Curiosity and Experience never get old. https://madplay.github.io/ Sun, 26 Apr 2026 14:39:43 +0900 Sun, 26 Apr 2026 14:39:43 +0900 Jekyll v3.10.0 Agent Teams, where multiple Claude Code instances divide tasks and exchange messages. Most of the time, subagents suffice, but there are moments when you need a team. Sun, 01 Mar 2026 21:06:00 +0900 https://madplay.github.io/en/post/claude-code-agent-teams https://madplay.github.io/en/post/claude-code-agent-teams claude-code agent-teams multi-agent collaboration AI 여러 Claude Code 인스턴스가 태스크를 나누고 메시지를 주고받는 Agent Teams. 대부분은 서브에이전트로 충분하지만, 팀이 필요한 순간이 있다. Sun, 01 Mar 2026 21:06:00 +0900 https://madplay.github.io/post/claude-code-agent-teams https://madplay.github.io/post/claude-code-agent-teams claude-code agent-teams multi-agent collaboration AI Why does Claude Code spin up subagents instead of doing everything alone? What are the roles of the three built-in agents, and how do you build custom ones? Sun, 22 Feb 2026 08:52:00 +0900 https://madplay.github.io/en/post/claude-code-subagent-architecture https://madplay.github.io/en/post/claude-code-subagent-architecture claude-code subagent agent-architecture worktree parallel AI Claude Code는 왜 혼자 다 하지 않고 서브에이전트를 띄울까? 내장 에이전트 세 종류는 각각 어떤 역할이고, 커스텀 에이전트는 어떻게 만들까? Sun, 22 Feb 2026 08:52:00 +0900 https://madplay.github.io/post/claude-code-subagent-architecture https://madplay.github.io/post/claude-code-subagent-architecture claude-code subagent agent-architecture worktree parallel AI Let's look at the concept of harness engineering, why it emerged, and how it differs from prompt engineering and context engineering. Sun, 15 Feb 2026 21:14:37 +0900 https://madplay.github.io/en/post/harness-engineering https://madplay.github.io/en/post/harness-engineering harness-engineering context-engineering prompt-engineering ai-agent AI 하네스 엔지니어링(Harness Engineering)의 개념과 등장 배경, 그리고 프롬프트 엔지니어링과 컨텍스트 엔지니어링과의 차이를 알아보자! Sun, 15 Feb 2026 21:14:37 +0900 https://madplay.github.io/post/harness-engineering https://madplay.github.io/post/harness-engineering harness-engineering context-engineering prompt-engineering ai-agent AI Just by prepending 'rtk' to a command, the agent's token consumption was reduced by 80%. We explore the operational mechanics of the open-source CLI proxy RTK (Rust Token Killer), how to install it, and where it hits its limits. Sun, 08 Feb 2026 22:15:00 +0900 https://madplay.github.io/en/post/rtk-reduce-ai-coding-agent-token-usage https://madplay.github.io/en/post/rtk-reduce-ai-coding-agent-token-usage rtk token-optimization ai-coding-agent cli-proxy AI 명령어 앞에 rtk 하나 붙였을 뿐인데 에이전트의 토큰 소비가 80% 줄었다. 오픈소스 CLI 프록시 RTK(Rust Token Killer)의 동작 원리와 설치법, 어디서 한계에 부딪히는지까지 함께 살펴본다. Sun, 08 Feb 2026 22:15:00 +0900 https://madplay.github.io/post/rtk-reduce-ai-coding-agent-token-usage https://madplay.github.io/post/rtk-reduce-ai-coding-agent-token-usage rtk token-optimization ai-coding-agent cli-proxy AI What is OpenClaw, which has captured the attention of developers in a short period of time? Sun, 01 Feb 2026 11:32:18 +0900 https://madplay.github.io/en/post/openclaw-overview-and-usage https://madplay.github.io/en/post/openclaw-overview-and-usage