更新。

[主要更新]

• 史诗级优化：难以置信的性能提升！
• UI全面进一步美化：聊天窗口 UI 全面升级，优雅、美观、现代！
• VRM也能跳舞：新增 VRM 模型的舞蹈动作支持
• 聊天窗口内嵌音乐卡片，实时同步播放状态
• Live3D待机动画支持多选 + 动画组随机轮换
• 现在聊天窗口和角色形象可以分别在不同屏幕上显示。

[次要更新]

• 音乐推荐系统统一重构，新增网易云音乐登录功能
• 修复 VRM/MMD 模型锁定后淡化机制失效的问题
• 修复自定义 API 中 TTS 被误判为 GPT-SoVITS
• 修复通义千问对话中偶发的 400 错误
• 修复 Steam 上传页面无法识别 MMD 模型文件
• 修复 Agent 任务无限重触发循环
• 更多 bug 修复和细节优化

[Major Updates]

• Epic Optimization: Unbelievable performance boost!
• Further UI Polish: Chat window UI fully revamped — elegant, beautiful, and modern!
• VRM Can Dance Too: Added dance action support for VRM models
• Embedded music card in chat window with real-time playback status sync
• Live3D Idle animations now support multi-select + random group rotation
• Chat window and character model can now be displayed on separate screens

[Minor Updates]

• Music recommendation system unified with new Netease Cloud Music login support.
• Fixed VRM/MMD model lock causing fade mechanism to break.
• Fixed custom API TTS being misidentified as GPT-SoVITS.
• Fixed occasional 400 errors in Qwen conversations.
• Fixed Steam upload page failing to recognize MMD model files.
• Fixed agent task infinite re-trigger loop.
• Additional bug fixes and detail optimizations.

What's Changed

• 修复 analyze_image_with_vision_model() 中的 httpx 连接泄漏问题 by @icyaaaww in #640
• fix: 记忆模块防幻觉——fact提取忽略gibberish，角色prompt禁止编造主人信息 by @wehos in #642
• feat: add avatar reaction bubble system by @yiyiyiyiGKY in #632
• feat: 文本模式记忆整理增加轮次/上下文长度触发机制 by @wehos in #644
• 优化：在 _validate_image_data() 中添加 image.load() 强制完整像素解码 by @icyaaaww in #643
• feat: RNNoise 打包修复 + VAD 持续阈值 + 前端降噪开关 by @wehos in #647
• 优化：在 DirectTaskExecutor 中缓存 AsyncOpenAI 客户端，防止连接池泄漏 by @icyaaaww in #648
• feat: 音乐秒关自动冷却 + 阿里百炼/国际版拆分 by @wehos in #646
• 修复：保护所有 asyncio.create_task() 后台任务不被 GC 回收 by @icyaaaww in #650
• feat: 增强主动问候系统——时段感知、节日/假期提醒、meow多语种 by @wehos in #651
• Claude/fervent lichterman by @wehos in #652
• Fix/websocket frame guard and lite denoiser by @wehos in #654
• fix: disable hover-fade on mobile, clamp audio schedule time, and loa… by @wehos in #656
• fix: harden avatar reaction bubble cleanup by @yiyiyiyiGKY in #657
• 修复 Electron 环境下角色卡导出 0 字节且重复弹窗的问题 by @MingTianSang in #661
• 版本迭代：修改自动启动，修改napcat启动方式，剔除本地数据 by @zhaijiunknow in #660
• fix: suppress mic input during proactive nudge injection by @wehos in #662
• feat: multi-window IPC support (routes, templates, TTS retry, VRM res… by @wehos in #515
• feat: MMD/VRM 待机动画改为多选，支持定时随机轮换 by @wehos in #649
• 修复自定义API配置保存后API Key无法正确显示的Bug by @MingTianSang in #665
• 优化声音克隆页面UI;在页面右上角添加"前往API设置"快捷按钮 by @MingTianSang in #663
• 修复：在 FactStore 中添加单字符级线程锁，防止读写竞争条件 by @icyaaaww in #653
• 修复steam上传页面无法显示识别MMD和VRM模型导致无法上传的问题 by @MingTianSang in #664
• 修复 VRM 锁定模型后左键点击导致淡化机制失效的问题，并为 MMD 新增同等的锁定淡化功能 by @MingTianSang in #666
• 添加MMD回退机制 by @MingTianSang in #670
• feat(music): Add Netease Cloud Music credential page & expand music keywords by @Himifox in #668
• feat(mijia): 米家插件增强 - 设备控制可靠性与别名功能 by @wangjunyu200708 in #674
• Add new SVG icons for sharing, image, and import functionalities by @LyaQanYi in #672
• 紧急修复：qq插件掉文件 by @zhaijiunknow in #673
• fix: 修复 create_response 在对话中间插入 SystemMessage 导致通义千问 400 错误 by @wehos in #675
• fix: agent task tracking + two-stage plugin filtering + pyncm_async compat by @wehos in #677
• 为淡化功能添加开关 by @MingTianSang in #676
• 优化UI,增加歌曲动作管理器,支持导入导出,可播放时调整进度条 by @KirisameYuumio in #667
• 为部分语言添加喵喵喵 by @MingTianSang in #682
• 将 idleAnimations / mmd_idle_animations 加入保留字段过滤列表 by @MingTianSang in #678
• feat: 添加聊天窗口大小调整功能和暗色模式样式（N.E.K.O-PC #30） by @LyaQanYi in #679
• 优化API设置页面UI，为核心API和辅助API填加胶囊样式，为下拉菜单填加圆角样式 by @MingTianSang in #683
• 统一 LLM 客户端调用、抑制爬虫日志噪音、修复测试基础设施 by @wehos in #685
• feat: 健康检查新增api域名 + 遥测版本号取changelog + 重试静默前3次 by @wehos in #686
• feat: 移除不必要的 streamingStatusLabel 属性，并在 React 聊天窗口中添加音乐卡片消息功能 by @LyaQanYi in #688
• fix: prevent agent task infinite re-trigger loop by @wehos in #687
• 将跟踪鼠标和全屏跟踪/局部跟踪按钮放到不同行 by @MingTianSang in #690
• i18n: 视觉描述 prompt 国际化（保留安全水印） by @wehos in #689
• 修复自定义API设置中普通TTS被误判为GPT-SoVITS的问题 by @MingTianSang in #691
• feat: 待机动画在循环结束时切换，避免动作中途跳变 by @wehos in #693
• 歌曲推荐播放器宽度自适应,曲名过长时滚动显示 by @KirisameYuumio in #692
• Feat(music card): sync music-card playback status by @Himifox in #695
• refactor(music): unify music API route, domain pool and code structure by @Himifox in #694
• 修复设置面板相关开关在选项之间慢速移动时鼠标光标短暂变为箭头的问题 by @MingTianSang in #699
• feat(react-chat): 补回点歌台与头像生成按钮至 React 聊天窗口顶栏 by @wehos in #697
• Add new send icon in PNG format to static icons directory by @LyaQanYi in #700
• fix: /chat 独立窗口头像无法加载 — 补齐缺失脚本 + BC 跨窗口转发 by @wehos in #696
• feat: React 聊天窗口最小化恢复为左下角悬浮球 + 呼吸灯动画 by @wehos in #698
• Add avatar preview button to chat header and index templates; include new icon asset by @LyaQanYi in #705
• fix: 修复 React 聊天窗口最小化悬浮球定位、图标缺失及 /chat 页面兼容问题 by @wehos in #706
• 修复 Live2D 与 VRM/MMD 来回切换后待机动作丢失并显示为无动作的问题 by @MingTianSang in #702
• fix: chat 独立窗口不再弹 toast，并禁止路由保留名用作角色名 by @wehos in #707
• feat(react-chat): 在 Composer 工具栏添加常驻翻译开关按钮 by @wehos in #708
• 重构 MiniMax TTS WebSocket 连接生命周期并修复空闲超时问题 by @MomiJiSan in #704
• fix: 调整消息列表的内边距以改善布局 by @LyaQanYi in #711
• 优化后端网页设置弹窗与聊天窗口局部 UI 细节 by @MomiJiSan in #709
• 修复 Live3D 光照设置在模型切换后重置为错误默认值的问题 by @wehos in #710
• fix: VRM 模型发送消息后消失 + Task HUD 轮询冗余清理 by @wehos in #712
• fix: 移除聊天窗口和消息列表的空文本属性，调整背景样式为透明 by @LyaQanYi in #713
• feat(react-chat): 为标题栏与输入栏背景添加 Liquid Glass 毛玻璃质感 by @wehos in #714
• feat: 添加窗口拖拽功能并调整样式层级 by @LyaQanYi in #715
• fix: 统一 LLM/TTS 重试错误上报策略，区分立即上报与延迟上报 by @wehos in #717
• fix: 统一 LLM JSON 响应容错解析 (robust_json_loads) by @wehos in #718
• feat: 添加多窗口模式下的头像预览功能支持 by @LyaQanYi in #720
• 修复模型管理页的初始化时序问题，避免 VRM 光照控件和待机轮换状态在声明前被访问；恢复 MMD 模型加载后的待机动作恢复与轮换逻辑，解决… by @MingTianSang in #724
• Fix/floating trigger instant rotation by @MomiJiSan in #723
• 修复独立聊天窗口在角色切换时误触发本地 Live2D 模型热切换，导致未加载 Live2D 运行时的页面抛出 Live2DManager … by @MingTianSang in #726
• fix: 修复 Windows 下角色数据库初始化失败（目录缺失 + 反斜杠 URI） by @wehos in #727
• feat: v0.7.4 — i18n 国际化 + 多窗口增强 + 模型/音乐/UI 修复 by @wehos in #728
• fix: 修理教程高亮范围，优化角色模型页面教程逻辑 by @KirisameYuumio in #730
• fix: 修复频繁切换角色时 TTS 错误级联 + 假性 API Key 拒绝弹窗 by @wehos in #731
• feat: 添加独立的 Jukebox 管理器页面和窗口支持 by @LyaQanYi in https://github.com/Project-N-...

