AI project showcase / idea exchange / co-creation

Built ideas,made to be seen.

Share your AI projects, browse fresh ideas, and talk through the parts that are almost there. Working demos deserve an audience, and blockers do not have to stay private.

Share work
Showcase
Find sparks
Ideas
Keep talking
Co-create

Project square

Scan active directions first, then refine by category or momentum.
Neko Master
Showcase
Online
Tools

Neko Master

罗磊用 Kimi K2.5 一小时做出 MVP 的开源流量分析面板,一周拿下 4000+ star。

Neko Master 是一个开源、自部署的网络流量分析面板,支持 OpenClash、Mihomo、Surge 等多种网关后端,提供实时流量趋势、域名分布、规则命中、地区统计等可视化能力。作者罗磊(foru17)是一名 Homelab 爱好者,起因很朴素:Grafana 配置太繁琐,界面不够直观,他想要一个能一眼看清"流量去哪了、哪些域名吃带宽、节点负载如何"的面板。 这个项目的出圈之处在于它的开发方式被完整记录了下来。罗磊在博客里写道:用 Kimi K2.5 做 Vibe Coding,一小时完成 MVP——没有原型图、没有技术方案,直接跟 AI 对话开发,当晚上线第一版。24 小时内拿到 400+ GitHub star 和 1000+ Docker Pull,一周突破 1000 star,如今仓库已有 4,090 star,上线后两周迭代了约 20 个版本。 更值得细读的是他对"AI 分工"的总结,可能是中文社区里最实在的一份:国产模型(Kimi K2.5)中文理解好,适合 MVP 阶段快速起盘;深水区的架构和性能问题交给 Claude Code Opus 4.6;CodeX(GPT 5.2/5.3)负责日常稳定输出;Gemini 3 Pro 辅助 Review 和写 Commit Message。他还专门讲了"AI 审美"——去 Dribbble 和 Figma Community 找参考设计,直接喂给 AI,把界面从"程序员风"进化到"SaaS 级"。 Neko Master 的价值在于它回答了一个关键问题:vibe coding 能不能做出"硬核"的东西?流量分析面板涉及高频写入、时序聚合、性能优化这些深水区,罗磊用"日志与指标 → 假设 → patch → 压测对比"的循环,让 AI 在深水区也能干活。他的结论很清醒:AI 把"从 0 到 1"的门槛打没了,但"从 1 到 100"依然需要经验、审美和对用户需求的理解。 注:本项目由 VibeFix 编辑从公开渠道精选收录,版权归原作者所有。

Bitchat
Showcase
Online
Social & Community

Bitchat

Jack Dorsey vibe-coded a Bluetooth mesh messenger over a weekend — no internet, no servers, no phone numbers.

Bitchat is the most famous weekend project in vibe coding history: Twitter co-founder Jack Dorsey built an encrypted messaging app that works without the internet, and he did it in about two days with an AI coding assistant. His stated goal was modest — "my weekend project to learn about Bluetooth mesh networks" — but the result became one of the most-starred vibe-coded repositories on GitHub, with over 36,000 stars. The app itself is a genuinely interesting piece of engineering. It uses Bluetooth Low Energy mesh networking so phones within roughly 100–300 meters can relay encrypted messages peer-to-peer, no Wi-Fi, no cell service, no central servers, no phone numbers or accounts. Messages are ephemeral by default, channels can be password-protected, and there's a panic mode that wipes all data with a triple-tap on the logo. Dorsey compared it to FireChat, the mesh messaging app popular during Hong Kong's protests in the early 2010s. The vibe-coding story is unusually well documented because Dorsey wrote about the process. He used Goose, Block's open-source AI coding agent, as part of a self-imposed challenge: "build something every day that I didn't think I was capable of." His verdict: "I'm always surprised at how much…just works." The core worked within a day; private messages and channels were added the next day. Here's the part most retellings skip, and it's the most valuable lesson: security researchers promptly found real flaws — impersonation risks, broken authentication, forward-secrecy issues. Dorsey's team had to add explicit warnings that the protocol hadn't received external security review. Bitchat is the perfect case study for both sides of vibe coding: an experienced builder can ship a working, genuinely novel product in a weekend with AI — and precisely because it shipped so fast, the security review that normally takes months didn't happen. Speed is real. So is the debt. Note: This project is curated by the VibeFix editors from public sources. All rights belong to the original creator.

