지석

[지석 @rockycloud77]

캠페인 참여 홈 :: 해피빈 https://t.co/yzzcyG8PjJ

[Nate Herk @nateherk]

Just dropped a 2+ hour free course on how to build an AI operating system with Claude Code.

I go over all the frameworks, all the things to think about, and I give away a free GitHub repo to get you started.

Give this a read.

[阿川 | AI thinking @AI_jacksaku]

微软又搞事情了。

一个低调的Python工具，悄悄爬到GitHub趋势榜第一。

10万+ stars。

它叫MarkItDown，干什么的？

把任何文件变成Markdown。

PDF、Word、PPT、Excel、图片...

扔进去，出来就是干净的Markdown。

听起来很简单对吧？

但你知道这解决了多大的痛点吗？

RAG系统最头疼的就是：怎么喂数据给AI。

PDF格式乱七八糟，Word藏满隐藏样式，

PPT全是图片没文字...

现在，一行代码搞定。

更狠的是，这是微软官方出品，

完全免费，开源，可商用。

我试了一下，一个200页的PDF，

3秒转完，格式干净得不像话。

这就是大厂的实力：

不做则已，一做就是行业标准。 https://t.co/wYANKoFiOl

[Tesla Europe, Middle East & Africa @teslaeurope]

De toekomst van mobiliteit is aangebroken

FSD Supervised has been approved in the Netherlands 🇳🇱 & will begin rolling out in the country shortly! 

Trained on billions of kilometers of real-world driving data, it can drive you almost anywhere under your supervision – from residential roads to city streets & highways

No other vehicle can do this. 

We're excited to bring FSD Supervised to more European countries soon

[lucas @lucas_flatwhite]

📱 Google AI Gallery + Gemma 4
https://t.co/AGXDEsTqnE

지금 무료로 다운로드 받으세요! 이건 구글에서 공식적으로 출시했고, 내 폰에서 Gemma 4를 직접 돌릴 수 있어요.

정말 좋은건 단순히 챗봇이 아니라 완전 오프라인으로 작동한다는 점이죠.

비행기 모드에서도, 데이터가 없어도 개인정보 걱정 없이 AI와 대화할 수 있습니다!

많은 분들이 클라우드 기반으로 AI를 쓰실텐데, 로컬에서 AI를 돌리는건 프라이버시, 속도, 비용 측면에서 장점이 많기도 하죠.

↓

🛠️ 먼저 Google AI Gallery 앱을 설치하세요!

iOS
https://t.co/TSAuewHEUI

Android
https://t.co/E9cvXkhbMm

완전한 100% 오픈소스이고, 어떤 데이터도 외부로 전송하지 않아요.

↓

⚡ 앱을 열어서 Gemma 4 모델을 다운 받으세요!

Gemma 4 는 이번에 구글에서 발표한 모델 시리즈예요.

E2B, E4B 모델은 온디바이스, 엣지 환경에 최적화된 모델이죠.

Gemma 4 E2B
- 보급형 스마트폰, IoT, 웹브라우저
- 극강의 효율성, 실시간 음성 처리

Gemma 4 E4B
- 플래그십 스마트폰, 노트북
- 균형 잡힌 성능, 복잡한 지시 이행

초경량화는 물론이고, 멀티모달도 지원해요. 굉장합니다 진짜!

텍스트뿐만 아니라 이미지, 오디오, 비디오 입력을 기본적으로 이해하구요.

특히 E2B와 E4B는 오디오 처리 능력이 뛰어나서 실시간 음성 인식이나 번역을 잘해요.

여기에 기본적으로 Thinking Mode 지원합니다! 에이전트 워크플로우에 최적화되어 있음..

다양하게 활용해보자구요!!!

[Andrej Karpathy @karpathy]

Wow, this tweet went very viral!

I wanted share a possibly slightly improved version of the tweet in an "idea file". The idea of the idea file is that in this era of LLM agents, there is less of a point/need of sharing the specific code/app, you just share the idea, then the other person's agent customizes & builds it for your specific needs.

So here's the idea in a gist format: https://t.co/NlAfEJjtJV

You can give this to your agent and it can build you your own LLM wiki and guide you on how to use it etc. It's intentionally kept a little bit abstract/vague because there are so many directions to take this in. And ofc, people can adjust the idea or contribute their own in the Discussion which is cool.

[lucas @lucas_flatwhite]

📋 Claude, ChatGPT, Gemini 위한 20가지 강력한 에이전트 스킬

"생산성을 높이기 위해 모든 AI 모델에 추가할 수 있는 20가지 강력한 에이전트 스킬을 큐레이션했습니다."

