mikasa

no

[陈成 @chenchengpro]

5 个最近讨论比较多的 skill：

1/ eli5
https://t.co/SlzpHhSiwc

2/ show-me
https://t.co/KHSJjetZWw

3/ skill-doctor
https://t.co/56hWhwHoV1

4/ retro
https://t.co/uFiRBivwHM

5/ unslop
https://t.co/CIlou9KzPQ

[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。

[Charlie Hills @charliejhills]

You installed Claude Code and stopped there.

35 add-ons that turn it into a company (links below):

If you're new to Claude Code.
Start here → https://t.co/YbdDKUJsSM

A SENIOR ENGINEER

1. superpowers
→ https://t.co/tGNwFVYsEa

2. agent-skills
→ https://t.co/n3GAaErDOH

3. karpathy-skills
→ https://t.co/z0R8ozhtmb

4. gstack
→ https://t.co/XPJ0gmpOIx

A PRODUCT DESIGNER

5. ui-ux-pro-max-skill
→ https://t.co/54NvwKmBix

6. taste-skill
→ https://t.co/gFud3Sa93j

7. impeccable
→ https://t.co/fFDkTq8wQR

A QA TESTER

8. Codex
→ https://t.co/PPPmwl9F1p (paid)

9. playwright-mcp
→ https://t.co/C9b2nZxiTF

10. skills
→ https://t.co/iR9fyxUFvL

A DOCS TEAM

11. markitdown
→ https://t.co/nJzODQKIDl

12. skills
→ https://t.co/Nj47wymxJR

13. frontend-slides
→ https://t.co/cjLcONPQIT

A MARKETER

14. marketingskills
→ https://t.co/ZyoMWfvgf8

15. claude-seo
→ https://t.co/h481BlnvgB

16. humanizer
→ https://t.co/3a9HKsFIYX

17. Apify
→ https://t.co/yn7rQKdnnz (paid)

A SOCIAL MANAGER

18. social-media-skills
→ https://t.co/ZwNtoUTdVm

19. ManyChat
→ https://t.co/vBo70XUjQK (paid)

20. GPT Image 2 →https://t.co/Ptbm54A58B (paid)

A MOTION DESIGNER

21. HiggsField
→ https://t.co/B3eLcc2Kn0 (paid)

22. hyperframes
→ https://t.co/ogS5d5ityj

23. OpenMontage
→ https://t.co/H8lDc70xNW

24. OpenCut
→https://t.co/9YLKSKYHCt

A RESEARCHER

25. Agent-Reach →https://t.co/1mFzc91H9X

26. last30days-skill
→ https://t.co/TdrSYOFQsT

27. NotebookLM
→ https://t.co/8ZW2a795nX (free)

AN OPS MANAGER

28. Notion MCP →https://t.co/OmXaT3B4iE (free)

29. github-mcp-server
→ https://t.co/ITX95VKOZJ

30. claude-mem
→ https://t.co/emHpdPtiUO

31. mem0
→ https://t.co/snaZn591kH

THE WHOLE AGENCY

32. agency-agents
→ https://t.co/9XD1C4BsmQ

33. awesome-claude-skills
→ https://t.co/hI8v1hrMkb

34. agents
→ https://t.co/0LNt2gpXGH

35. claude-skills
→ https://t.co/SqRj6vAzYh

What each one does is on the graphic.

Every repo is free. Five of the tools are paid.

Seven are ones I already run.

The rest are the ones I keep sending people.

[SONIA @S0N_IA]

EL INGENIERO QUE CREÓ CLAUDE CODE ACABA DE SUBIR 28 MINUTOS DE ORO PURO

sacó un video completo sobre cómo escribir prompts que realmente entregan resultados, no genéricos de manual.

hay cursos de 300 dólares que no llegan ni a la suela de lo que explica en los primeros 10 minutos.

archivos CLAUDE.md, atajos de memoria, sesiones paralelas, patrones de prompting que casi nadie usa.

