kamikhanz

[Miguel Ángel Durán @midudev]

¡Esta skill de Claude Code es una maravilla!

Se llama anatomy y crea ilustraciones isométricas interactivas para explicarte cualquier concepto.

Genera SVG con efectos 3D y es de código abierto: https://t.co/6YjyY7xzRm

[Daily Fitness @DailyFitHQ]

Dear men,

Do 200 push ups everyday and see the changes in 2 week.

[Thariq @trq212]

a big advantage of doing planning like this is that it’s a lot more token efficient compared to raw HTML-

the model doesn’t need to remake the components or logic to do common things like state machines, diagrams, code snippets, etc.

[Zam @zamarrowski]

Si no lees el código al menos ten guardrails en tu proyecto, te dejo algunos:

🔎 knip: mira el código y te chiva que archivos no se usan, que exports que nadie importa, etc.

🧬 jscpd: encuentra código duplicado para evitar que haya varias funciones haciendo lo mismo

👇

[Hernán @hernan__cc]

ayer escuché una analogía muy interesante para pensar la adopción de IA:

"si mañana te doy un asistente personal (humano) que puede aprender cualquier cosa en minutos, que harías hoy para delegarle efectivamente tu trabajo?"

- documentar tus proyectos
- darle un espacio de trabajo
- escribir documentos de onboarding
- ordenar tus archivos
- darle acceso a tus herramientas de trabajo

o le dir��as "encargate de esto" sin ningún contexto?

lo mismo tenes que hacer para aprender a usar agentes IA efectivamente.

el 90% de nosotros nunca tuvo gente a cargo, por eso nos cuesta entender cómo delegar efectivamente.

ahora podemos tener equipos enteros trabajando 24/7. para llegar ahí primero tenemos que ordenarnos nosotros.

[Jasper Kasparov @JasperKasparov]

I started taking B12 in the morning, omega-3 at lunch, and magnesium before bed.

Without exaggerating, my personality changed 180 degrees.

1. B12, in the morning

[Vivek Karuturi @VivekxK]

We do almost no code review.

Most of our review happens in the planning process with https://t.co/RG0yIZGuAl and Fable, not at the PR stage.

After we’ve aligned on key decisions/tasks as a team, I use Ref’s built-in orchestrator to fan out multiple tasks in parallel to Opus 5.

Once the initial implementation is done, the PRs all go through several AI-driven reviews to ensure:
* The implementation matches the intent of the plan (Ref conformance checks).
* General bugs are caught, fixed and noted in the plan (Codex is great here)
* Security review (Codex also great)
* Linting, unit and integration test suites are green (standard CI practices)

The orchestrator agent is typically still managing all of this when it’s in-flight. It notifies me if anything needs my input.

We also have preview deploys on all PRs so we can always manually verify features end to end ourselves when needed.

[Mark Phelps @itsmarkphelps]

I think I may have found the perfect agentic planning process.

1. Use @mattpocockuk 's /grill-me skill to flesh out the initial PRD
2. Use @thekitze 's /council skill (see below) to cross check the PRD with multiple models/agents (ie: claude sends to codex)
3. Use plannotator to annotate the plan with any feedback you have and send it back
4. Go back to step 2. Loop until you are happy with it

[Jordan | 先生 @jordanovweb3]

Everyone is still optimizing prompts

The people running 20 agents stopped doing that a year ago

Prompt → Context → Harness → Loops

Matt Pocock put his whole workflow in one hour and called it "just copy him"

384,000 watched it, almost none copied the third layer

a rule in a prompt is a suggestion

a rule in a harness is a wall

that's the entire difference

you probably don't have an hour right now

bookmark it before it's gone

watch, then read the full guide below ↓

[Aniket Pawar @alaymanguy]

Seeing a lot of screenshots tagging @rauchg about exhausting Vercel limits

Introducing Vercel Doctor ▲

Scans your codebase to help optimize your bill:
- Large static assets → CDN
- Link prefetch false by default
and more...

Run as a CLI or agent skill. Fully open source.

[Next.js @nextjs]

We closed 1,500 issues on 𝚟𝚎𝚛𝚌𝚎𝚕/𝚗𝚎𝚡𝚝.𝚓𝚜 in one month by building a 𝚌𝚕𝚘𝚜𝚊𝚋𝚒𝚕𝚒𝚝𝚢 agent with @eve.