> 한국어 PDF
https://t.co/fE4tgUmtoZ

> 마크다운이 필요하시다면
https://t.co/C2Vc66EsGv

어디든 이 프롬프트를 복사해서 활용하거나 개선할 수 있어요. 스킬들은 5가지 카테고리로 분류했네요.

SCQA는 글쓰기나 프레젠테이션을 위한 커뮤니케이션 구조화 프레임워크 인데,,

이런 접근이 흥미로워서 정리해봤어요.

[CY₿ER⚡WALL @BTCcyberwall]

법제처 챗봇 만들어봤습니다.
대한민국의 모든 법령을 기반으로 작동합니다.
https://t.co/1ZyV6yFbvz

DB는 https://t.co/Mx4oCds38R 여기 있는 md파일로 만들었구 누락된 법령 1400여 건은 https://t.co/hEQ6BJ5MDJ 요걸로 채워넣었습니다

[lucas @lucas_flatwhite]

우오! 국가법령정보 MCP
https://t.co/24w8khxKrN
= 한국 법령 175,064건!

헌법부터 판례까지, AI가 한 줄이면 다 찾게 생겼네요.

민원 답변에 정확한 조문 번호를 넣어야 하는데, 법제처 사이트에서 클릭을 열 번은 해야 원하는 조항을 찾을 수 있었죠.

감사 지적을 받고 근거 법령을 급하게 뒤져야 할 때, 조례를 제개정하면서 상위법 위임 근거를 확인해야 할 때..

이런 순간마다 느끼는 비효율을 해결하기 위해 7년차 지방공무원이 퇴근 후 직접 만든 도구!!!

korean-law-mcp는 법제처 Open API를 감싸서 64개의 법률 도구를 하나의 MCP 서버로 제공해요!

다루는 범위가 상당히 넓네요?

헌법 (1), 법률 (1,706), 위임법령 (3,480), 행정규칙 (9,827), 자치법규 (158,863), 판례 (1,187)

그래서 총 175,064 ㄷㄷ

헌법부터 판례까지, 대한민국 법령 체계 전체를 검색하고 비교하고 분석할 수 있어요!

약칭도 잘 알아듣는다고 해요. 법률 실무에서는 약칭을 훨씬 많이 쓰니까요.

지방계약법, 개보법, 행기법, 전자정부법, 화관법 같은 약칭을 입력하면 자동으로 정식 명칭으로 변환해서 검색해줘요. 법제처에서 일일이 정식 명칭을 입력할 필요가 없음 ㄷㄷ

검색 뿐만 아니라 비교하고, 마크다운 변환하고, 워크플로우도 제공하니 아주 좋네요!

[Sawyer Merritt @SawyerMerritt]

NEWS: Dutch regulators (RDW), the key authority that would clear the path for Tesla FSD approval across much of Europe, has issued a response to @Tesla’s 𝕏 post today.

"We know that Tesla's application has the interest of many people. The RDW gets a lot of (media) questions about this. Given this great interest and the many speculations, we would like to give a short response to Tesla's message and request. Normally, we never make statements about requests from manufacturers because of the market-sensitive information.

Assessment process:
In the message, Tesla states that they are in the final phase of the assessment process. That's right, Tesla and the RDW are currently going through the final stages of the assessment process. About 18 months ago, the joint intensive testing program began. During the term and in this final phase, the RDW thoroughly looks at the test results and analyzes the data. During this final phase, our inspectors will review all data and test results and, after completion of this process, a decision will be made on the approval of the driver's assistance system FSD Supervised. For the RDW, (traffic) safety is paramount."

[Vishakha Singhal @vishisinghal_]

Most people think using Claude Code is about writing better prompts.

It’s not.

The real unlock is structuring your repository so Claude can think like an engineer.

If your repo is messy, Claude behaves like a chatbot.

If your repo is structured, Claude behaves like a developer living inside your codebase.

Your project only needs 4 things:

• the why → what the system does
• the map → where things live
• the rules → what’s allowed / forbidden
• the workflows → how work gets done

I call this:

The Anatomy of a Claude Code Project 👇

━━━━━━━━━━━━━━━

1️⃣ CLAUDE.md = Repo Memory (Keep it Short)

This file is the north star for Claude.

Not a massive document.

Just three things:

• Purpose → why the system exists
• Repo map → how the project is structured
• Rules + commands → how Claude should operate

If CLAUDE.md becomes too long, the model starts missing critical signals.

Clarity beats size.

━━━━━━━━━━━━━━━

2️⃣ .claude/skills/ = Reusable Expert Modes

Stop repeating instructions in prompts.

