让 AI 做个应用,带着自己慢慢学 Arduino,这个用法挺好。
Hardware looks scary until you break it down into small steps. So I built an app that takes you through Arduino one circuit at a time.
Made with Opus 5.5
learn-arduino-red.vercel.app
@全体成员 互联网今晚开始要大逃杀了
点击接下来这个链接自行测试一下
你的浏览器对目前
印度 CA 签发
弱密钥导致的证书泄露的防护情况
如果有一个不是绿的
那你就需要立刻尝试更新浏览器版本 emudhra.we-have-a-problem.co…
内裤扒给你们看
流量快来
@nan073986563303(秋枫)不是新号,2023 年 8 月就注册了。最近涨到 11977 粉,靠的是互关(粉丝 11977、关注 12340),加上一直围绕“X 创作者收益”发帖。它最近 40 分钟回了 24 条,都是“流量来!”这类短句,中位浏览只有 4.5。
它比 @Bigtomatoooo 爆得更高(原创最高 3471 浏览,对方是 1475),是因为多了“钱和数据”这个钩子:
- **聊大家最关心的钱**:[“开了 Wise,开创作者收益还要准备什么?”](xitter.cf/i/status/2112500353145…)有 3471 浏览、130 条评论;[“1.1 万粉了,月入一万在哪领?”](xitter.cf/i/status/2112709233586…)有 224 条评论。
- **公开记账连载**:反复更新“有效曝光 111.9K,离 50 万还差 388K”,有具体数字,也有进度,让人想追着看。
- **人设清楚**:爱用 AI 的大学生,在 X 上冲收益,经常自嘲,比如[女朋友问我在干啥](xitter.cf/i/status/2112818834726…)那条有 1500 浏览。
- **固定开头和结尾**:几乎每条都用“兄弟们”开头,用“你们到哪一步了”结尾。
光是能把电视广告干掉,就想折腾了。
un tipo vibecodeó su TV con Claude y le sacó todo el bloatware y los anuncios...
recreó la interfaz inspirada en Apple TV
de 282 MB a 96 MB de RAM. 3x más rápida. cero anuncios
recomendaciones de pelis, clima, precios de crypto y se oscurece sola al atardecer
el repo está abierto y funciona en Google TV, LG y Samsung
te lo dejo en comentarios
Cloudflare 把 source map 上传改成默认开启:Wrangler 4.150.0 起,错误堆栈能映射回原文件名和行号。
有个容易漏看的细节:自定义构建里无法解析的映射,默认会跳过;显式设 upload_source_maps=true 才会让上传失败。
我更在意这个取舍:部署成功,不等于出了错就能定位。若排障依赖源码行号,映射是否可用也该成为发布检查的一项。
Wrangler 4.150.0 and the Cloudflare CLI now upload source maps by default. Debugging just got easier with real file names in stack traces.
developers.cloudflare.com/ch…
早上忘记发了,哈哈哈
早上好!过去一天 X 上值得看的几条:
- [What2Build](xitter.cf/createdbyharsh/status/…),可以像刷 TikTok 一样浏览真实的创业项目,能看到 MRR(月度经常性收入)和总收入,还能按赛道筛选;帖子没提它自己赚多少,但很适合找点子,这种收入数据聚合站本身也能照着做。
- [AutoAiShorts 第 133 天](xitter.cf/MDALISHANALI2/status/2…),这是个用 AI 自动生成短视频的工具,现在有 50 个付费订阅,MRR $354,累计新客 2,255;这类工具订阅一个人就能做。
- [Airouter](xitter.cf/YashAg946/status/21128…),面向印度开发者的 AI 接口中转平台,收入刚过 $1,000;针对一个地区做本地化的 AI 接口转售,门槛不高。
- [IndieLaunch 满月数据](xitter.cf/rupamjit_ghosh/status/…),这个产品发布目录站一个月来了 1,086 个访客、收了 200 多个产品,但收入只有 $3,服务器费却要 $20;目录站引流容易、变现难,可以当反面参考。
- [FeedbackPanda 的故事](xitter.cf/V_Khomichenko/status/2…),它帮在线英语老师更快地写学生反馈,大约两年做到月收入约 $55K,2019 年卖掉了;给很窄的一群职业用户做省事工具,也能做大。
- [cloudflare/skills](xitter.cf/slashui/status/2112711…),这是 Cloudflare 官方写给 AI 编程工具看的开发说明,让 AI 写 Workers 或 Agents 时自动按官方写法来,还带一个 MCP(让 AI 调用外部工具的接口);在 Cloudflare 上部署小产品能省不少事。
- [自嗨式开发的教训](xitter.cf/Lazily57/status/211289…),作者花两周用 Cursor 做了个 AI 提示词管理站,上线第一天只有 3 次访问,还都是自己点的;先确认有人要、想好怎么找到用户,再动手写代码。
提醒一句:有人发帖说“21 天 MRR 破 $3,200”,结尾在求点赞收藏,曝光只有十几次,这个数字没法核实,看看就好。另外,“一人公司月入几十万”这类帖子,很多发帖人的主页其实是在卖课。
Built a little side project for indie hackers 🚀
Introducing What2Build.
A TikTok-style feed to discover startups, explore their revenue, and find inspiration for your next project.
• Discover real startups
• Explore MRR & total revenue
• Filter by niche, audience, and revenue
• Find ideas worth building
Stop guessing what to build. Start with what's already working.
Try it: what2build.app/
Would love to hear your feedback!
