oozoofrog

Guidance on designing hierarchical context architecture for large-scale projects — structuring CLAUDE.md, CONTEXT.md, AGENTS.md to maximize agent reasoning precision and token efficiency. Use when the user asks about "컨텍스트 아키텍처", "context architecture", "계층적 컨텍스트", "CONTEXT.md 설계", "주의력 예산", "attention budget", "컨텍스트 엔지니어링", "context engineering", "토큰 효율성", "컨텍스트 부패", "context rot", "점진적 노출", "progressive disclosure". Common requests include "CLAUDE.md가 너무 길어요", "프로젝트 컨텍스트 파일을 어떻게 구성하죠?", "My CLAUDE.md is too long", "Set up context architecture for my project".

Analyze the project and scaffold a layered context architecture made of CLAUDE.md, subdirectory CLAUDE.md, .claude/rules/, and AGENTS.md. Preserve and augment existing files. Triggers — 컨텍스트 초기화, 컨텍스트 아키텍처, CLAUDE.md 생성, 스캐폴딩.

Runs 3-stage verification of a hierarchical context architecture — reference integrity, code reference validity, and content accuracy checked in sequence to produce a report. Triggers — 계층적 컨텍스트 검증, 참조 무결성, 코드 참조 검증, 내용 정확성, 컨텍스트 아키텍처 검증, ctx-verify.

Design research orchestrator — study famous designers/artists and codify their design traits into quantitative tokens. Use for '디자인 리서치', '디자이너 연구', '디자인 토큰 생성', '화가 분석', '디자인 시스템 구축' requests. Auto-triggers before the design-craft skill when no reference exists.

Browse Claude Code session logs in a TUI. Use on requests like "세션 로그 보기", "session viewer", "세션 뷰어", "claude code 로그 TUI", "이전 대화 보기", "내 세션 목록", "지난 세션 검토", "로그 인터랙티브 보기". Toggles between all-project sessions and sessions matching the current working directory.

Extracts project context into a structured research prompt for GPT-PRO and copies it to the clipboard. Use for requests that delegate research to GPT or extract project context as a structured prompt, such as "GPT에 물어봐", "GPT 리서치", "GPT-PRO", "리서치 위임", "컨텍스트 추출", "gpt research", "GPT에게 넘겨", "GPT한테 질문", "GPT 프롬프트", "research prompt", "컨텍스트 뽑아줘", "GPT용 프롬프트", "외부 리서치", "프롬프트 생성", "GPT 위임".

Automates the macOS app release pipeline — Xcode version bump, Release build, DMG packaging, local install, GitHub Release, Homebrew Cask publishing, and GitHub Actions CI/CD. Use for macOS app distribution requests. 트리거: "릴리스", "release", "버전 업데이트", "version bump", "배포", "deploy", "홈브루 업데이트", "brew cask", "DMG 만들기", "새 버전 배포", "publish", "CI/CD", "workflow", "자동 배포", "GitHub Actions", "새 버전 배포해주세요", "릴리스 스크립트 만들어주세요", "홈브루에 올려주세요", "배포 자동화 워크플로우 만들어주세요".

Learning-science-based code review and educational content generation. /yoda review for Before/After/Why reviews, /yoda share to spread to a team in 4 formats (md/web/wiki/slides). 코드 리뷰, 코드 분석, 코드 설명, 팀 공유, 온보딩, 기술 세미나, 지식 전파 요청 시 활성화

Karpathy의 LLM Wiki 아이디어를 바탕으로 raw source와 LLM-maintained wiki를 분리한 지속형 지식 베이스를 설계·부트스트랩합니다. markdown/git 기반의 범용 구조, `AGENTS.md` 운영 규칙, ingest/query/lint 워크플로우, Obsidian 같은 선택적 어댑터 분리가 필요할 때 사용합니다.

Use this skill when the user explicitly wants a goal-directed research loop that keeps proposing, testing, logging, and refining ideas toward a stated objective across repeated iterations or autonomous runs. Do not use it for one-shot summaries or when no evaluation signal can be defined.

Codex instruction 구조의 밀도와 건강도를 감사합니다. AGENTS.md 길이, 계층 깊이, 중복 규칙, 커버리지 부족을 찾을 때 사용합니다.

