만들며 배운 것을 남깁니다.

Category

Search

1
개발할 때는 Next.js, 실행할 때는 FastAPI 하나만 남기기
Python 백엔드에 웹 화면을 붙이면 개발 중에는 서버 두 개가 자연스럽습니다. Next.js 개발 서버가 화면 변경을 반영하고 FastAPI가 데이터를 처리합니다. 그런데 사용자가 로컬 도구를 실행할 때도 Node 서버와 Python 서버를 각각 켜야 할까요? 화면을 정적 파일로 만들 수 있다면 배포 후에는 FastAPI가 화면과 API를 같은 주소에서 제공하도록 구성할 수 있습니다. 개발 환경의 편의를 유지하면서 실행할 서버를 줄이는 선택입니다. 다만 개발 서버가 해 주던 일을 무엇이 대신할지 명확히 나눠야 합니다. 개인 사진 분류 프로젝트의 웹 실행 경로는…
- python
- javascript
KO2026-10-29
2
한글 입력 중에만 단축키가 안 된다면, 키와 문자를 따로 보세요
영문 입력 상태에서는 작동하던 ⌘I가 한글 입력 상태에서만 조용해집니다. 같은 키를 눌렀고 Command도 누르고 있으니 단축키 설정을 다시 저장하거나 앱을 재시작하고 싶어집니다. 하지만 앱이 비교하고 있는 값이 i가 아니라 ㅑ라면 설정을 다시 저장해도 맞아떨어지지 않습니다. macOS의 입력 이벤트에는 키 코드, 입력 문자, 수정 키를 일부 제외한 문자처럼 서로 다른 값이 들어 있습니다. 단축키 문제를 좁히려면 먼저 앱이 어느 값을 키 이름으로 쓰는지 확인해야 합니다. Apple은 NSEvent.charactersIgnoringModifiers를 Shift를…
KO2026-10-28
3
파일은 그대로인데 결과는 달라진다: 캐시에 처리 버전이 필요한 이유
이미지 파일의 SHA-256이 어제와 같으면 오늘도 같은 분석 결과를 써도 될까요? 파일이 바뀌었는지 확인하는 데는 도움이 됩니다. 하지만 모델을 바꾸거나 판정 기준을 조정했다면 같은 입력에서도 결과가 달라질 수 있습니다. 캐시를 재사용하려면 “같은 파일인가?”와 함께 **“같은 방식으로 처리했는가?”**를 물어야 합니다. 이 두 조건을 하나의 해시가 모두 설명한다고 생각하면 설정을 고쳤는데도 이전 결과가 남는 이유를 찾기 어려워집니다. 예를 들어 점수 0.7을 얻은 항목을 임계값 0.6에서는 통과시키고 0.8에서는 제외하는 처리기를 생각해 봅시다. 입력과 점수가…
- caching
- python
KO2026-10-27
4
Vitest Is Collecting Unrelated Tests? Start with the File List
If Vitest reports a failure in a test file you do not recognize, inspect the file path before changing assertions, timeouts, or worker settings. A test collected from a tool's cache needs a different fix from a failing test in your application. The useful question is specific: which files did this invocation discover,…
EN2026-10-23
5
One Repo, Two Agents, No ~/.claude: 5 Rules for Making Claude Code and Codex Behave the Same
What I found when I tried to make a repository carry its own agent rules for both runtimes My agents are well trained, as long as they run on my Mac. On 2026-10-01 my ~/.claude held a 22,449-byte CLAUDE.md, 12 rule files, 7 agents, hooks on five events and 24 enabled plugins. My ~/.codex held a 31,128-byte AGENTS.md,…
EN2026-10-22
6
Your Global Claude Code Skills Shouldn't Know Your Framework
They should know how to ask your repo — five times mine knew too much On 2026-06-12 I rewrote three skills in my global Claude Code folder, because one of my own projects was living inside them. Not a reference to it. The project itself. My semantic-commit skill, which I had added on 2026-04-19, told the agent to…
EN2026-10-22
7
7 Silent Failures in My Flutter App That No Test Caught
A token refresh that never ran, a crash filter that threw away crashes, and a test that defended a bug On 2026-10-03, four tests for a review screen in the Flutter app I maintain started failing. They couldn't find a snackbar. The snackbar was fine. The real error was this: The app decides whether it's running under…
EN2026-10-21
8
I Vendored Two Third-Party Claude Code Skills. A 49-Day Census Found Zero Calls.
A counter-list to the "plugins every developer should install" posts: five things my own transcripts taught me My feed keeps serving me lists of Claude Code plugins and skills I apparently cannot live without. I read them. I also install things: at my 2026-09-03 census, my global skills folder held 36 user skills. So…
EN2026-10-21
9
4 Things Claude Code Subagents Don't Inherit (and 1 Thing They Do)
What I learned when my own subagents ignored my rules, skipped my gates, refused my approval, and read my answer key On 2026-10-02 I was testing a new Claude Code skill by running it through subagent scenarios. One of the subagents did exactly what the skill told it to do: it stopped at the plan-approval gate, showed…
EN2026-10-20
10
pnpm 12 Quietly Blinds Your Dependency Scanner (and the Fix Has a Missing Step)
Your lockfile still pins the vulnerable versions, and your scanner still says "No issues found" On 2026-10-04 I was moving a repository I maintain from Yarn Classic to pnpm 12.9.1. Boring work: delete three yarn.lock files, pin "packageManager": "pnpm@12.9.1" in each package.json, regenerate, update the setup docs. By…
EN2026-10-20
11
7 Flutter App Patterns Worth Copying Into Your Next App
What a survey of my Riverpod + go_router app says to keep, and what to fix before you paste it My test harness has a doc comment that says (in Korean) it initializes ProviderScope, ScreenUtilInit and EasyLocalization. It does not wrap anything in EasyLocalization. Screen tests that use it had been running without…
EN2026-10-19
12
Your Claude Code Skills Are Invisible Past 1% of Context: 5 Checks Before You Rewrite a Description
A skill that never fires may not have a weak description. It may have no description at all. You write a skill. You give it a careful description. Then you wait for Claude to reach for it, and it never does. The natural next step is to rewrite the description: make it sharper, add trigger words, maybe add an…
EN2026-10-19