[v0.7.3] You can really dance! - QQ Bilibili 米家 plugins added!

[主要更新]

• 首次连接或切换角色时，AI 会根据主动和你打招呼
• 语音通话模式同样支持主动搭话
  -全新的长期记忆系统已悄然实装，我们暂时不打算透露细节，敬请期待后续调整
• 基于阿里 CoPaw 新增云端 OpenClaw 接入，支持将任务路由至云端执行
• 新增基于 OpenFang 的本地虚拟机，让猫娘可以在后台做自己想做的事情
• 新增角色卡导出/导入功能，支持"仅设定"和"完整角色卡"两种模式
• API Key 现在以脱敏形式显示，操作更安全
• 新增 API Key Book 快速填充与跳转功能
• 新增 MiniMax 语音克隆。给国外用户更多选择！
• MMD 模型新增跳舞功能，现在可以在点歌台点一支舞

[次要更新]

• 新增3个内置插件，QQ，Bilibili，米家！
• 新增 Steam 状态检查，未登录时自动提示
• 修复断连重连时的会话超时问题
• MMD 模型现在可以正确看向鼠标
• VRM 模型支持更多色调，并可调整描边粗细
• 新增版本更新提醒
• 更多bug修复和细节优化

[Major Updates]

• Proactive Greetings: Neko will now proactively greet you when connecting for the first time or switching characters.
• Voice Call Initiative: Proactive conversation is now supported in Voice Call mode as well.
• Long-term Memory: A brand-new Long-term Memory system has been quietly implemented. We aren't revealing the details just yet—stay tuned for future adjustments.
• OpenClaw Cloud Integration: Added cloud OpenClaw access based on Alibaba CoPaw, supporting task routing to the cloud for execution.
• Local Virtual Machine: New local VM based on OpenFang, allowing your cat-girl to do her own thing in the background.
• Character Card Import/Export: New functionality supporting both "Settings Only" and "Full Character Card" modes.
• API Key Masking: API Keys are now displayed in a masked format for safer operation.
• API Key Book: Added a quick-fill and navigation feature for your API Key Book.
• MiniMax Voice Cloning: Added MiniMax voice cloning to provide more options for international users!
• MMD Dancing: Added a dancing feature for MMD models; you can now request a dance via the jukebox.

[Minor Updates]

• 3 built-in plugins added, Bilibili-danmaku, QQ, Mijia
• Added Steam status checks with automatic prompts when not logged in.
• Fixed session timeout issues during reconnection.
• MMD models can now correctly follow the mouse cursor with their gaze.
• VRM models now support more color tones and adjustable outline thickness.
• Added version update notifications.
• Additional bug fixes and detail optimizations.

