Merge pull request #2 from claude-code-expert/develop

docs : 수정사항 업데이트

Author: villainscode

CHANGELOG.md

@@ -2,6 +2,20 @@
 
 All notable changes to this project will be documented in this file.
 
+## [1.3.1] - 2026-04-11
+
+### Changed
+
+- **README restructure**: `README.md` → Korean (default), `README.en.md` → English (previously `README.ko.md` removed)
+- **Hook mechanism docs**: Added "동작 원리 / How It Works" section explaining UserPromptSubmit context injection flow
+- **Agent tool reference**: Updated all docs from "Task tool" to "Agent tool" to match current Claude Code API
+- **Version section**: Replaced token economics with version history table in both READMEs
+
+### Fixed
+
+- **shellcheck SC2221/SC2222**: Reordered glob patterns — `*"문서화"*` before `*"문서"*` to prevent unreachable pattern
+- **shellcheck SC1071**: Excluded zsh script (`subagent-chain.sh`) from CI shellcheck (unsupported shell)
+
 ## [1.3.0] - 2026-04-01
 
 ### Added

README.en.md

@@ -73,6 +73,20 @@ Subagents are invoked via Claude Code's built-in **Agent tool** — not by runni
    - Subagent context is discarded
 ```
 
+### Version
+
+**v1.3.1** (current) — [Full changelog](CHANGELOG.md)
+
+| Version | Key Changes |
+|---------|-------------|
+| v1.3.1 | README restructure (KO/EN split), CI shellcheck fix, hook mechanism docs |
+| v1.3.0 | Squad Router hook (80 keywords, 3-phase matching, context injection) |
+| v1.2.1 | Cross-platform notifications (macOS/Linux/WSL), subagent verification |
+| v1.1.3 | SubagentStart/Stop banners, SHA256 checksum, auto hook registration |
+| v1.1.0 | Agent safety rules, pipeline context, tool ordering standardization |
+| v1.0.0 | 8 agents, 9 commands, SubagentStop hook, install script |
+
+<!--
 ### Honest Token Economics
 
 > **"Subagents save tokens" is a common misconception. They actually use MORE.**
@@ -120,6 +134,7 @@ Per Anthropic docs, multi-agent workflows use roughly **4-7x more tokens** than
 | Enforcing tool restrictions (Read-only) | Tasks needing all tools |
 
 > **Bottom line:** Subagents are a **context hygiene tool**, not a token savings tool. They keep your main session clean so quality doesn't degrade. You pay more tokens total, but you get a better workspace.
+-->
 
 ### Why Use Subagents?
 

README.md

@@ -73,6 +73,20 @@ bash install.sh
    - 서브에이전트 컨텍스트는 폐기
 ```
 
+### 버전
+
+**v1.3.1** (현재) — [전체 변경 이력](CHANGELOG.md)
+
+| 버전 | 주요 변경 |
+|------|----------|
+| v1.3.1 | README 구조 개편 (KO/EN 분리), CI shellcheck 수정, 훅 메커니즘 문서화 |
+| v1.3.0 | Squad Router 훅 추가 (80 키워드, 3단계 매칭, context injection) |
+| v1.2.1 | 크로스 플랫폼 알림 (macOS/Linux/WSL), 서브에이전트 검증 완료 |
+| v1.1.3 | SubagentStart/Stop 배너, SHA256 체크섬, 자동 훅 등록 |
+| v1.1.0 | 에이전트 안전 규칙 추가, 파이프라인 컨텍스트, 도구 순서 표준화 |
+| v1.0.0 | 8개 에이전트, 9개 커맨드, SubagentStop 훅, 설치 스크립트 |
+
+<!--
 ### 솔직한 토큰 비용 구조
 
 > **"서브에이전트가 토큰을 절약한다"는 흔한 오해입니다. 실제로는 더 씁니다.**
@@ -120,6 +134,7 @@ Anthropic 문서에 따르면 멀티 에이전트 워크플로우는 단일 에
 | 도구 제한 강제 (Read-only) | 모든 도구 필요 작업 |
 
 > **결론:** 서브에이전트는 **컨텍스트 위생 도구**입니다. 총 토큰은 더 쓰지만, 메인 세션이 깨끗하게 유지되어 세션 후반부 품질 저하를 방지합니다.
+-->
 
 ### 쓰는 이유
 

VERSION

@@ -1 +1 @@
-1.3.0
+1.3.1

install.sh

@@ -14,7 +14,7 @@ RELEASE_BASE="https://github.com/$REPO/releases"
 if [ -n "${BASH_SOURCE[0]:-}" ] && [ -f "$(dirname "${BASH_SOURCE[0]}")/VERSION" ]; then
   VERSION=$(cat "$(dirname "${BASH_SOURCE[0]}")/VERSION")
 else
-  VERSION="1.3.0"
+  VERSION="1.3.1"
 fi
 
 AGENTS_DIR="$HOME/.claude/agents"