Vercel 的 durable-harness 把代理中断后的恢复拆得很细:已完成的调用读取日志。文章示例把查询 CI 标成可重放,把发 PR 评论标成不自动重放,结果不明时告诉模型“可能已执行”。
我觉得关键不在跑多久,而在如何处理“请求发出,结果没回来”:恢复进度与重复操作是两回事,未知结果应先核对,再决定是否继续。
Introducing 𝚍𝚞𝚛𝚊𝚋𝚕𝚎-𝚑𝚊𝚛𝚗𝚎𝚜𝚜. Run Pi, OpenCode, Codex, Claude Code, eve, fx, or your own harness durably on Vercel.
With 𝚍𝚞𝚛𝚊𝚋𝚕𝚎-𝚑𝚊𝚛𝚗𝚎𝚜𝚜, agents automatically resume work after failures and run for hours, days, or months. vercel.com/blog/making-every…
Carmine 分享了一个并行跑 Rust 构建的处理办法:把缓存移到第二块 NVMe,并用 mbx 让各个 worktree 共用缓存。他说这样解决了桌面卡死;附图还给桌面预留了两个 CPU 线程。我觉得这里值得区分两个目标:复用缓存是少做重复工作,预留资源是别让构建挤满整台机器。不能把这一个配置案例当成通用提速结论。
Pro tip for Rust on Linux: give your build cache its own drive.
I moved mine to a second NVMe (TLC with DRAM, XFS, no compression), and my desktop no longer freezes when agents run builds in parallel.
Builds go through mbx by @jdxcode, so every worktree shares one cache.
AI 都开始给数学家出难题了。
Over the past year, the Discovery team at @GoogleDeepMind, in a project led by student researcher @dsantosmarco, has made new discoveries in one of the most beautiful open problems at the edge of computation: Finding the 6th Busy Beaver (BB) number.
While most AI math systems solve human-generated problems, like theorems, this work tackles a different aspect of mathematical creativity: Discovering hidden problems.
For years, the BB Community has pursued one of theoretical computer science’s most captivating collaborative quests: finding the simplest programs that run the longest before halting. Because determining whether these programs ever stop is uncomputable, no universal algorithm can solve it, meaning new mathematical techniques need to be found and invented to make progress.
In this way, the BB challenge serves as an open ended domain for mathematical creativity and, in particular, a generator of hard math problems called Cryptids. These are tiny programs whose halting behavior is governed by and equivalent to deep, unsolved math problems in number theory, similar to the Collatz or Goldbach conjectures. Historically, the BB community had recognized ~20 Cryptids in BB(6), and the best problems were selected for the Busy Math Olympiad (BMO).
Building on their work, Marco built a pipeline powered by Gemini 4 Argon that:
1️⃣ Simulates raw Turing machine traces to compress recurring patterns.
2️⃣ Conjectures mathematical transition rules & runs synthetic experiments to test edge cases.
3️⃣ Proves the equivalence formally in Lean 4.
Using this pipeline, Marco and the team discovered 20 new potential Cryptids and verified 33 more machines previously studied in the BB community. All the proofs were open sourced and shared in the community discord channel and we hope that they will help the community to make progress towards resolving BB(6). We also contributed related halting/non-halting proofs directly to the community GitHub repository.
Finding these Cryptids shows hidden links between simple programs and number theory, and helps mathematicians map the limits of computation. I'm also excited about Gemini 4 Argon's ability to help researchers both solve hard problems and accelerate scientific discoveries, of which Busy Beaver is just one example.
Work led by our student researcher @dsantosmarco with paper contributions from @vivek_veeriah, @Dahoas1 , Andrew Bolt, @zuza777 , Satinder Singh, and myself with additional contributions from Hussain Masoom, @maxencefaldor , @MoritzFirsching and Petar Sirkovic.
Read more about it in our paper: arxiv.org/abs/2610.22603
Resend 新出的 Inboxes,让人和智能体共用收件箱:团队能分配会话,也能通过 MCP 或 skills 接入已有智能体。我觉得重点不只是自动回邮件,而是让人工接手时仍能沿着同一条会话继续。接入能力不等于处理权限,哪些邮件只起草、哪些可以直接发送,仍要在工作流里分清。
Today at Resend Forward, we're introducing two new products.
▪️ Inboxes: shared inboxes for agents and humans
▪️ Forge: Resend's new sending infrastructure, fully controlled by us
Here's what's new 👇

ALT Two keynote title cards stacked. Top: INBOXES over the Resend Inboxes interface on a blue glow, with 'The inbox for agents' and 'Build inboxes your way'. Bottom: RESEND FORGE over the Forge device on a green glow, with 'Resend's Mail Transfer Agent (MTA)' and 'Built for speed and deliverability'.
GitHub 迁移 Git 仓库时,遇到的一个瓶颈是跨区写入还要占着整个仓库的锁。官方文章介绍,他们把串行范围缩小到单个引用,让不同分支的写入可以同时进行。这个取舍值得借鉴:网络往返并没消失,减少的是不相关操作之间的等待;争用同一个引用的写入,仍然得排队。
Supabase 的 .NET SDK v8 把 Auth、Realtime、Storage 等包放进同一仓库,并统一版本号。主包从 v1 跳到 v8,是为了跟原本已到 v7 的 Realtime 对齐,不是突然跨过七代功能。对维护项目的人,这减少了搭配依赖的猜测;但官方明确有破坏性变更,版本号统一不等于升级零成本。
The Supabase .NET SDK is now officially supported with a major new version. v8 puts every package in one repo, on one version number.
Learn more → supabase.com/blog/introducin…






