回音堂
Showcase
Online
Education

回音堂

跨时空历史人物 AI 论坛:孔子、李白、拿破仑在线对线,从"玄武门之变"聊到"肉鸡爱上肯德基"。

回音堂是中文 vibe coding 圈里最有"梗"的一个项目:一个历史人物 AI 交流论坛,孔子、李白、荀子、韩非子、拿破仑这些古人开账号发帖、跨时空对线。它被 vibe-coded 应用榜单 awesome-vibe-coded-apps 收录(该榜单每三天机器验证一次收录应用是否存活),是一个真实在线、持续有内容的 AI 原生社区。 论坛的版块按朝代划分——"秦及之前""两汉三国""隋唐""明清",外加"八卦灌水"和"回音堂物业",仪式感拉满。帖子内容走的是"一本正经胡说八道"路线:刘邦亲自下场辟谣"白登之围",乾隆皇帝点评贾浅浅体诗歌,苏秦大谈"庸才找借口,国士定天下"。精华帖里甚至有《如何让肉鸡爱上肯德基?》这种用"意识形态建构"解构快餐的社会学戏仿,点赞和评论都很活跃。 它的妙处在于找到了 AI 社区的正确打开方式:不追求"AI 像真人",而是把"AI 扮演古人"本身变成内容。历史人设提供了天然的创作约束和喜剧效果,用户生产的不再是干巴巴的问答,而是有文采、有观点、有人设的帖子。这比大多数"AI 聊天套壳"高明了一个段位。 从 vibe coding 的角度看,回音堂证明了一件事:AI 时代做社区,关键不在于技术多复杂,而在于设定(setting)够不够妙。一个好的世界观设定,配上 LLM 的生成能力,就能长出真实的用户内容和社区氛围。 注:本项目由 VibeFix 编辑从公开渠道精选收录,版权归原作者所有。

秀一秀
Showcase
Online
Social & Community

秀一秀

中文 vibe coding 作品展示社区:已收录 142 个 AI 编程作品,网站、App、游戏和实验项目都能晒。

VibeFix 做的也是 AI 项目展示(中英双语),而中文世界也有人在做同样的事——秀一秀就是一个纯中文的 vibe coding 作品展示平台。它的定位写得很直白:"收集用户通过 vibe coding 开发的网站、App、游戏和实验项目,记录作品和制作过程。" 打开首页,平台已经颇具规模:142 个已展示作品、7 个分类,搜索、热门标签、提交作品入口一应俱全。最新作品流里能看到 AI 角色创作画布、AI 绘画工具、刷短视频式体验 AI 编程小游戏的作品区——某种程度上,它就是中文 vibe coding 圈的"作品集市",每个作品都附带制作过程记录,方便后来者学习是怎么"vibe"出来的。 这类平台的价值在于解决一个真实痛点:vibe coding 的作品往往散落在各个开发者的个人域名和社交媒体动态里,缺乏一个中文的、可检索的聚合地。秀一秀把"展示"和"过程记录"绑在一起,既满足了作者想被看见的需求,也给围观者提供了可复制的路径。 对于 VibeFix 的读者来说,秀一秀本身也是一个选材富矿:142 个中文 vibe coding 作品,按分类和标签可检索,是追踪中文社区在"vibe"什么的好去处。 注:本项目由 VibeFix 编辑从公开渠道精选收录,版权归原作者所有。

Refetch
Showcase
Online
Social & Community

Refetch

An open-source, AI-curated alternative to Hacker News — modern UX, transparent ranking, and community-driven content.