Use this skill when the user wants to audit a Codex skill or skill repository for official structure compliance, discovery readiness, frontmatter quality, oversized SKILL bodies, stale `agents/openai.yaml`, duplicate skill names, or missing repo-level AGENTS guidance.

Codex 플러그인/스킬 저장소를 감사하고 개선합니다. `.codex-plugin/plugin.json`, `.agents/plugins/marketplace.json`, `.agents/skills`, `.codex/agents`, legacy Claude 산출물을 점검할 때 사용합니다.

Apple 플랫폼 관점의 코드·PR 리뷰를 수행합니다. Swift/SwiftUI/UIKit/AppKit 코드의 정확성, API 사용, 빌드 위험, 테스트 누락을 점검할 때 사용합니다.

Codex 플러그인/스킬 저장소를 감사하고 개선합니다. `.codex-plugin/plugin.json`, `.agents/plugins/marketplace.json`, `.agents/skills`, `.codex/agents`, legacy Claude 산출물을 점검할 때 사용합니다.

AGENTS.md 기반 instruction 구조의 정합성을 검증합니다. 링크 무결성, 코드/명령 참조, 문서 주장 정확성을 점검할 때 사용합니다.

Bridge Codex with the local Claude Code CLI for real analysis, implementation, and review handoffs. Use when Codex should call `claude` directly from the shell to analyze a codebase, draft or apply targeted code changes, produce a second-opinion review, or compare Codex and Claude outputs in trusted local repositories.

Configure and troubleshoot Ghostty terminal settings with a safe, repeatable workflow for themes, profiles, accessibility/readability, and include chains. Use when Codex needs to create, update, or debug Ghostty config files (for example `config`, `themes/*.conf`, `profiles/*.conf`), adjust contrast/font/cursor/padding behavior, or switch and verify theme/profile setups.

End-to-end App Store and TestFlight release preparation for iOS and watchOS apps. Use when Codex needs to refresh or automate App Store screenshots, add or update fastlane screenshot lanes, update version/build numbers, write What's New or release notes, sync metadata source files, validate release readiness, and build or upload binaries with fastlane, Xcode, or the repo's existing App Store delivery workflow.

Coordinate multi-agent iOS/Swift development by splitting work across task framing, codebase analysis, implementation, critical review, and review incorporation. Use when Codex should orchestrate separate planner/reviewer and builder agents for SwiftUI/UIKit, networking, concurrency, feature work, bug fixes, refactors, or release-sensitive changes instead of letting one agent both analyze and implement.

Route iOS/Swift tasks between multi-agent delivery workflow and deep Swift domain expertise. Use when Codex needs a single entry point to decide whether to apply `ios-multi-agent-dev`, `swift-master`, or both for iOS bug fixes, features, refactors, SwiftUI/SwiftData/concurrency issues, Swift 6 migrations, release-sensitive changes, or code review.

Review, optimize, migrate, guide, and generate Swift/iOS code across Swift 6, SwiftUI, Swift Concurrency, SwiftData, MVVM/TCA architecture, Pure DI, and Combine-to-AsyncSequence conversion. Use when Codex needs domain-specific Swift expertise for `.swift` files, iOS app codebases, SwiftUI state management, concurrency correctness, Swift 6 migration, architecture choices, dependency injection, or repo-wide code review.

Apply this skill whenever async work, task orchestration, cancellation, or actor correctness is involved.

Apply this skill whenever the task touches app structure, screen composition, or responsibility boundaries.

Interactive setup wizard for GitHub sync

Pull latest configuration from GitHub

Push configuration changes to GitHub

Show current sync configuration

Open GitHub sync repository page in browser

Show sync status and pending changes

Guide for syncing with upstream Swift compiler demangling changes

Validate mangled symbols against expected demangled output

Format and create a Stage X.Y commit for swiftmcp using the established message format (HEREDOC subject + bullet body + test-count line) and push to origin/main. Wraps git add + commit + push in one user-driven step. User-only invocation — commits land on a public GitHub repository.

Capture verbatim swiftc output for a given option set against the locally installed toolchain. Use BEFORE introducing a new tool whose parser/regex/sanity-check depends on swiftc's exact output shape — prevents format guesses that may drift from the real toolchain. Examples — probe `-warn-long-expression-type-checking` warning format, capture `-emit-sil` heading, verify `frontend -interpret` runtime behavior on host SDK.

{{SWIFTNEST_AGENT_SKILL_DESCRIPTION}}