openclaw ai-agent gateway llm self-hosted AI 짧은 시간에 개발자들의 시선을 사로잡은 오픈클로(OpenClaw)는 무엇일까? Sun, 01 Feb 2026 11:32:18 +0900 https://madplay.github.io/post/openclaw-overview-and-usage https://madplay.github.io/post/openclaw-overview-and-usage openclaw ai-agent gateway llm self-hosted AI How can powerful AI destabilize society? This post reviews the five critical risks Dario Amodei identifies in 《The Adolescence of Technology》. Sat, 31 Jan 2026 22:54:03 +0900 https://madplay.github.io/en/post/dario-amodei-the-adolescence-of-technology https://madplay.github.io/en/post/dario-amodei-the-adolescence-of-technology ai anthropic dario-amodei llm agi governance AI 강력한 AI는 어떻게 사회를 흔들까? 다리오 아모데이(Dario Amodei)의 《기술의 사춘기》에서 꼽은 5가지 치명적 위험. Sat, 31 Jan 2026 22:54:03 +0900 https://madplay.github.io/post/dario-amodei-the-adolescence-of-technology https://madplay.github.io/post/dario-amodei-the-adolescence-of-technology ai anthropic dario-amodei llm agi governance AI We examine the structure and core APIs of the Claude Agent SDK, exploring how to control agents with custom tools and hooks. Fri, 09 Jan 2026 20:34:00 +0900 https://madplay.github.io/en/post/claude-agent-sdk-tutorial https://madplay.github.io/en/post/claude-agent-sdk-tutorial claude-agent-sdk ai-agent mcp typescript anthropic AI Claude Agent SDK의 구조와 핵심 API를 살펴보고, 커스텀 도구와 훅으로 에이전트를 제어하는 방법을 알아본다. Fri, 09 Jan 2026 20:34:00 +0900 https://madplay.github.io/post/claude-agent-sdk-tutorial https://madplay.github.io/post/claude-agent-sdk-tutorial claude-agent-sdk ai-agent mcp typescript anthropic AI Cursor is an IDE. Claude Code is a terminal agent. Same model, different results. Here is why. Tue, 30 Dec 2025 08:14:00 +0900 https://madplay.github.io/en/post/cursor-vs-claude-code https://madplay.github.io/en/post/cursor-vs-claude-code cursor claude-code ai-coding-tool comparison ide AI Cursor는 IDE, Claude Code는 터미널 에이전트. 같은 모델인데 결과가 다른 이유를 짚어 본다. Tue, 30 Dec 2025 08:14:00 +0900 https://madplay.github.io/post/cursor-vs-claude-code https://madplay.github.io/post/cursor-vs-claude-code cursor claude-code ai-coding-tool comparison ide AI Isolate recurring rules and procedures into reusable skills. Learn the structure and authoring points of Agent Skills. Sun, 28 Dec 2025 14:22:07 +0900 https://madplay.github.io/en/post/ai-agent-skills-and-skill-md https://madplay.github.io/en/post/ai-agent-skills-and-skill-md agent-skills skill-md claude-code cursor codex ai-coding AI 매번 다시 설명하던 규칙과 절차, 스킬로 분리하면 재사용이 쉬워진다. Agent Skills의 구조와 작성 포인트를 알아보자! Sun, 28 Dec 2025 14:22:07 +0900 https://madplay.github.io/post/ai-agent-skills-and-skill-md https://madplay.github.io/post/ai-agent-skills-and-skill-md agent-skills skill-md claude-code cursor codex ai-coding AI From permission blocking to environment variable management, learn how to build safe execution boundaries by separating team security rules from personal local settings. Fri, 17 Oct 2025 23:11:00 +0900 https://madplay.github.io/en/post/claude-code-settings-json https://madplay.github.io/en/post/claude-code-settings-json claude-code settings-json permissions hooks security AI 권한 차단부터 환경 