28 minutos. GRATIS. sin relleno.

no importa si llevas años programando o recién empiezas con Claude: esto te cambia el flujo de trabajo desde hoy.

la documentación oficial te dice qué existe. esto te enseña cómo usarlo de verdad.

guárdalo. míralo. compártelo.

[reira @reiraxbt]

Andrew Ng just released a free 2-hour course on complete Graph Engineering

How to go from one prompt to 100 agents running inside self-improving graphs:

09:14 - Build your first agent from scratch
33:11 - Master loop engineering
1:02:46 - Turn agent loops into graphs
1:30:15 - Build agents that rewrite themselves
1:49:05 - Run the full graph system without you

Prompts → Context → Workflows → Feedback → Loops → Graphs

Most graph tutorials stop at the diagram

This one has you running a real graph within the first 20 minutes

Most people are still prompting one agent at a time

Andrew Ng is already teaching the layer above:

Self-improving systems that build and run themselves

Watch this brilliant course and build the graph

Then read the full architecture below ↓

[Kirill @kirillk_web3]

A Chinese developer just explained the shift from Loop Engineering to Graph Engineering better than anyone.

most people are still building agents the way that's about to be obsolete.

> why single-agent loops break and go "goal blind"
> the 4 parts of a graph: nodes, edges, state, policy
> 3 topologies that run everything: diamond, supervisor, pipeline
> Anthropic's 5 official workflow patterns

the punchline: it's not how many agents you run. it's the determinism you build with verifiers, code fallbacks, and reality anchors.

I broke the same architecture down with Kimi K3. Full A-Z guide below.

[mikasa @tannn172]

Let's build the next CAT trend! 🐱🚀
@ASCIIcats_

[mikasa @tannn172]

@ASCIIcats_ something big is coming

[ASCII Cats (=^.^=) @ASCIIcats_]

success = community

help us summon more believers and spread the message. follow your fellow cats. more of us, louder we purr

light a candle, join the cult: https://t.co/PzHlitFE0t

new cult. new order. new success. tomorrow. =^.^=

[mikasa @tannn172]

@Zorathzzz you are legendary

0xB104eD986F051379D58A75F0F37f5f3f4B29435d

[Zora @Zorathzzz]

Okey, I made way too much from the market.

time to give some of it away.

yesterday i gave away $1,000.

today, let’s do it again.

$1,000 giveaway.
5 winners.
$200 each.

follow me, repost this, and leave your wallet in the comments.

easy money.

winners will be picked in 24 hours.

[mikasa @tannn172]

@cryptobeastreal let go

[Zealy @zealy_io]

We’re hyped to announce @MangoOS_Network community-building efforts here on Zealy!

Mango Network is a Layer 1 blockchain built on the Move language, aiming to be a transactional omni-chain infra network.

beyond their strong community-building efforts, we’re excited to work together on market promotion and dive into new possibilities for ecosystem integration. mango network is building something special, and we’re proud to support them every step of the way.

nice to see u here @MangoOS_Network 🔥

[mikasa @tannn172]

@zealy_io @MangoOS_Network boost up

[mikasa @tannn172]

@MangoOS_Network @zealy_io nice

[Mango Network @MangoOS_Network]

Been thinking about it for so long, and finally I'm working with Zealy now.

Nice to See U here @zealy_io

[Mango Network @MangoOS_Network]

To better involve everyone in our testnet, we've produced this video covering the important points to note for the initial operations.

Additionally, we've also completed the optimization of the points rules. Even if all your daily tasks are in a loading state, your daily points will still be recorded.

[Mango Network @MangoOS_Network]

We've got two quick tutorial videos to help with the front-end verification issues you've been facin.

Check them plz 👇

(don't mind the accent of the AI lol)

[mikasa @tannn172]

@MangoOS_Network wow

[mikasa @tannn172]

@MangoOS_Network @takeshi_kasai go