The workflow:
1. Search related work
2. Try to reproduce bugs across versions in a sandbox
3. Return context for review

https://t.co/uuz2BYDHjf

[Harshil Tomar @Hartdrawss]

Steal this clean mobile app UI system :

- Titles: Inter Tight 22/28px Semi Bold, -0.7px letter spacing
- Section headers: Inter Tight 19/24px Semi Bold, and 0.6px letter spacing
- Body/day labels: Inter Tight 14.5/18px Regular, and 0.3px letter spacing
- Buttons: 38-46px circular, 6px inset highlight + soft drop shadow, no flat fills
- Cards: 12-18px radius depending on depth, layered box-shadows not borders
- Text: var(--text) for primary, var(--muted) for secondary, var(--strong) for icons
- Dividers: 1px var(--track), never a visible border color

Tight tokens, layered shadows instead of borders, negative tracking on every heading.

That's what makes dense mobile screens still feel calm.

[Cam @Cameron_C2]

Vitest 5 is out!!

Some super nice perf improvements in this release

https://t.co/jCL0oupai9

[Matt Pocock @mattpocockuk]

/show-me is a phenomenal skill

Makes PR descriptions extremely easy to read

Basically a toolbox of "nice ways to look at code"

Nice work, @dexhorthy

https://t.co/phB1dAPwgG

[Peter Yang @petergyang]

If you're trying out Fable 5.1 I highly recommend running:

/claude-api prompt-audit

on your skills. It finds a bunch of redundancies and rules to remove for the latest models.

Running it for all my skills now.

[Matt Pocock @mattpocockuk]

This has ended up being better than expected and fills an interesting hole in my skills.

Most folks don't have an AFK agent workflow set up yet. You should. It's better than /implement-spec.

BUT running /implement-spec is good, especially while tokens are cheap, and very simple to set up.

I'll be graduating this into my public skill set once I'm back from holiday.

[alex shapalov @alxshp]

https://t.co/rbsinUJrsG + https://t.co/ii8QrPI9YP almost crossed 1,000 GitHub stars combined in just two weeks! ⭐️⭐️⭐️⭐️

Two open source tools for PostgreSQL:

→ pgbot Postgres diagnostics for humans and AI agents (Go)
→ pgterm: a fast Postgres terminal written in Rust

A huge thank-you to everyone who tried them, starred them, and shared feedback 🐘

Next stop: 2,000 ⭐ :)

https://t.co/4B5p3lgwn1

[Inspora @insporadesign]

Mini calendar by @jeetnirnejak

More on →https://t.co/QIsUfM3KGG

[Michael Guo @Michaelzsguo]

Lauren Tan @poteto 是 Cursor 的工程师，之前在 Meta 做 React Compiler，也在 Netflix 做过 tech lead 和工程经理。

她加入 Cursor 只有五个月。第一个月还在熟悉代码库，上个月已经合入了 1000 个 PR。这个月才过去 12 天，她又合入了接近 800 个。

这不是 AI slop code，而是你每天都在使用的 Cursor 的代码。

很多人，包括 Claude Code 的 Boris，都提过自己借助 coding agent 达到了类似的效率。但真正愿意把工作方法完整分享出来的人并不多。Lauren 在这个一小时的视频里，几乎是手把手讲了她怎么走到这一步。

她认为，用 AI coding 最大的问题不是生成代码，而是验证代码。

如果 agent 不能自己运行产品、操作界面、读取 CPU trace 和 heap snapshot、打开模拟器并复现问题，那么最后还是要由你来检查结果。你就是整个流程的 verifier，也是无法并行工作的瓶颈。

Lauren 的做法，是先给 agent 建立完整的验证能力：让它能通过 Chrome DevTools 或模拟器实际操作产品，再用 feature map 告诉它每个功能在哪里、怎么进入。这样即使同事只丢来一张截图，或者一句很模糊的 bug 描述，agent 也能找到对应功能，复现问题并验证修复。

每当她发现 agent 在猜测、漏读代码或走错方向，就把这个失败模式写成一条 skill。然后像测试代码一样测试这些 skill：让多个 sub-agent 分别执行任务，由 coordinator 制定 rubric，再让另一个模型交叉检查评分，反复迭代到结果足够稳定。

现在，她甚至允许 agent 自动合并 PR。有一天早上醒来，已经有 20 个 PR 自动进入 main；她直接在 main 上检查，结果都没有问题。

这套方法不是简单的提示词技巧，更像做工程管理：先设计好环境、流程和验收机制，再让团队并行工作。只不过这支团队，现在由几十个 coding agent 组成。

[Taras @tarasshyn]

Today I learned about Entity Stacking.

If you put these links in your Organisation schema under `sameAs`, Google will understand who the fuck you are far better than with 300 junk directory listings - and it’s free.

Cheers, @Charles_SEO 🎩