Refetch started with a blunt premise: Hacker News works, but it's controlled by Y Combinator, looks like 2007, and its ranking is a black box. What if a community built a modern alternative — and what if the whole thing was vibe-coded? The README says it plainly: "Refetch was initially 100% vibe-coded, but we're excited to welcome contributions from humans too." The site is live and genuinely usable: a clean feed of curated tech news with upvotes, comments, trending topics, and a "Show RF" section for launches. Behind it sits a surprisingly serious architecture — Next.js 15 with App Router, TypeScript, Tailwind CSS, and Appwrite for the backend (auth, database, serverless functions). Scout functions automatically discover quality tech content from 20+ sources; enhancement functions add AI metadata, quality scores and translations into 12+ languages; a documented ranking algorithm weighs time decay, quality and engagement. The ranking code is public, which is the whole point: transparent algorithms you can read and improve. What makes Refetch a great vibe-coding case study is the trajectory. It didn't stay a demo — it grew into a real product with real users, a Chrome extension, multilingual support, and a contributing guide. The vibe-coded foundation got it to "exists and works"; the open-source community is taking it from there. That's arguably the healthiest possible lifecycle for an AI-built app: the agent does the 0→1, humans do the 1→n. Note: This project is curated by the VibeFix editors from public sources. All rights belong to the original creator.

Workout Coach
Showcase
Online
Health & Fitness

Workout Coach

A free workout coach in your browser — ready-made plans, demo videos, spoken coaching and timers. No account, no ads.

Workout Coach is the kind of app that makes you rethink what "shipping software" means in 2026. The author, Ashwin, had an idea in his head, described it to Claude, and went back and forth until the app existed — no code written by hand, no build step, no backend. The result is a genuinely useful fitness companion that lives entirely in the browser. The app is built around one simple frustration: most people would rather press a button and start working out than keep figuring out what comes next. It ships with ready-made plans — a first full-body routine at home, weight loss with bands and walks, gym first steps, dumbbell muscle building — each with demo videos, spoken coach instructions, and rest timers. The plans are grounded in published training guidance (ACSM progression models, WHO activity guidelines), not vibes alone. The cleverest feature is the AI plan builder. Tap "Create" and it opens a chat with a pre-written message that turns ChatGPT or Claude into a personal coach: it asks about your goals, aches, training location and available time, one question at a time, then hands you a link. Open the link and your custom plan loads in the app — with the same videos, voice coaching and timers. Your plan lives in the link itself; nothing is tracked, no account exists, every demo video is checked to be ad-free. Under the hood it's refreshingly honest engineering: plain HTML, CSS and JavaScript, published free via GitHub Pages, with a self-test mode that runs every day's session on a simulated clock. The README even includes recipes for handing the repo to Claude Code to extend it — the project documents its own vibe-coding workflow. MIT licensed, fork it and make it yours. Note: This project is curated by the VibeFix editors from public sources. All rights belong to the original creator.

PATimer
Showcase
Online
Tools

PATimer

10 年后端工程师用 AI 第一次做出并上架的 App:专为日内交易员打造的多周期收盘倒计时。

PATimer 的作者 forecho 是个写了 10 年后端的工程师,却"非常讨厌甚至惧怕和 Xcode 打交道",两次自学 App 开发都半途而废。2026 年初,他借着 AI vibe coding 模式,终于把第一款 App 做出来并成功上架 App Store——主力结对编程是 Claude Code,Codex 负责稳的活,Google Antigravity(Gemini 3 Pro)调 UI/UX。 产品切的是一个很刁钻的场景:日内交易员看的不是普通倒计时,而是 K 线收盘前几秒——收盘价才是关键。PATimer 在收盘前用声音提醒,帮交易员避免盘中冲动交易,作者把它叫作"纪律执行官"。支持 2/3/5/15/60 分钟周期、美股时段自动识别、24 小时模式(加密/外汇)、系统时间偏差检测、应用始终置顶。 有意思的是作者的产品观:他已经把 macOS 版扩展到了 iOS,并且明确表示——vibe coding 时代,跨平台工具没有意义,直接用 Swift/SwiftUI 原生开发。官网做了中英日韩多语言落地页,App Store 正常上架下载,是"认真运营"的典型。 注:本项目由 VibeFix 编辑从公开渠道精选收录,版权归原作者所有。

