
StateSync | Real-Time State Persistence for AI Agents
Prevent context loss and AI agent amnesia. StateSync provides real-time state persistence and agent handoffs over MCP.
@StateSync_ · X
The full gallery
Tech stack
16 projects

Prevent context loss and AI agent amnesia. StateSync provides real-time state persistence and agent handoffs over MCP.
@StateSync_ · X

Wager digital assets in fast, skill-based typing matches. Real-time competition and onchain rewards turn every keystroke into a spectator-ready earning experience.
@weston_union · X

Animated thinking orbs component for AI interfaces with six states and automatic theming.
@MengTo · X
These will instantly make your vibe-coded projects look less generic. Free, open-source libraries from the same creator: • • •

Tools for tuning trading card game decks and pricing singles for play.
@BinderCurve · X

Make music with webcam hand gestures—chords, harmony, theremin pitch, no signup needed.
lisheng3698 · V2EX
我把摄像头变成了乐器:从手势识别到 Web Audio,分享 5 个真正踩过的坑 最近做了一个有点不一样的浏览器小项目:用摄像头识别双手动作,再把手的位置和姿势映射成和弦、音高、音量和音色。 不需要安装软件,也不需要麦克风。打开网页、允许摄像头后,就可以隔空演奏。 项目地址:[GestureSynth.me]( https://gesturesynth.me/) 建议使用最新版 Chrome 或 Edge 。摄像头画面只在本地处理,不会上传。 ## 真正难的不是“识别到手” 最开始我以为,只要 MediaPipe 能返回手部关键点,剩下的事情就是把坐标转换成音符。 实际做下来发现:**识别出一只手很容易,让它变成一个“能演奏”的控制器才难。** 摄像头每一帧都会返回一组手部关键点,但这些坐标会轻微抖动,偶尔还会丢失。如果直接把原始坐标连接到音高或者音量,听起来会非常不稳定。 最后形成的处理链路大致是: ```text 摄像头 → 手部关键点识别 → 提取手掌位置、距离和手势 → 平滑、死区和状态保持 → 映

React components for building AI agent user interfaces with streaming, tool approvals, and cross-platform deployment.
@sameerpallav · X

Learn American Sign Language with interactive 3D hands, structured lessons, and camera practice.
@EntangleITcom · X
I vibe coded an American Sign Language tutor app:

Explore guitar scales, chords, and intervals on an interactive fretboard.
@PLaclede62953 · X
I’m a full-time developer, father, hobby guitarist, and regular church musician. FretFlow is the practice tool I wanted: quick fretboard visualization without jumping between tabs, charts, and theory sites. #GuitarPractice #WorshipGuitar #BuildInPublic

Practice whistling with real-time pitch feedback and progressive exercises.
x1shir · V2EX
HowToWhistle:像唱 KTV 一样练习吹口哨 一直想提高吹口哨技巧,搜了一圈发现网上的教程都是文字和视频,练的时候根本不知道自己吹得对不对。 想到可以像 KTV 一样,有个音轨对比,于是干脆自己搓了一个练习网站:[https://howtowhistle.org/]( https://howtowhistle.org/) 核心思路是用浏览器的麦克风做实时音高检测,你对着麦克风吹,页面会显示你当前的音高和目标音的偏差,类似 KTV 唱歌时的实时评分。功能大概有: - 实时音高反馈,能看到自己吹的是哪个音、偏高还是偏低 - 从单音长音开始的渐进式练习,到音阶、再到简单旋律 - 纯前端实现,声音不上传,无需注册 目前还比较简陋,欢迎大家试玩和提建议。

SMTRY builds sovereign AI infrastructure: the memory layer for AI you own. Anamnesis gives your AI persistent, encrypted memory.
@SMTRYcorp · X

Paste a playlist link, get a distraction free course with progress tracking.
Sekhmet · HN
Adept – convert any YouTube playlist into a Udemy-like course

Translate locale files with context and tone awareness for games and software.
@polycontext · X
Context-aware localization for apps, games, and software.