반응형
블로그 이미지
개발자로서 현장에서 일하면서 새로 접하는 기술들이나 알게된 정보 등을 정리하기 위한 블로그입니다. 운 좋게 미국에서 큰 회사들의 프로젝트에서 컬설턴트로 일하고 있어서 새로운 기술들을 접할 기회가 많이 있습니다. 미국의 IT 프로젝트에서 사용되는 툴들에 대해 많은 분들과 정보를 공유하고 싶습니다.
솔웅

최근에 올라온 글

최근에 달린 댓글

최근에 받은 트랙백

글 보관함

카테고리


반응형

Study Group 에 Join 하면서 Obsidian 이라는 것을 알게 되었는데 이게 배우면 배울수록 흥미롭네요.
모토가 A second brain 인데 AI 랑 합해 지면 진짜 나의 두 번째 두뇌가 될 수 있을것 같습니다.
Study Group 에 참여하면서 공부하는 내용을 공유 합니다.
여러분들도 흥미로우실 거예요.

🎓 Study Group 발표 자료 – Obsidian과 RAG 연동 방법 리서치 결과 발표 – PKM Project | My AI 2nd Brain 🧠🤖📚

이 영상은 시애틀 지역에서 활동 중인 PKM(Personal Knowledge Management) Study Group의 발표 자료로 제작된 콘텐츠입니다.
우리는 AI를 활용하여 나만의 제2의 뇌, 즉 AI 기반의 Second Brain을 만드는 것을 목표로 함께 공부하고 있습니다.

https://youtu.be/VyWi3C06Tvw?si=zlU5F28GTc0zbM7A

 

📌 본 영상에서는 다음과 같은 내용을 다룹니다:

🔹 Obsidian과 RAG(Retrieval-Augmented Generation)의 연동 방법
🔹 Standalone App vs. Plug-in 방식의 구현 차이
🔹 Local 방식과 Web 기반 방식의 비교
🔹 MCP, Claude, SuperWhisper를 활용한 자동 글쓰기 환경 구축 사례
🔹 다양한 Obsidian AI Plug-in 비교 분석 및 실습 대상 선정
🔹 향후 개발 방향과 전략 정리

📂 본 발표에 사용된 Google Slide 링크입니다:

🇰🇷 Korean Version:
https://docs.google.com/presentation/d/1T221IQ-6LllyzgIro3nsPfNxT41lK8XYRW4h59Xsjr4/edit?usp=sharing

 

PKM Project - RAG

RAG(Retrieval-Augmented Generation)의 이해와 Obsidian 적용 PKM (Personal Knowledge Management) Project https://docs.google.com/presentation/d/1VzvJ8-3NuqEcSClob9c9xqcfC4S9SF2m7iaSGqrP4iA/edit?usp=sharing https://changbal.slack.com/archives/C0566785M5

docs.google.com

🌎 English Version:
https://docs.google.com/presentation/d/1ZkUxJcmR9CeyR4X6RPbJP6khQ_PYYduMikvquFshNXY/edit?usp=sharing

 

[ENG] PKM Project - RAG

RAG(Retrieval-Augmented Generation)의 이해와 Obsidian 적용 PKM (Personal Knowledge Management) Project https://docs.google.com/presentation/d/1VzvJ8-3NuqEcSClob9c9xqcfC4S9SF2m7iaSGqrP4iA/edit?usp=sharing https://changbal.slack.com/archives/C0566785M5

docs.google.com

그리고 Youtube에서 최근 시작한 더빙 기능 덕분에 한국어로 만들어서 올려도 다른 나라말로 유투브가 자동으로 번역해서 음성으로 들려줘서 언어에 상관없이 더 많은 사람들에게 영상을 보여 줄 수 있어서 정말 좋아졌습니다.

🎓 Study Group Presentation – Research on Integrating Obsidian with RAG – PKM Project | My AI 2nd Brain 🧠🤖📚
This video is a presentation created for the PKM (Personal Knowledge Management) Study Group based in the Seattle area.
Our group studies how to build a personalized “Second Brain” using AI, aiming to enhance knowledge management and productivity.
📌 This presentation covers the following topics:
🔹 How to integrate Obsidian with RAG (Retrieval-Augmented Generation)
🔹 Differences between Standalone Applications and Plug-in-based implementations
🔹 Comparison of Local and Web-based approaches
🔹 Case study: Building an automated writing environment using MCP, Claude, and SuperWhisper
🔹 Analysis and selection of Obsidian AI plug-ins for hands-on practice
🔹 Future development strategies and direction

 

 

반응형


반응형

🚀 이번 주 Vibe Coding 라방 안내

이번 주 라방에서는 아래와 같은 작업을 하겠습니다:

🔁 Prompt-Driven Restart 방법론 정리!
📧 Feature Suggestion email 발송 결과 공유
🧑‍🏫 지난 시간 개발 된 영어 회화 Tutor 앱 기능 설명
🛠️ 영어 회화 Tutor 앱 기능 개선하기

🕔 방송 일정
📅 2025년 6월 29일 (일)
🕘 오후 9시 (KST, 한국) / 🕔 오전 5시 (PST, 시애틀)

🔔라방 알림 설정 : https://www.youtube.com/live/VgdZ3pA2TlA?si=eJDwbJHnDRZLI0kn


 

 

📂 GitHub Repository
▶️ https://github.com/solkit70/CUA_VibeCoding_02/tree/develop

 

GitHub - solkit70/CUA_VibeCoding_02

Contribute to solkit70/CUA_VibeCoding_02 development by creating an account on GitHub.

github.com

📣 라이브 방송에 참여해서 말만 하세요!
✨ 여러분의 아이디어가 실시간으로 구현될 수도 있습니다!

🔥 부담 없이 함께 해요!
🤖 Vibe Coding은 AI에게 코딩을 시키는 것이 아니라,
👥 AI와 함께 호흡하며 배우고 실험하는 새로운 방식입니다.
🙌 지금 바로 함께해 보세요!

 

반응형


반응형

Dear Vibe Coding Tool Providers,

 

My name is Changsoo Park, and I’m based in Redmond, Washington.

I’ve been running a YouTube channel where I host weekly livestream sessions on Sundays, focused on learning and experimenting with Vibe Coding—mainly with a Korean-speaking audience.

Over the past three months, through regular Vibe Coding sessions, I’ve gained a lot of practical insight and experience. Based on this hands-on learning, I’ve come up with a few ideas that I believe could significantly improve the effectiveness and usability of Vibe Coding tools like GitHub Copilot or Cursor AI.

I'd like to briefly share two of them with you today.

 

안녕하세요,
저는 미국 워싱턴주 레드먼드에 거주하는 박창수라고 합니다.

저는 현재 유튜브 채널을 운영하고 있으며, 올해 3월부터 매주 일요일마다 ‘Vibe Coding’을 주제로 한국어로 진행되는 실시간 라이브 방송을 진행하고 있습니다.
그동안 꾸준히 Vibe Coding을 실험하고 학습해 오면서, 실제 현장에서 느낀 문제점들과 함께 툴이 보완되면 훨씬 실용적일 수 있는 기능 아이디어들을 떠올리게 되었습니다.

오늘은 그 중 두 가지 아이디어를 공유드리고자 합니다.

 


1. Regression Feature Awareness in Vibe Coding Tools 🧪

One common issue I’ve encountered during Vibe Coding sessions is that adding new features often unintentionally breaks previously working ones. This is a typical regression problem well-known in software engineering.

In traditional QA workflows—something I’m familiar with from my 25 years in tech, including 7 years as a QA engineer—Regression Testing is a standard step. After developing new features, previously tested functionalities are re-tested to ensure they still work correctly.

Currently, most Vibe Coding tools don’t offer any sort of regression verification process. As a result, applying Vibe Coding in real production-level environments remains quite limited.

If regression awareness and testing support could be integrated into Vibe Coding tools—possibly powered by AI—it would make the development process more reliable and sustainable.

I believe this is technically feasible, especially since modern QA practices have long focused on automation. Incorporating even a lightweight regression tracking mechanism in Vibe Coding tools could greatly enhance their practical utility.

 

Vibe Coding으로 어플리케이션을 개발하다 보면 새 기능을 추가했을 때 이전에 잘 동작하던 기능이 갑자기 제대로 동작을 하지 않는 경우가 빈번하게 발생합니다.
이는 전통적인 소프트웨어 개발에서도 잘 알려진 ‘회귀(regression)’ 문제입니다.

저는 IT 업계에서 약 25년간 일했으며, 그 중 7년은 QA 엔지니어로 일했습니다. QA 업무의 중요한 프로세스 중 하나가 바로 Regression Test입니다.
새 기능을 개발한 후, 기존 기능들이 여전히 잘 작동하는지 확인하는 테스트 단계죠.

현재 대부분의 Vibe Coding 도구에는 이러한 회귀 검증 기능이나 흐름이 포함되어 있지 않아, 실제 현업에서 사용하기에는 한계가 존재합니다.
만약 AI 기반 Vibe Coding 툴이 기존 기능의 안정성을 확인해줄 수 있는 회귀 검증 기능을 제공한다면, 더 실용적이고 신뢰성 있는 개발 환경이 될 수 있다고 생각합니다.