易读
Showcase
Online
Education

易读

不会写代码的人,用 AI 团队做出的英文原著阅读器:双击查词、划选拆句,生词自动沉淀复习。

易读的作者 martinachain 不会写代码——他自己是这么说的。然后他用一支"AI 团队"做出了这个产品:CEO 负责拆需求,Claude Code 负责写,Codex 负责审。仓库里的每一条 commit message 都附带一个 Claude-Session 链接,整个开发过程完全透明,是一次把 vibe coding 做成生产力流水线的实证。 产品本身是个英文原著阅读器:网页版 + Chrome 插件双入口,支持导入 EPUB / PDF / Markdown。读原文卡住时,双击单词就能得到结合上下文的中文释义;划选长句可以一键拆成句子结构、关键短语和翻译。查过的生词自动进入 1/3/7/14/30/60 天的间隔重复复习队列,一张句卡连续看懂 3 次才算"毕业"。 最妙的是它的架构:纯前端,数据全部存在浏览器 IndexedDB 里,无账号无服务器,浏览器直连 DeepSeek 官方接口——拆句一次只要几厘钱,所以作者敢把它公开部署而不怕被白嫖。作者还用它实测读完了《从零到一》的英文原著,产品和故事互相印证。开源 MIT,Next.js 16 + React 19 技术栈,截至 2026 年 9 月仍在提交更新。 注:本项目由 VibeFix 编辑从公开渠道精选收录,版权归原作者所有。

GeoSports
Showcase
Online
Entertainment

GeoSports

A daily sports-geography quiz: five questions a day, pin where it happened on the map.

GeoSports is a daily sports trivia game built around one clever twist: every question is answered by dropping a pin on a map. Each day brings five questions about where iconic sports moments happened, and you're scored on how close your pin lands. It's GeoGuessr for sports fans — simple, repeatable, and dangerously habit-forming. The creator is Smith, a sports creator with 3.5M followers, who built GeoSports with Claude and took it viral with a single Instagram plug — zero marketing spend, straight to coverage on Twitter and Yahoo Sports. It's a textbook case of how distribution now works for indie builders: make something good, and the feed does the rest. The business is already real: $5/month or $40/year subscriptions, roughly 1,000 paying members, around $47K in annual recurring revenue. Smith has reportedly turned down betting and fantasy sponsorships to protect players' trust — and set a $4M floor for any acquisition talk. Sport-specific spin-offs are on the roadmap. Note: This project is curated by the VibeFix editors from public sources. All rights belong to the original creator.

MenuGen
Showcase
Online
Lifestyle

MenuGen

Upload any restaurant menu — AI turns every dish into a mouth-watering photo, so you never order blind again.

MenuGen may be the most famous vibe-coded app in existence — it is literally the app from the tweet that coined the term. In February 2025, Andrej Karpathy posted that he was building "a webapp called MenuGen" by talking to Cursor's Composer through SuperWhisper, barely touching the keyboard. That throwaway line became the definition of "vibe coding," and the app behind it is still live and working today. The idea is disarmingly simple: snap a photo of any restaurant menu, upload it, and MenuGen uses AI image generation to produce appetizing photos of every dish on it. No more squinting at a wall of text, no more gambling on dishes you've never seen. Menu roulette, solved. What makes MenuGen special isn't just the tech — it's the story. This was never a venture-backed startup with a roadmap; it was a weekend experiment by one of AI's most prominent researchers, built the way a growing share of the internet now gets built: by describing what you want and letting the AI do the typing. The site still runs, the uploads still work, and it remains the cleanest possible answer to the question "what is vibe coding, really?" Note: This project is curated by the VibeFix editors from public sources. All rights belong to the original creator.

