Aryu

@prisma What we want: - no breaking changes any more - have real performance improvement

Today, we’re sharing our vision for a new foundation for Prisma ORM. It keeps the experience you love and sets the direction for a more composable, extensible, agent-ready future! We'll let the code speak for itself ✨

@wong2__ 我用这个 github.com/neutree-ai/ope…

看到大家纷纷把http api包装成cli给agent用，我有个疑问：为什么不是api文档+curl呢

@forehalo @__oQuery @better_auth @coderabbitai @grok explain this

@__oQuery @better_auth the response from @better_auth dev reminds me what @coderabbitai CEO did months ago

I've been following @better_auth since the 0.x days. I was an early adopter, actively asking for help on their Discord, filing issues, and building with it. The library itself is great — a massive improvement over AuthJS auth/core. But their release engineering has been consistently problematic. Here's why. 1/

@loujaybee Thanks for your creation! BTW, could you share how did you design and implement the animation arts?

@Aryu0112 Thanks for sharing !

background-agents.com elegant

@xxholly32 因为现在少儿编程教的内容，很快就变成历史了

@Aryu0112 这个是什么逻辑

如果你的孩子喜欢历史，那现在就可以送他/她去学少儿编程

@libukai 做的真不错，但是感觉用来做设计这个领域，有点可惜了

Pencil.dev 这个 Agent Team 并行设计给人的感觉实在是太科幻了，😇。认真跑了几个测试项目，深入看了看大模型对话日志，可以说在整合 MCP/Skill/Agent 这几个概念的能力上，Pencil.dev 绝对是顶级模版，已经是那种我肯定是达不到的境界。

@prajdabre How many agents are answering this threads?

Technical interview question: Suppose you have 5 TB worth of text data and you want to count the total number of words, how will you do this?

@tkkong I still get a little shy speaking to agents in public～

holy shit dictation mode in Codex is game changer

@souporserious This is a big bet. Winning the ecosystem, especially at the library level, doesn’t guarantee the ultimate commercialization.

Vercel snagged agent on NPM 👀 npmjs.com/package/agent

@garrytan The auto complete feature is underrated, it generate the expected words for me in about 70% cases.

Is this an easter egg in Claude Code? It wants me to go to sleep. It's worried about my sleep habits.

@shafu0x Hard agree, the UI/UX shift is wild.

just build for agents

@dabit3 replay is a must-have feature for agent sessions

Devin lets you scrub, rewind, and replay agent sessions. Useful when running parallel agents, background agents, or scheduled sessions / cron jobs.

@addyosmani agent-friendly factories: - self-contained small-to-medium size repos/libs - 100% automated testing (function + performance) - feature requests as new automated test cases - test cases ensuring compatibility - README for both humans and agents - docs delivered as skills.

Every abstraction shift in software history made devs more productive by raising the level of intent. This is the next step: from writing code to orchestrating systems that write code (building "the factory" for your code). The unsolved problem isn't generation but verification. That's where engineering judgment becomes your highest-leverage skill. To truly scale, think "factory model" - orchestrate fleets of agents like a production line: clear specs as blueprints, TDD for quality control, strong architecture to amplify leverage.

@jakevin7 oh-my-opencode 和以前的 awesome-xxx 是类似的，满足了收集欲，解决的问题很有限。而且为了吸引流量，迭代不够克制，臃肿之后就成了影响 agent 泛化的杀手。而且现在迭代速度这么快，第三方很难超过第一方。

AI时代，真的特别容易fomo。 但是请注意好保护我们自己的注意力。 还记得之前爆火的 oh my opencode吗，现在呢。 本质上还是看需求，应该是拿 ai当工具而不是凑着去用。 多思考自己，思考问题。 而不是拿着锤子找钉子，致每个人。

@tomdale managers’ experience directing people seems to transfer pretty smoothly to directing agents now

Scariest impact of AI so far is managers are coding again.

@kasong2048 跟年龄没有必然关系，跟「有效踩坑+学习」经验有关。

有种预感，AI 接下来的发展对大龄程序员是重大利好 程序员之所以有35岁危机，是因为本职工作不如年轻人能卷。这种“靠个人时间堆积的 ‘卷’ ”在 AI Coding 带来的效率提升面前根本不够看。 除此之外，大龄程序员积累的多工种能力（比如 产品 + 前端 + 后端）可以快速转化为「指挥多工种 Agent 工作的能力」。 极端情况下，会出现“一个跨能力大龄打工人指挥多工种 Agent，独立完成产品的全生命周期”。 这种效率提升是 10x 以上的。毕竟，Agent 之间不用开会撕逼「狗头」 有人会问，年轻人不也能指挥 Agent 卷么？残酷的现实是，AI Agent 太好用了，很多公司已经不愿意培养新人了，新人已经失去了老人当年通过实践掌握各种能力的环境。

@Wattenberger AskUserQuestion tool can be extend to AskMyMatesQuestion

Screw Slack for agents, I want humans in my agent tools

@JFPuget Algorithm task is the sweet spot for coding agent. Because it's self contained, explicit to verify, loop fast.

I am using codex (or claude code) more and more for developing algorithmic code. Still doing it step by step, in order to validate each step thoroughly, but not writing code anymore, as soon as I am a bit tired, or running out of time. This is faster than I'd do by hand, especially when I am tired. There is a caveat though: I spend more time fixing typos in my task description that anything else. Codex is very sensitive to typos. One character change in a function name and it creates a new function without asking if it could be the existing one. I'm dyslexic, which is why I spend lots of time fixing typos: I make lots of typos. The flip side is that I also makes typos when coding. One could argue that I suffer equally from it, whether coding or describing what to code. Not true, with autocomplete typos are less likely in coding.

@BHolmesDev now we have remote solution in many levels: - network level - terminal level - agent level - agent orchestrator level

He's not exaggerating by the way. We set up shareable SSH sessions so anyone on a team can join an agent while it works. There's even multi-cursor to see what someone is typing in the prompt box 🤯