변수 관리까지, 팀의 보안 규칙과 개인의 로컬 환경을 분리하여 안전하게 실행 경계를 구축하는 기준을 알아보자! Fri, 17 Oct 2025 23:11:00 +0900 https://madplay.github.io/post/claude-code-settings-json https://madplay.github.io/post/claude-code-settings-json claude-code settings-json permissions hooks security AI Running /init first in Claude Code anchors project context in CLAUDE.md, eliminating repeated explanations and inconsistent first responses across sessions. Sat, 04 Oct 2025 21:00:00 +0900 https://madplay.github.io/en/post/why-init-first-in-claude-code https://madplay.github.io/en/post/why-init-first-in-claude-code claude-code init claude-md ai-coding context AI Claude Code에서 /init을 먼저 실행해야 하는 이유는 프로젝트 문맥을 CLAUDE.md로 고정하기 위해서다. 반복 설명과 흔들리는 첫 응답이 왜 줄어드는지 정리했다. Sat, 04 Oct 2025 21:00:00 +0900 https://madplay.github.io/post/why-init-first-in-claude-code https://madplay.github.io/post/why-init-first-in-claude-code claude-code init claude-md ai-coding context AI Let’s review the key changes in JDK 25, the fifth LTS Java release, based on the official release notes. Fri, 19 Sep 2025 21:32:10 +0900 https://madplay.github.io/en/what-is-new-java-25 https://madplay.github.io/en/what-is-new-java-25 java jdk25 openjdk Java/Kotlin 다섯 번째 LTS 자바 버전인 JDK 25 버전의 주요 변경사항을 공식 릴리즈 기준으로 살펴보자! Fri, 19 Sep 2025 21:32:10 +0900 https://madplay.github.io/post/what-is-new-java-25 https://madplay.github.io/post/what-is-new-java-25 java jdk25 openjdk Java/Kotlin Longer sessions slow down due to the context window. Learn how to manage token usage with /compact, /clear, and /cost. Mon, 25 Aug 2025 22:09:31 +0900 https://madplay.github.io/en/post/claude-code-context-and-token-optimization https://madplay.github.io/en/post/claude-code-context-and-token-optimization claude-code context token compact session AI 세션이 길어질수록 느려지는 이유는 컨텍스트 윈도우에 있다. /compact, /clear, /cost로 토큰 사용량을 관리하는 방법을 알아본다. Mon, 25 Aug 2025 22:09:31 +0900 https://madplay.github.io/post/claude-code-context-and-token-optimization https://madplay.github.io/post/claude-code-context-and-token-optimization claude-code context token compact session AI Explains context engineering, why it emerged, and how it differs from prompt engineering. Tue, 01 Jul 2025 22:39:00 +0900 https://madplay.github.io/en/post/context-engineering https://madplay.github.io/en/post/context-engineering context-engineering prompt-engineering llm rag AI 컨텍스트 엔지니어링은 LLM에 무엇을 넣고 무엇을 버릴지 설계하는 작업이다. 등장 배경과 프롬프트 엔지니어링과의 차이를 알아보자. Tue, 01 Jul 2025 22:39:00 +0900 https://madplay.github.io/post/context-engineering https://madplay.github.io/post/context-engineering context-engineering prompt-engineering llm rag AI With MCP, Claude Code can work with external systems as well. This post explains how to connect an MCP server. Sat, 21 Jun 2025 20:41:09 +0900 https://madplay.github.io/en/post/claude-code-mcp-setup https://madplay.github.io/en/post/claude-code-mcp-setup claude-code mcp anthropic automation AI MCP를 활용하면 Claude Code가 외부 시스템까지 다룰 수 있다. MCP 서버 연결 방법을 알아보자! Sat, 21 Jun 2025 20:41:09 +0900 https://madplay.github.io/post/claude-code-mcp-setup https://madplay.github.io/post/claude-code-mcp-setup claude-code mcp anthropic automation AI