What's Changed

• 新增当前猫娘页面背景和详情区背景变为青绿色的视觉区分效果 by @MingTianSang in #540
• 为MMD添加待机动作设置功能 by @MingTianSang in #542
• 修复VRM的模型识别问题；修复模型无法删除的问题 by @MingTianSang in #541
• 修复函数的字符验证规则,使其兼容"()" by @MingTianSang in #545
• Minimax voiceclone api support by @rophec in #528
• fix: resolve i18n labels at runtime for character settings sidepanel … by @rophec in #547
• 修复切换画质时MMD描边设置被覆盖的问题 by @MingTianSang in #548
• feat: integrate OpenFang as headless agent execution backend by @wehos in #557
• refactor: rename OpenFang display name to 虚拟机 (Virtual Machine) acros… by @wehos in #558
• bugfix(plugin) by @wislap in #549
• 重构触摸设置页面UI by @MingTianSang in #551
• refactor: API设置页重构 — 统一密钥管理 + 服务商扩展 by @wehos in #550
• fix: 音乐/meme分享默认开启，兼容新老用户 by @rophec in #553
• VRM/MMD: 支持局部跟踪模式;Live2D: 支持全屏跟踪模式 by @MingTianSang in #554
• feat(music): add volume control slider by @Himifox in #555
• feat:添加前端live2d/vrm/mmd头像获取功能 by @wislap in #556
• feat(memory): 三层记忆架构 — Facts/Reflection/Persona + prompt拆分i18n by @wehos in #563
• feat(memory): Reflection生命周期V2 — suppress全局化/冷却/软persona/定期反驳 by @wehos in #566
• 支持云端OpenClaw接入NekoClaw by @Krabbypattyl in #546
• fix: 修复模型数据中的动画属性名称，确保正确使用闲置动画 by @CN-Zephyr in #564
• 更新docker编译使用的node版本，追平主流开发需求 by @wuhins163 in #569
• 添加角色卡导出功能，并将它分为仅设定和完整角色卡 by @MingTianSang in #565
• Refactor/music UI async allowlist by @Himifox in #568
• Fix/api settings i18n ux v2 by @wehos in #571
• Add Korean, Russian, and Traditional Chinese language packs for local… by @LyaQanYi in #572
• fix(config): agent model respects enableCustomApi toggle + add 27 uni… by @wehos in #574
• Fix/memory hotswap timing by @wehos in #575
• feat(steam): 添加Steam状态检查功能及相关国际化支持 by @LyaQanYi in #577
• docs: 全面更新四语言README，新增俄语版 by @wehos in #578
• fix: entry metadata timeout 应作为上限，调用者不可绕过 by @wehos in #579
• feat(render): 帧率设置新增无限制(VSync)档位 by @wehos in #581
• 在关闭自定义API时自动从Key Book填充已保存的API Key;优化保存逻辑，仅在API Key非空时才同步到Key Book，避免… by @MingTianSang in #582
• 修复角色管理页主人档案重命名功能返回404错误 by @MingTianSang in #576
• 修复 VRM 模型路径解析错误导致无法加载内置模型的问题 by @MingTianSang in #583
• 修复保存初始VRM模型然后关闭页面时会显示模型路径为空和Live2D加载异常 by @LyaQanYi in #584
• 修复 MMD 模型切换时的状态残留问题 by @MingTianSang in #585
• feat: 添加静止自动淡化功能，优化鼠标交互体验 by @CN-Zephyr in #580
• 修复 openclawCheckbox 未定义导致的 ReferenceError by @MingTianSang in #588
• feat: 新增 OpenClaw CoPaw 通道脚本与使用手册 by @Krabbypattyl in #592
• 支持通过音频直链URL直接注册音色，无需本地上传 by @MingTianSang in #593
• 加入CCDIK by @KirisameYuumio in #596
• 模型评分系统更新 by @TL0SR2 in #598
• 文档、Github工作流、Docker编排更新 by @wuhins163 in #595
• 完成点歌台测试版 by @KirisameYuumio in #600
• fix: persona promotion 矛盾检测误判与静默丢数据 by @wehos in #602
• feat: 点歌台增强 + MMD视线追踪 + 色调映射/描边设置优化 by @wehos in #605
• feat: CosyVoice 克隆模型可配置化 by @wehos in #606
• fix: 断连重连 session 启动超时 + 重试需重启应用 by @wehos in #607
• Fix submenu overlay visibility on right-side menus by @yiyiyiyiGKY in #604
• fix: proactive chat屏幕感知独立节流 + 默认视觉间隔下调至10秒 by @wehos in #608
• refactor: 合并 Agent 路由为统一评估 + ComputerUse 鼠标轨迹光圈 by @wehos in #610
• feat: add Russian (ru) localized characters config by @wehos in #611
• feat: 首次连接/切换角色时主动搭话 + 对话间隔提示下调至30分钟 by @wehos in #612
• fix: TTS诊断日志区分"跳过(原生语音)"与真正启动 by @wehos in #615
• Feature/voice proactive chat by @wehos in #614
• Claude/flamboyant elion by @wehos in #616
• fix: 将自定义 OpenClaw 频道重命名为 neko by @Krabbypattyl in #617
• fix:Unify character name validation across services by @yiyiyiyiGKY in #618
• fix: Qwen TTS 空闲超时不再报错 + 杂项优化 by @wehos in #619
• fix: 模型管理页未保存提示全面化 — 快照比对替代零散 dirty 标记 by @wehos in #620
• feat: 版本更新日志系统 + CosyVoice 检测软化 + Jukebox/通知 UX 优化 by @wehos in #621
• Fix(mmd): reset idle animation for MMD request stage; fix eye jitter on motion switch by @Himifox in #622
• 免费声音预设名称本地化支持 by @MingTianSang in #623
• 修复VRM模型描边粗细设置在主页面不生效的问题 by @MingTianSang in #624
• fix(music): add Netease music proxy to resolve browser playback issue by @Himifox in #625
• 修复声音克隆自定义前缀未填写时的提示信息 by @MingTianSang in #627
• 将档案名长度限制从20个字符增加到60个字符 by @MingTianSang in #628
• fix(ui): 修复锁定状态管理，确保状态恢复正常 by @LyaQanYi in #630
• feat: refine proactive chat timer triggers during voice chat by @wehos in #631
• fix: reflection 重复合成 + facts/reflections 归档机制 by @wehos in #635
• fix: 更新 neko voice_id + goodbye 状态下禁止主动搭话 by @wehos in #636
• Add Bilibili danmaku by @wangjunyu200708 in #633
• 修复屏幕共享弹窗布局问题 by @MingTianSang in #637
• fix(screenshot): 优化截图逻辑 by @LyaQanYi in https://github.com/Project-...