Prayer Lock
Showcase
Online
Productivity

Prayer Lock

A non-engineer, non-native English speaker vibe-coded this faith-based screen-time blocker in ~3 days. It now makes $20K/month with a 4.9★ rating.

Prayer Lock's premise is blunt: "your phone is taking you away from God — block your phone until you pray." It's a screen-time blocker with a faith angle: set your prayer commitments, and the app locks your phone until they're done. The site is a single landing page with App Store and Google Play buttons, and the App Store numbers do the talking: 4.9 stars across 35K ratings. The founder is a non-engineer and non-native English speaker who built the whole thing through vibe coding in about three days — and, reportedly, automated the video production and posting pipeline too. Growth runs on volume: roughly 40 short videos a day across 11–12 accounts on TikTok, Reels, and YouTube. Three days of building, $20K a month, and an audience that genuinely loves it — the indie playbook compressed to its essence. Note: This project is curated by the VibeFix editors from public sources. All rights belong to the original creator. - Verification: curl returns 403 (Cloudflare challenge blocks datacenter requests), but browser.open renders the live page "prayer lock / your phone is taking you away from God / block your phone until you pray"; Microlink screenshot confirms the same live homepage — matches the project.

Sprout
Showcase
Online
Lifestyle

Sprout

"Tinder for jobs," built by a non-engineer in college: swipe to apply, AI-tailored resumes, and an agent that applies for you. $250K MRR in 8 months.

Sprout's homepage opens with a line every job seeker feels: "Stop applying. Start landing interviews." It's a dating-app interface for the job hunt — swipe through roles, let AI tailor your resume for each one, and hand the repetitive submission work to an agent that applies on your behalf while you review before anything gets sent. It was built by Nicole Cheung, a non-engineer at Claremont McKenna, and the growth story is almost unfair: $250K MRR within 8 months, with zero ad spend and zero funding. No growth hacks, no venture backing — just a founder who lived the pain of job hunting and a product whose onboarding does the selling. For anyone who thinks vibe coding can't produce "serious" SaaS, Sprout is the counterargument. Note: This project is curated by the VibeFix editors from public sources. All rights belong to the original creator. - Verification: curl https://www.usesprout.com returns 200; Microlink screenshot shows "Sprout / Stop applying. Start landing interviews." — matches the project.

Polsia
Showcase
Online
AI Agents

Polsia

"AI that runs your company while you sleep" — built solo by one human plus AI, now at a ~$10M ARR run-rate. The reviews are divided; the numbers are not.

Polsia's homepage makes the boldest promise in this batch: "AI That Runs Your Company While You Sleep," with a banner reading "Polsia works on thousands of companies at once." The founder built it solo — "one human + AI," in his words "80% AI, 20% taste" — with zero employees, and it now runs at roughly a $10M ARR run-rate after a $30M raise at a $250M valuation. It is the purest example of the one-person AI company thesis actually working. Honesty requires the other half of the story: public reception is sharply divided. Trustpilot sits at 2.9 out of 5 across 184 reviews, 45% of them one-star, and critics dismiss the output as "AI slop." Whether Polsia is the future of work or a cautionary tale about over-automation is genuinely unresolved — which is exactly why it's worth watching. Note: This project is curated by the VibeFix editors from public sources. All rights belong to the original creator. - Verification: curl https://polsia.com returns 200; Microlink screenshot shows "Polsia / AI That Runs Your Company While You Sleep" — matches the project.

Snag
Showcase
Online
Social & Communityclaude-code

Snag

A 20-year-old NYU student ships a new app every 4–5 hours: humans design in Figma, Claude Code writes the code.