기술적으로도 이는 충분히 가능하다고 생각합니다. 기존 QA 자동화 시스템에서도 회귀 테스트는 이미 자동화가 활발히 이루어지고 있으며, 이를 Vibe Coding 툴에도 접목할 수 있을 것입니다.


2. Prompt-Driven Restart: A Refactoring Alternative 🔁

The second idea comes from repeated observations while building applications via Vibe Coding.

Often, when errors occur during development, users try to fix them by tweaking code or asking the AI to debug. But this can become inefficient and drain energy—contrary to the core philosophy of Vibe Coding: low-effort, high-leverage coding.

Instead, I started experimenting with restarting from scratch using the original prompt, slightly modified to avoid previous errors. This approach often yields better, cleaner results. I call this method Prompt-Driven Restart.

 

두 번째로 소개드릴 아이디어는 제가 직접 이름 붙인 개념인 Prompt-Driven Restart입니다.

Vibe Coding을 하다 보면 새로운 기능 개발 중 오류가 발생하고, 이를 고치려는 시도를 반복하는 경우가 많습니다.
하지만 프롬프트만으로 해결되지 않는 경우에는 AI가 만든 소스코드를 분석하거나 유도 질문을 통해 문제 해결을 시도해야 합니다.

이러한 과정은 많은 에너지를 소모하게 되며, 오히려 Vibe Coding의 간결하고 효율적인 코딩이라는 장점이 희석되는 경우가 많습니다.

그래서 저는 기존 소스코드를 계속 수정하기보다는, 처음에 사용했던 프롬프트를 보완하여 새로 시작하는 방식을 사용해 보았습니다.
그 결과, 훨씬 효율적이고 오류도 적은 결과를 얻을 수 있었고, 이 방법론에 Prompt-Driven Restart라는 이름을 붙였습니다.

 

To support this methodology, I suggest the following features:

 

이 방식을 Vibe Coding 툴에서 보다 쉽게 활용할 수 있도록, 아래와 같은 기능이 지원되었으면 합니다:

  • 🗃️ Prompt History Management: The ability to save and easily retrieve prompts previously used in a project.

  • 🔖 Restart Point Markers: A feature to mark key restart points in a session, allowing users to jump back to a clean slate when necessary.

  • ✍️ Prompt Patching Interface: An easy way to modify a prior prompt by adding clarifications or constraints to prevent known issues.

  • 🗃️ 프롬프트 히스토리 관리 기능: 지금까지 사용한 프롬프트를 저장하고 쉽게 불러올 수 있도록
  • 🔖 재시작 포인트 지정 기능: 특정 시점에서 다시 시작할 수 있도록 체크포인트 설정
  • ✍️ 프롬프트 보완 도우미: 오류가 발생하지 않도록 이전 프롬프트에 쉽게 지시문을 추가할 수 있는 인터페이스

Together, these features would empower users to avoid long debug cycles and instead take advantage of Vibe Coding’s core strengths: fast iteration, flexible creativity, and low cognitive overhead.

 

이러한 기능들이 더해진다면, 사용자는 반복된 디버깅에 갇히지 않고 보다 전략적이고 창의적인 방향으로 Vibe Coding을 활용할 수 있게 될 것입니다.


To give you context, I document my progress and share it with my audience via Google Slides and YouTube livestreams. Most of my content is in Korean, but I’ve provided the links below for your reference. They should translate well via built-in tools.

 

제가 지금까지 진행해 온 Vibe Coding 실험은 Google Slide로 정리해 두었으며, YouTube에 관련 방송도 아카이빙 해두었습니다.
모두 한국어로 되어 있지만, 요즘은 번역도 쉬워졌기에 함께 공유드립니다:

 

 

Vibe Coding

 

www.youtube.com

 

 

Vibe Coding

Vibe Coding 재미로 하는 Vibe Coding

docs.google.com

 

I’m sharing these ideas in the hope that they may help push Vibe Coding tools further—making them more powerful, reliable, and user-friendly. I would be happy to share more insights in the future if helpful.

 

Vibe Coding이라는 새로운 개념과 방식이 더 발전하고, 실제 현장에서도 널리 사용될 수 있기를 진심으로 바라며,
작은 경험이나마 의미 있는 피드백이 되었기를 바랍니다.

 

Thank you very much for your time and the incredible tools you’ve built. I look forward to your thoughts.

 

앞으로도 계속 좋은 아이디어가 떠오르면 공유드리겠습니다.

 

Warm regards,
Changsoo Park
Redmond, WA
YouTube Channel: Catch Up AI

 



반응형