[v0.7.2] MMD Support and Meme Spamming

[主要更新]

• 支持MMD模型，现在NEKO已经可以同步支持Live2D，VRM以及MMD了！
• 插件SDK架构升级。现在SDK相对稳定，可以被二次开发者使用了！
• 语音对话新增静音按钮。现在可以在不中断会话的情况下临时静音了。

[次要更新]

• UI改进：重构了屏幕源选择列表和按钮处理逻辑，修复了暗色模式下的背景遮挡问题；统一了设置面板的展开动画。
• 数据管理改进：为角色管理页面添加了输入框失去焦点时的自动保存机制，实现了全局对话设置与服务器的定期同步，并对缓存系统进行了性能优化。
• 技术文档与规范化：全面更新了智能体、记忆系统以及插件扩展的说明文档；在语音功能中增加了音频采样率的强制校验，并修正了插件静态资源的路径引用。
• 引入了遥测系统。现在系统会统计不同模型的token用量及缓存命中率。

[Major Updates]

• Support for MMD models. NEKO now supports Live2D, VRM, and MMD!
• Plugin SDK architecture upgrade. The SDK is now stable and ready for secondary developers!
• Added a mute button for voice conversations. You can now temporarily mute without interrupting the session.

[Minor Updates]

• UI improvements: Refactored screen source selection list and button handling; fixed background occlusion in dark mode; unified expansion animations for settings panels.
• Data management: Added auto-save on blur for character management; implemented periodic sync of global settings with the server; optimized cache system performance.
• Documentation and standardization: Fully updated documentation for the agent, memory, and plugin systems; added mandatory audio sample rate validation; corrected plugin static resource paths.
• Telemetry system: Introduced tracking for token usage and cache hit rates across different models.

What's Changed

• Floating Button Submenu Positioning and Style Fix by @yiyiyiyiGKY in #446
• Fix/memory and screen permission by @Tonnodoubt in #450
• fix(session): add expected_session lifecycle guards to prevent stale … by @wehos in #460
• 添加音频 MD5 去重功能，优化音色注册逻辑 by @LyaQanYi in #429
• fix: prevent Live2D model from disappearing when zoomed out off-screen by @yiyiyiyiGKY in #452
• 解决连续更改live2d画质会导致模型消失的问题 by @MingTianSang in #455
• fix(css): 优化GPU合成层隔离，减少重绘对canvas的影响(neko-pc #15) by @LyaQanYi in #459
• fix(showLive2d): 避免模型淡入动画导致的闪烁，优化可见性检查 by @LyaQanYi in #457
• refactor: extract music-ui from common-ui and reset positioning by @Himifox in #458
• Refactor/decompose app js by @wehos in #466
• Move translation box by @qiguang113 in #454
• Fix(music): prevent music from refreshing while playing by @Himifox in #463
• add skip button and enhance tutorial reset button with disabled state by @yiyiyiyiGKY in #464
• 优化手机界面UI;修复手机网页无法点击翻译按钮的问题 by @MingTianSang in #465
• feat(port): 添加从 Electron userData 目录读取端口覆盖配置（neko-pc #16） by @LyaQanYi in #467
• fix(uploadItem): 移除上传时解锁成就的逻辑，改为在有订阅物品时解锁创意工坊成就 by @LyaQanYi in #468
• fix(cfa): Windows反勒索防护(CFA)下Live2D模型加载失败的问题 by @wehos in #469
• feat(websocket): 添加翻译功能以处理状态消息并更新消息显示 by @LyaQanYi in #470
• fix: handle edge cases in player logic to prevent crashes by @Himifox in #471
• 将Agent模式触发顺序改为插件系统，网页，鼠标控制 by @MingTianSang in #472
• feat(config): 添加 assistApiKeyKimi 字段到核心配置 API by @LyaQanYi in #474
• Update README.MD by @WUHINS in #476
• feat: Agent结果解析器 + LLM上下文防污染 + i18n全覆盖 by @wehos in #475
• 完成live2d自定义触摸动画设置 by @KirisameYuumio in #478
• fix: adjust selection rules for proactive music recommendations by @Himifox in #480
• feat: 拦截外部链接，使用系统默认浏览器打开 by @LyaQanYi in #481
• 为Docker部署访问添加SSL by @WUHINS in #479
• feat: 更新任务HUD逻辑，优化任务卡片更新和样式，增强暗色模式支持 by @LyaQanYi in #482
• 猫娘名称修改时可以同步修改记忆名称；防止猫娘名字中出现"."符号导致无法保存记忆；优化主人与猫娘档案键值显示 by @MingTianSang in #484
• feat: token telemetry, staggered launcher startup & character naming fixes by @wehos in #486
• Refactor proxy headers and add project address echo by @WUHINS in #488
• chore: remove memory/semantic and memory/router, drop semantic_store by @wehos in #489
• fix(music): prevent AI from repeating song info and improve search diversity by @Himifox in #490
• 移除触摸区自定义动画 by @KirisameYuumio in #491
• feat: 添加 VRM 模型扫描和删除功能，增强用户模型管理 by @LyaQanYi in #492
• fix(ui): close floating button popups when clicking outside by @yiyiyiyiGKY in #493
• fix(achievement): unify Steam achievement system and fix triggering bugs by @yiyiyiyiGKY in #494
• 优化部分报错的提示让它更加直观 更易理解 by @MingTianSang in #496
• 修复API相关的报错(如免费API时间耗尽、API欠费)无法正确显示的问题 by @MingTianSang in #497
• fix: prevent proactive cascade from crashing system and isolate plugi… by @wehos in #498
• exe模式吞时间戳与蝴蝶结问题 by @KirisameYuumio in #495
• fix: TTS Worker 崩溃后自动重试，防止语音永久静默 by @wehos in #500
• 修复与重加回自定义触摸动画 by @KirisameYuumio in #499
• fix: Fix skip button click hotspot issue in Steam Overlay environment by @yiyiyiyiGKY in #501
• 修复与重加回自定义触摸动画 by @KirisameYuumio in #502
• 现在情感配置标题始终固定在页面顶部关闭按钮随时可见 by @MingTianSang in #503
• fix(tutorial): fix overlay border on resize and skip button unclickable on desktop by @yiyiyiyiGKY in #505
• 修复部分页面因标题而出现的问题（下拉框层叠问题；情感配置间距问题） by @MingTianSang in #509
• 禁用部分浏览器快捷键防止出现意外情况 by @MingTianSang in #508
• Fix(cookies login page): replace dom purify cdn with local file for packaged app compatibility by @Himifox in #506
• fix: 解除猫娘声音注册时处理会话结束和前端通知 by @LyaQanYi in #511
• bugfix:linux port issue. by @wislap in #512
• 为角色管理页面添加输入框失去焦点时的自动保存机制 by @MingTianSang in #513
• refactor(plugin): introduce async-first sdk_v2 with unified error model and stricter developer facade by @wislap in #456
• 将主页面的角色设置在鼠标悬浮在上面时向下展开的机制改为和对话设置画质设置主动搭话`自主视觉一样的展开方式 by @MingTianSang in #517
• feat(voice-clone): normalize audio sample rate and reject files below 16kHz by @rophec in #473
• fix(plugin): 删除无效 metrics 并修正静态资源路径 by @Himifox in #520
• MMD模型支持，顺便合并了一手common-ui代码 by @LyaQanYi in #510
• fix: 重新应用PR#517角色设置侧边栏改动 by @wehos in #523
• 在语音控制按钮左侧添加静音麦克风按钮 by @MingTianSang in #519
• bugfix : mcp_adapter by @wislap in #522
• 修复"请他离开"后再回来时静音按钮错误显示的问题 by @MingTianSang in #524
• 添加MMD模型说话时口型同步功能 by @MingTianSang in #525
• feat(settings): 添加全局对话设置的加载和保存功能，并实现定期同步到服务器 by @LyaQanYi in #521
• feat(mcp-adapter): expose MCP tool results to LLM;feat(plugin):improve result system by @wislap in #526
• feat(proactive): implement meme proactive chat with regional separation and proxy support by @Himifox in #516
• 将VMD动画上传功能从下拉选择器分离为独立按钮;补充VMD动画相关的所有本地化字符串;新增删除VMD动画按钮 by @MingTianSang in #530
• feat(model-manager): 支持在 Live3D 模式下加载 MMD 模型 by @LyaQanYi in #532
• 修复暗色模式下保存按钮区域背景遮挡内容的问题;暗色UI优化 by @MingTianSang in #531
• 无色调映射时禁用曝光滑块防止误导 by @MingTianSang in #534
• 修复角色卡模型未设置时预览环形文字不更新的问题 by @MingTianSang in #535
• Feat/cache optimization 3 by @Himifox in #533
• 修复MMD物理模拟开关关闭后进入角色外形时物理效果仍然存在的问题 by @MingTianSang in #536
• Fix/mmd physics and misc refactor by @wehos in #538