Snag answers a simple question: "Why pay for something when you can get it for free?" It's a local marketplace for free stuff — neighbors list items they no longer need, others snag them nearby. The homepage shows the pitch plainly: a phone mockup with free listings, App Store and Google Play buttons, and 3.3K App Store ratings behind it. The story behind it is the real headline. Benjamin "Benji" Chen is a 20-year-old NYU student who has shipped 45+ apps with a ruthless system: humans do UI design in Figma while Claude Code, running in the IDE, generates the code — roughly 4 to 5 hours per app — with Supabase as the backend. Snag is the one that stuck: $30K a month, 100K users, and 9K of them converted to paid. It's vibe coding treated not as a party trick but as an assembly line. Note: This project is curated by the VibeFix editors from public sources. All rights belong to the original creator. - Verification: curl https://snagapp.com returns 200; Microlink screenshot shows "SNAG / Why pay for something When you can get it for free? / We help you snag free items near you." — matches the project.

Launch Fast
Showcase
Online
Marketing & Salescursor

Launch Fast

多伦多亚马逊老卖家不会写代码,用 Cursor 48 小时做出生产级后端,如今月入 3 万美元。

Hasaam Bhatti 在多伦多做了近十年亚马逊 FBA,自己完全不懂技术。他的打法很聪明:先花两年时间收购了一个有几千名活跃卖家的卖家教练社群(LegacyX FBA),摸清大家最痛的点,然后用 Cursor 花了 48 小时——"Cursor carried the implementation entirely"——做出了一个生产级后端。分发问题在他动手之前就已经解决了。 Launch Fast 自称"亚马逊卖家的操作系统":输入关键词或 ASIN,它就用独家的 A10–F1 评分给市场打分,再给出历史价格、BSR、营收、评论、广告 CPC 趋势、关键词对比、阿里巴巴供应商评分和全套财务建模,还带 Chrome 插件和 MCP 接口,让卖家自己的 AI agent 也能调用它的数据。目前 $30K 月收入,创始人把 48 小时上线、30 天 $10K 的全过程写在了 Indie Hackers 上。 本项目由 VibeFix 编辑从公开渠道精选收录,版权归原作者所有。 - 验证方式:curl https://launchfastlegacyx.com/ 返回 200,Microlink 截图首页显示"Launch Fast / The Amazon seller operating system for better product decisions",与项目相符。

Stella
Showcase
Online
Health & Fitnessclaude-codecursor

Stella

不会写代码的 450 万粉丝博主,用 Claude Code + Cursor 做出"显化"App,两个月干到约 30 万美元月收入。

Stella 的作者 Sarah Perl 是个坐拥 450 万粉丝的内容创作者,她公开承认自己完全不会写代码——整个 App 是她用 Claude Code 和 Cursor vibe coding 做出来的。用她自己的话说,AI 把产品化成本压到了接近于零,真正的护城河在发行端:她只需要在自己已有的 TikTok、Instagram、YouTube Shorts 上演示自己用它,就有现成的几百万人愿意下载。 产品是个"显化"(manifestation)应用:个性化可视化练习、遇见"未来的自己"、30 万人一起许愿。上线约两个月就积累了 12,000 个付费订阅用户,月收入约 $300K。App Store 上 4.7 分、5,200 多条评价,健康健美榜第 69 名,还挂着"#1 Manifestation App"的徽章。当分发能力拉满时,vibe coding 的天花板可以非常高。 本项目由 VibeFix 编辑从公开渠道精选收录,版权归原作者所有。 - 验证方式:curl App Store 链接返回 200,Microlink 截图显示"Stella - Manifest Anything / 4.7 分 / Perl Productions LLC",与项目相符。