Turn common workflows into reusable skills.

Examples:

• code review checklist
• refactoring playbook
• debugging workflow
• release procedures

Now Claude can switch into specialized modes instantly.

Result:

More consistent outputs across sessions and teammates.

━━━━━━━━━━━━━━━

3️⃣ .claude/hooks/ = Guardrails

Models forget.

Hooks don’t.

Use hooks for things that must always happen automatically.

Examples:

• run formatters after edits
• trigger tests after core changes
• block sensitive directories (auth, billing, migrations)

Hooks turn AI workflows into reliable engineering systems.

━━━━━━━━━━━━━━━

4️⃣ docs/ = Progressive Context

Don’t overload prompts with information.

Instead, let Claude navigate your documentation.

Examples:

• architecture overview
• ADRs (engineering decisions)
• operational runbooks

Claude doesn’t need everything in memory.

It just needs to know where truth lives.

━━━━━━━━━━━━━━━

5️⃣ Local CLAUDE.md for Critical Modules

Some areas of your system have hidden complexity.

Add local context files there.

Example:

src/auth/CLAUDE.md
src/persistence/CLAUDE.md
infra/CLAUDE.md

Now Claude understands the danger zones exactly when it works in them.

This dramatically reduces mistakes.

━━━━━━━━━━━━━━━

Here’s the shift most people miss:

Prompting is temporary.

Structure is permanent.

Once your repository is designed for AI:

Claude stops acting like a chatbot...

…and starts behaving like a project-native engineer. 🚀

[Tesla Korea @tesla_korea]

한국 테슬라 오너들이 단 1개월 만에 FSD(감독형)으로 누적 주행거리 100만 km를 돌파했습니다.
이는 대한민국을 약 480바퀴 돌고도 남는 거리입니다!

*섬을 제외한 공식 해안선 길이와 북쪽 국경 기준, 1바퀴당 약 2,413 km

Korea Tesla owners have surpassed 1 million km of cumulative driving distance with FSD (Supervised) in just one month.
This distance is enough to circle the entire country of South Korea approximately 480 times—with some to spare!

*Based on the official coastline length (excluding islands) and the northern border, one full lap is approximately 2,413 km.

[이소영 국회의원 (Soyoung Lee) @im_soyounglee]

드디어 오늘, 서울에서 테슬라 FSD 체험 했습니다.

JiDal Papa님의 모델S 협찬에 힘입어^^ 파파님 정말 감사합니다.

국회 -> 망원시장 -> 홍익대 -> 국회 복귀 코스였고요.

이미 무인 로보택시를 타봐서 그런지 신기함은
덜했지만, 웬만한 사람만큼 운전을 잘하네요.

이미 완성된 기술이라고 느껴져 생각이 많아집니다. 앞으로 실제 보급이 확산되면 우리 삶의 모습이 많이 달라질 것 같습니다.

장롱면허인 저도 굳이 수동차 운전을 배울 이유가 없겠다 싶네요.

[Elon Musk @elonmusk]

Starlink now available in South Korea 🇰🇷

[Tsla Chan @Tslachan]

$TSLA 🇰🇷
BREAKING : I just experienced FSD (Supervised) for the first time in Korea.

The vehicle is Model X and the FSD version is *V14.1.4.

We have yet to see any official cases of V14 being applied outside of the United States and Canada.

I've just completed my first FSD run for about an hour.
(I've experienced about 10,000km of FSD in the U.S.)

Except for the parking lot, there was never a case where I intervened during the FSD drive.

The first impression of the FSD that works in Korea is that it is very amazing and a great achievement of the Tesla AI team.

Although there was one case where I went off the road without following the navigation, the FSD thought to itself and drove on the right path.

The overall experience in Korea was very good, such as turning left, turning right, U turn, joining lanes, and changing lanes.

It's kind of like... an American driver driving in Korea on an international driver's license.

If more data is generated in the future and the number of users increases, higher achievement of FSD in Korea is very much expected.

I can't wait for more people to experience FSD in Korea!
(Tesla Korea is a HW4 vehicle that purchased FSD and will update the FSD (Supervised) in the future for vehicles suitable for VIN number 5or7.

[Tesla Korea @tesla_korea]

FSD 감독형
Next Destination: Korea
Coming Soon

[Palantir @PalantirTech]

Seoul 🇰🇷

[Elon Musk @elonmusk]

Starship

[Tesla Japan @teslajapan]

FSD (Supervised)のテスト走行を本格開始

*国内リリース時期は、弊社開発状況及び規制当局の許認可に依存します。