New Contributors

• @qiguang113 made their first contribution in #454

**Full Changelog...

[v0.7.1] NekoClaw & Music Share

[主要更新]
Agent功能重置为NekoClaw，自定义插件上线，用户现在可以通过插件接入各种功能！
第一批官方内置自定义插件为：定时器（支持闹钟、倒计时、循环提醒）和网络搜索。
主动搭话中新增了发送音乐的功能，猫娘现在可以分享音乐啦！
创意工坊界面重做且更加稳定。
自定义语音更新至CosyVoice 3.5。

[次要更新]
优化了VRM模型的拖拽判定逻辑，有利于稳定帧率。
强制禁用了本地连接的http代理，防止代理软件误伤本地连接。
对角色设置页UI各项细节进行了优化，加入一个小彩蛋。
对前后端直接连接的稳定性进行了增强，避免断连。
优化了角色的退场动画。
对错误提示的前端展示做出了一系列优化。

[Major Updates]

Agent functionality has been reset to NekoClaw, and custom plugins are now live — users can now integrate various features through plugins!
The first batch of official built-in plugins includes: Timer (supports alarms, countdowns, and recurring reminders) and Web Search.
A music-sharing feature has been added to the proactive chat function — the cat girl can now share music!
The Creative Workshop UI has been redesigned and is now more stable.
Custom voice has been updated to CosyVoice 3.5.

[Minor Updates]

Optimized the drag detection logic for VRM models to help maintain a stable frame rate.
Forcibly disabled HTTP proxies on local connections to prevent proxy software from interfering with local traffic.
Various UI details on the character settings page have been refined, along with a small easter egg.
Improved the stability of direct front-end to back-end connections to prevent disconnections.
Optimized the character exit animation.
Made a series of improvements to how error messages are displayed on the front end.