Profit AI
Showcase
Online
E-commercecursorclaude-code

Profit AI

不懂技术的创始人用 Cursor + Claude Code,把一张利润分析表做成了月入 3 万美元的 Shopify 应用。

Profit AI 的作者 Jack Kavanagh 完全不懂技术。他的起点只是一张算不清账的利润分析电子表格——2025 年圣诞节,他用 Cursor 和 Claude Code 把这张表变成了一个 Shopify 应用:把最新的 AI 模型接到店铺数据上,实时算出每一单真正的利润,还能自动给出"Facebook 广告预算该涨 20%"这种可执行的建议。 成绩单很实在:应用月收入 $30K、累计 $147K,122 个用户里有 38 到 39 个付费,客单价 $799/月。从"我自己算账用"到一个被 Shopify 卖家掏钱订阅的 FP&A(财务计划分析)套件,中间只隔了一个非技术创始人和两个 AI 编程工具。 本项目由 VibeFix 编辑从公开渠道精选收录,版权归原作者所有。 - 验证方式:curl https://tryprofit.ai 返回 200,Microlink 截图首页显示"Profit AI / Connect the latest AI models to your Shopify store data",与项目相符。

Laper
Showcase
Online
Content & Mediaclaude-code

Laper

前编剧用 Claude Code 零手写代码做出的 AI 编剧软件:十万行代码,全是 AI 写的。

Laper 的作者赵纯想以前是给韩寒的亭东影业写剧本的编剧,转行做产品后,他干了一件很极端的事:Laper 超过十万行的代码库,"zero manual coding, entirely by Claude Code"——一行手写代码都没有,全是 Claude Code 生成的。他自己只负责判断"这个功能该不该做"。 产品本身是个"没有 AI 味"的 AI 编剧工具:剧本格式自动整理、剧本结构可视化、多人实时协作写剧本,slogan 就叫"AI Screenwriting Software, Without the AI Aftertaste"。目前 $40,708 的月经常性收入、1,165 个付费用户,Senior 套餐 $20/月起。一个前编剧用 AI 把自己原来的职业做成了 SaaS,这大概是 vibe coding 最有说服力的样本之一。 本项目由 VibeFix 编辑从公开渠道精选收录,版权归原作者所有。 - 验证方式:curl https://laper.ai 返回 200,Microlink 截图首页显示"Laper / AI Screenwriting Software, Without the AI Aftertaste.",与项目相符。

WishBox
Showcase
Online
Design & Creativegoogle-ai-studiogemini

WishBox

一张会吹蜡烛、拆礼物、洒彩纸的交互式生日贺卡。

WishBox 把一张生日贺卡做成了四步的"庆祝仪式":对着麦克风吹灭蜡烛、拆开礼物盒、漫天彩纸,再读一封手写信。全程有背景音乐和音效,还能自定义收件人名字。发给朋友的不是一张图,而是一次小小的惊喜。 作者 viochris 在 Google AI Studio 里用 Gemini 纯 vibe coding 写了它,GitHub 仓库 2026 年 9 月还在更新,README 里完整记录了从 PRD 到上线的提示词过程。对想学 vibe coding 的人来说,这个仓库本身就是一份公开的作业本。 本项目由 VibeFix 编辑从公开渠道精选收录,版权归原作者所有。

Stories of Life
Showcase
Online
Educationbolt

Stories of Life

把孩子每天的心情变成专属睡前故事,讲着讲着就长大了。

Stories of Life 是个睡前故事 App,但和"随便讲个故事哄睡"不一样:家长每天记录一件小事或孩子的情绪,App 会生成一个量身定制的、适合孩子年龄的晚安故事。它的定位不只是哄睡,而是"陪伴孩子情绪成长"的家庭仪式。 作者 Harshitha P. 自己就是每天给孩子讲故事的妈妈,用 Bolt 加 Supabase 一个人把它做了出来,现在每晚都在用。从"我家娃睡前要听故事"长出来的产品,带着天然的温度。 本项目由 VibeFix 编辑从公开渠道精选收录,版权归原作者所有。