What's Changed

• feat: add Russian (ru) language support across frontend and backend by @wehos in #349
• 增加了对多个模型进行评分的代码系统 by @TL0SR2 in #337
• fix: resolve missing bilibili-api files after packaging by @Himifox in #346
• 切换画质时保存当前模型的大小和位置，防止切换后大小位置被重置 by @Chiya-mt in #347
• fix: harden browser session lifecycle and polish chat UI by @wehos in #350
• Feat/ci cross platform build by @wehos in #353
• Feat/ci cross platform build by @wehos in #354
• Feat/ci cross platform build - add browser use deps by @wehos in #355
• Feat/ci cross platform build - manually copy playwright_browsers on macOS by @wehos in #357
• chore: set timeout-minutes to 300 for Nuitka build, 180 for Electron by @wehos in #358
• 更改live2D的初始位置防止其加载到屏幕外 by @Chiya-mt in #360
• 聊天框收缩和展开动画期间按钮将无法被再次点击 by @Chiya-mt in #361
• Feat/ci cross platform build by @wehos in #362
• feat: implement UGC cache warmup and character card synchronization on server startup by @LyaQanYi in #356
• add GPT-SoViTS voice id adaptation by @wislap in #344
• 增强 Live2D 交互模块的文档说明 by @Himifox in #363
• feat: fix VRM and live2d initialization issue by @wehos in #365
• Feat/ci cross platform build by @wehos in #364
• Feat/avatar popup shared behavior by @wehos in #370
• 修复没有订阅物品时“您还没有订阅任何创意工坊物品”i18n不立即应用的问题 by @Chiya-mt in #369
• feat: add audio silence remover with unit tests by @LyaQanYi in #367
• feat(cookies_login): add i18n, optimize UI, fix memory leak and adapt mobile by @Himifox in #366
• 完成api保存按钮的位置优化 by @KirisameYuumio in #368
• 解决调整画质导致的语音按钮与屏幕共享按钮状态重置的问题;添加鼠标跟踪开关 by @Chiya-mt in #371
• 修复创意工坊文件与我的文档不一致问题，尝试修复SSL加载失败问题、增加SSL预检，统一保留字过滤 by @wehos in #375
• Refactor character data handling to support new reserved field structure by @wehos in #376
• feat: 增强滑块功能，添加音色设定文本 by @LyaQanYi in #377
• 解决模型设置无法保存的问题 by @Chiya-mt in #378
• wip: prompts_sys, core, system_router, memory/semantic, memory_server by @wehos in #379
• 修复live2d模型在模型设置页无法被保存的问题 by @Chiya-mt in #381
• 添加一个弹窗提示，当猫娘名称出现“/”等字符时发出提示 by @Chiya-mt in #382
• Wehos/wip live2d utils by @wehos in #383
• 添加对steam创意工坊目录的扫描防止模型管理页面找不到模型 by @Chiya-mt in #384
• Persist Steam workshop path as startup fallback by @wehos in #387
• 优化模型加载过程。 by @LyaQanYi in #386
• Fix TTS interrupt handling to prevent stale audio leakage and avoid m… by @wehos in #388
• feat(tts): improve audio chunk handling and speech id management by @wislap in #389
• 增强平滑恢复功能，优化动作和表情的过渡处理，确保视觉一致性 by @LyaQanYi in #390
• 如果配置了阿里API并且为当时注册音色时使用的也可以在列表中选择对应音色；修改音色选项显示内容 by @Chiya-mt in #392
• (可选)为模型管理界面中的steam创意工坊模型添加标识 by @Chiya-mt in #393
• Pr182 by @wislap in #380
• 为模型设置部分按钮添加最大显示的字数限制；鼠标悬浮时显示为选中的模型与表情动作名称 by @Chiya-mt in #394
• 新建猫娘与确定猫娘时使页面滚动让新建的猫娘在页面中心 by @Chiya-mt in #395
• feat(proxy): 添加运行时热切换代理模式的API by @LyaQanYi in #397
• 修复自动收起小三角后导致语音或视觉按钮重置的问题 by @Chiya-mt in #398
• fix(vision): cache proactive vision screen capture stream to avoid re… by @Tonnodoubt in #399
• feat(browser): 添加浏览器控制功能及相关逻辑 by @LyaQanYi in #402
• update plugin system by @wislap in #404
• 修正Browser首次启动卡顿的问题 by @rophec in #401
• 优化对话框的展开与收缩时的动画 by @Chiya-mt in #405
• 对捏脸界面UI进行大调整优化整体操作舒适度；添加小彩蛋 by @Chiya-mt in #406
• 美化VRM和live2d表情管理中的下拉表单；为下拉表单添加有滚动条时右侧变为直角的机制 by @Chiya-mt in #407
• 为创意工坊添加兜底机制防止偶发性的模型管理页下拉菜单不更新 by @Chiya-mt in #411
• fix: ensureWebSocketOpen prevents 'WebSocket未连接' on auto-reconnect by @wehos in #412
• 优化看板娘和VRM容器的最小化状态，使用平滑滑出效果替代直接移出屏幕 by @LyaQanYi in #413
• 修改角色管理页选择VRM模型后的显示使得更加简洁 by @Chiya-mt in #414
• 修正情感管理步骤中的模型选择元素ID，并添加跳过自动显示的选项 by @LyaQanYi in #415
• Quota by @LyaQanYi in #418
• refactor:plugin server by @wislap in #410
• 内网 httpx 禁用代理, TTS config (cosyvoice 3.5 update, international user detect), UI (important notice) by @wehos in #428
• Redesign Steam Workshop Manager UI and Unify Theme Styling by @yiyiyiyiGKY in #419
• 优化VRM canvas的渐入动画，改用CSS transition替代setInterval，提升性能和流畅度；调整浮动按钮的可见性恢复… by @LyaQanYi in #423
• 优化角色管理页面UI并添加隐藏猫娘功能 by @Chiya-mt in #424
• 优化api key设置中 保存成功或失败 的提示位置 by @KirisameYuumio in #425
• refactor: 更新错误消息以使用统一的代码和描述 by @LyaQanYi in #427
• feat: 新增 APlayer 音乐推荐体系与个人动态搭话完善，修复搭话自检 Bug by @Himifox in #416
• 增加媒体凭证扫描登录模块 by @KirisameYuumio in #426
• 优化CSS和JavaScript中的文本标签，统一格式和样式(neko-pc #14) by @LyaQanYi in #432
• fix(build): correct SteamworksPy.dylib filename for macOS packaging by @Tonnodoubt in #433
• feat(logging): add brace-style format compatibility for logging by @wislap in #434
• fix: 修正用户插件仪表板重定向路径 by @LyaQanYi in #435
• SteamWorkShop ui optimization by @yiyiyiyiGKY in #436
• fix: regenerate speech_id before proactive TTS to avoid CosyVoice con… by @Tonnodoubt in #437
• fix: correct sys.path level for plugin module imports by @Tonnodoubt in #438
• fix: 防止重复发送 TTS 完成信号和提交缓冲区 by @LyaQanYi in #439
• feat(task-hud): deferred completion, task restore, timeout, and thread-safe cache by @Tonnodoubt in #440
• 情感配置页面下拉菜单如果会超过页面底部那么将向上展开 by @Chiya-mt in #442
• refactor(plugin): split built-in and user plugin roots by @wehos in https://...

[v0.7.0] Computer/Browser Use, GPT-SoVITS

[主要更新]

1. Agent 模块大幅重构，更加稳定。
2. 现在猫娘不仅可以操作键盘、鼠标，还可以通过主动打开并用脚本控制浏览器。
3. 主动搭话可以选择信息源。
4. 新增画质调节面板，支持调节 VRM/Live2D 的帧率与画质。
5. 现在支持接入 GPT-SoVITS 语音，全界面韩语实装。
6. 界面增加深色模式。

[次要更新]

1. 部分猫娘消息框位置调整，不再遮挡住视线。
2. VRM模型现在支持头部追踪。
3. 界面新增韩语支持。
4. 千问更新至qwen3.5，辅助API额外增加kimi支持。

还有更多更新内容请参考commit记录。

[Major Updates]

1. Agent Module Overhaul: Significant refactoring for greatly improved stability.
2. Advanced Interaction: The AI (Neko) can now perform keyboard and mouse operations and autonomously launch and control browsers via scripting.
3. Proactive Engagement: Added the ability to select specific information sources for proactive "small talk" or interactions.
4. Graphics Control Panel: New settings to adjust frame rates and rendering quality for VRM and Live2D models.
5. New Voice & Language Support: Integrated GPT-SoVITS audio support and implemented full UI localization for Korean.
6. Dark Mode: Added a native Dark Mode toggle for the entire interface.

[Minor Updates]

1. UI Optimization: Adjusted the position of certain message boxes to ensure they no longer obstruct the view.
2. Head Tracking: VRM models now support real-time head tracking.
3. Language Expansion: Added Korean language support to the interface.
4. LLM Upgrades: Updated Qwen to qwen3.5; added Kimi support for auxiliary APIs.

For a full list of changes, please refer to the latest commit history.

What's Changed

• Complete ko translation i18n files by @Tonnodoubt in #302
• 修复角色管理页面和声音克隆界面可以多开的问题 by @Chiya-mt in #304
• feat: add sidebar white panel treatment and prevent button text clipping by @Tonnodoubt in #305
• feat: 为初始角色的默认档案内容值和键名提供i18n by @CN-QanYi in #308
• Add a system tray note to the home tutorial and simplify the tray preview by @Tonnodoubt in #303
• 加了两个issue模板，方便分类issue中的需求或bug by @CO2mega in #311
• utilize bilibili-api-python library for bilibili scraping by @Bill-Haoyu-Lin in #309
• feat: Reduce keyword repetition in memory summaries with pronoun subs… by @Tonnodoubt in #310
• 修复在教程期间请她离开、拖动角色之类的可能造成的问题 by @wehos in #314
• Fix voice chat state reset issue after session timeout by @Tonnodoubt in #315
• Let the expand icon same as back icon by @Tonnodoubt in #316
• 修复屏幕缩放不同对模型位置和UI大小的影响 by @killerppo in #317
• feat: 添加扬声器音量控制功能 by @killerppo in #312
• feat: truncate long text chats and sync rewritten memory by @killerppo in #306
• Add GPT-SoVITS support by @wislap in #298
• Add floating hover effect to persistent expression list items by @Tonnodoubt in #318
• 为音色注册页面添加声音预览功能 by @Chiya-mt in #319
• fix:voice disconnect errors recovery in onclose by @Tonnodoubt in #322
• fix i18n state & add i18n skill by @Tonnodoubt in #324
• 修改记忆后及时刷新获取最新的数据防止出现删除后上下文还在的情况 by @Chiya-mt in #323
• Feature vrm头部跟踪，待机动作控件，修改了默认光照参数，添加了一些新的vrma文件。 by @zyumo777 in #325
• feat: Implement VRM emotion configuration system with i18n support by @Tonnodoubt in #326
• fix：动态检测 VRM 模型前方向，兼容 0.x 和 1.0 VRM版本的头部跟踪 by @zyumo777 in #327
• 捏脸界面选择模型后防止i18n翻译覆盖模型名称 by @Chiya-mt in #328
• 修复语音按钮与屏幕共享按钮右边的三角展开按钮在主页教程期间可以被点击的问题 by @Chiya-mt in #329
• fix: resolve voice mode button sync and session state issues on auto mic close by @Tonnodoubt in #331
• 对主动搭话功能进行拆分，可以随意组合视觉搭话、新闻搭话和视频搭话 by @Chiya-mt in #330
• 完成分离文本对话api与纠错api by @KirisameYuumio in #321
• feat: add dark mode support across multiple templates and implement theme manager by @CN-QanYi in #332
• feat(web_scraper): 新增自主搭话中B站和微博个人动态内容支持 by @Himifox in #333
• 测试了对话评分系统，在控制台中增加了对简单笑话和图片识别过程的print显示 by @TL0SR2 in #334
• 解决角色管理中API设置入口按钮文本闪烁的问题 by @Chiya-mt in #335
• 添加健康检查端点和端口工具，支持服务健康检测与启动锁机制 by @CN-QanYi in #336
• Fix Bilibili dynamic mobile UA mismatch and error log leak by @Himifox in #338
• docs: add VitePress developer documentation site (62 pages) by @wehos in #340
• docs: add VitePress documentation site with i18n (en/zh-CN/ja) by @wehos in #341
• fix(docs): change VitePress base to '/' for custom domain by @wehos in #343
• 解决VRM的悬浮按钮UI无法显示开关状态的问题 by @Chiya-mt in #345

New Contributors

• @CO2mega made their first contribution in #311
• @Bill-Haoyu-Lin made their first contribution in #309
• @killerppo made their first contribution in #317
• @zyumo777 made their first contribution in #325
• @KirisameYuumio made their first contribution in #321
• @Himifox made their first contribution in #333

Full Changelog: v0.6.3...v0.7.0

[v0.6.3] Gemini, VRM, UI Update

中文Major Change log

更新内容太多，只挑一些重点

1. UI再次大幅度优化，尤其是设置页面。
2. 支持Gemini Live API及Gemini系列模型。
3. 支持VRM (3D) 模型
4. 完善i18n，尤其是日语支持
5. 新增面向新用户的内置Tutorial

English Major Change Log

There are quite a few updates, but here are the highlights:

1. Major UI Overhaul: Significant design improvements, particularly within the Settings page.
2. Expanded Model Support: Now supports the Gemini model series and the Gemini Live API.
3. VRM Support: Added compatibility for 3D (VRM) models.
4. Enhanced i18n: Improved internationalization, with a specific focus on Japanese language support.
5. New User Onboarding: Added a built-in tutorial designed for new users.

Full Changelog: v0.6.2...v0.6.3

[v0.6.2] 本地语音和全局快捷键！

中文Change log

[主要更新]

• 新增本地 TTS 语音支持。
• 创意工坊角色卡底层重构。
• 新增文本拟真输出模式，提供更自然的交互体验。
• 新增全局快捷键功能及专为直播设计的主播模式。
• 屏幕分享功能升级，现在支持选择特定窗口进行分享。

[次要更新]

• 支持国际化热门信息爬取。
• 优化 Live2D 交互逻辑：在锁定状态下按住 Ctrl 键可实现点击穿透。

[Bug修复]

• 修复了“粘手”异常 Bug。
• 修复了模型显示重叠（影分身）的 Bug。
• 修复了切换麦克风设备时可能导致的失效问题。

English Change Log

[Major Updates]

• Added support for local TTS (Text-to-Speech).
• Refactored the Steam Workshop character card system.
• Introduced a new Realistic Output Mode for more natural interactions.
• Added global hotkeys and a dedicated Streamer Mode.
• Enhanced screen sharing: You can now select specific windows for sharing.

[Minor Updates]

• Added support for international trending information crawling.
• Optimized Live2D interaction: Holding the Ctrl key in locked state now allows for click-through.

[Bug Fixes]

• Fixed the "sticky hand" (cursor dragging) bug.
• Fixed the "shadow clone" (duplicate model rendering) bug.
• Fixed issues related to microphone device switching.

New Contributors

• @rophec made their first contribution in #192
• @RechardLLee made their first contribution in #224

Full Changelog: v0.6.1...v0.6.2

[v0.6.1] N.E.K.O. (Macbook Support)

中文Change log

[主要更新]

• 支持多屏幕。
• 新增Macbook支持。
• 新增翻译字幕。当对话语言与系统语言不同时，弹出翻译提示。
• 移除底层傲娇属性。

[次要更新]

• 更好的初始化逻辑，降低首轮语音加载时间。
• 更好的热重置逻辑，降低热重置期间不回应语音的概率。
• 更好的TTS打断逻辑。不会再遗漏句子。
• 自定义语音（Cosyvoice）现在会自动预热。

[Bug修复]

• 修复了Live2D参数编辑器返回问题。
• 修复了内部更新失败提示。
• 修复了某些情况下语音模式开启主动视觉无法触发主动截图的问题。

English Change Log

[Major Updates]

• Added support for multiple displays.
• Added MacBook support.
• Added translated subtitles. When the conversation language differs from the system language, a translation prompt will pop up.
• Removed the systemetic Tsundere personality.

[Minor Updates]

• Improved initialization logic to reduce first-round voice loading time.
• Improved hot reset logic to reduce the chance of no voice response during hot resets.
• Improved TTS interruption handling; sentences will no longer be skipped.
• Custom voices (CosyVoice) are now automatically pre-warmed.

[Bug Fixes]

• Fixed the return issue in the Live2D parameter editor.
• Fixed incorrect internal update failure notifications.
• Fixed an issue where proactive screenshots could not be triggered in certain cases when voice mode and proactive vision were enabled.

New Contributors

• @coderabbitai[bot] made their first contribution in #199
• @wsqstar made their first contribution in #195
• @TL0SR2 made their first contribution in #211

Full Changelog: v0.6.0...v0.6.1

[v0.6.0] N.E.K.O. (Proactive Era)

中文Change log

[主要更新]

• 新增主动视觉功能（默认关闭）。猫娘会在主动搭话时获取屏幕信息，或在语音对话中时不时窥屏。
• 更加完善的自定义API支持。现在你可以使用自定义API来自由搭配服务商。你可以使用免费API搭配阿里的TTS语音。
• 新增基于循环神经网络的主动降噪系统，有助于提升语音识别表现。

[次要更新]

• 更新了阿里对话模型Qwen3-Omni-1201的默认参数，缓解复读问题。
• 文本模式下新增了一系列复读检测，触发时会自动清除复读对话并提醒用户对策。
• 新增插件系统（处于实验状态）。
• 语音克隆时可以选择参考语音的语言了。
• 使用Ogg Opus而非PCM传输TTS数据，显著降低带宽压力。
• 使免费/Step模型的文本模式下同样支持Web Search。
• Docker容器新增nginx反向代理。

[Bug修复]

• 修复了UI界面中一系列点击事件和状态重置问题。
• 当猫娘被打断时，不会再出现重复的记忆块。
• 麦克风静音检测不再计算语音系统准备期间的时间。

English Change Log

[Major Updates]

• Added Active Vision feature (disabled by default). The catgirl can capture screen information when initiating conversations, or occasionally peek at the screen during voice chats.
• Improved Custom API support. You can now freely mix and match service providers using custom APIs. For example, you can pair a free API with Alibaba’s TTS service.
• Introduced an active noise reduction system based on recurrent neural networks, helping to improve speech recognition performance.

[Minor Updates]

• Updated the default parameters for Alibaba’s dialog model Qwen3-Omni-1201, reducing repetition issues.
• Added a series of repetition-detection mechanisms in text mode. When triggered, repeated dialogue will be automatically cleared and users will be notified with suggested countermeasures.
• Added a plugin system (experimental).
• Voice cloning now allows selecting the language of the reference audio.
• Switched TTS data transmission from PCM to Ogg Opus, significantly reducing bandwidth usage.
• Enabled Web Search in text mode for free/Step models as well.
• Added an nginx reverse proxy to the Docker container.

[Bug Fixes]

• Fixed a number of click-event and state reset issues in the UI.
• Prevented duplicate memory blocks from appearing when the catgirl is interrupted.
• Microphone mute detection no longer counts the system’s voice preparation time.

What's Changed

• Docker封装，部署和启动流程的优化。Docker部署文档更新 by @WUHINS in #157
• fix: prevent click-through in advanced settings and improve UX by @Tonnodoubt in #158
• 修复免费版本 UI 显示和保存后状态显示错误 by @Chiya-mt in #159
• Added noise reduce funcionality with RNNoise. by @wehos in #168
• 添加可以根据屏幕内容主动搭话的功能 by @Chiya-mt in #160
• fix: 在加载新模型和刷新页面时关闭设置窗口 by @CN-QanYi in #165
• Add user-facing plugin subsystem (plugin server, registry, runtime state) by @wislap in #161
• Added repetition detection and notification. by @wehos in #170
• Support custom API. Screenshot utils unified usage. by @wehos in #171
• Refactor main_server. Adjust build profiles. by @wehos in #173
• 解决了自定义API设置视觉模型后模型选择错误的问题 by @Chiya-mt in #174
• feat(交互): 在拖拽模型时临时禁用按钮事件拦截 by @CN-Zephyr in #175
• feat(主动对话):基于标题搜索的主动对话功能 & 将热门内容改为首页推荐并更新相关逻辑 by @CN-QanYi in #176
• 新增语音聊天时如果开启了主动视觉则每十秒共享一次屏幕（1帧画面） by @Chiya-mt in #179
• feat(系统路由): 添加AI回复内容验证规则 by @CN-Zephyr in #181
• Enhance denoise and AGC quality for better VAD. by @wehos in #184
• 修复语音或屏幕共享按钮状态异常的问题 by @Chiya-mt in #189

Full Changelog: 0.5.5...v0.6.0