Claude Skills by toongri

Use when implementing any feature, adding code, or modifying existing code in this Java/Spring project. Triggers on write operations like adding entities, services, facades, controllers, or any domain logic.

Use when writing tests in this Java/Spring project, generating test skeletons, deciding mock strategies, learning test patterns, or understanding test levels. Triggers include "테스트 작성", "테스트 패턴", "mock 전략", "test skeleton".

Use when facing trade-offs, subjective judgments, uncertain decisions, or when diverse viewpoints would improve judgment quality. Triggers include "council", "다른 의견", "perspectives", "what do others think".

MANDATORY before any implementation when request is vague or has multiple interpretations. Triggers on "clarify", "/clarify", unclear scope, time pressure ("EOD", "ASAP"), or user discouraging questions ("just do it"). If you're about to assume something, STOP and use this skill.

Use when reviewing code changes for quality, correctness, and production readiness before merge

Use when collecting, curating, or organizing job descriptions (JDs) — triggers include "JD 모으고 있어", "JD 수집", "JD 큐레이션", "JD 정리하고 있어", "오늘 수집 정리해줘", "오늘 본 JD", "관리 중인 JD", "쌓아둔 JD", "내 프로필에 맞는 JD 쌓아줘", "내 이력에 맞는 JD 큐레이션", and "싹 돌려" (in JD rescan context). Do NOT trigger on discovery phrases claimed by resume-apply ("JD 찾아줘", "JD 골라줘", "공고 뭐 있지", "지원할 곳", "어디 넣을까") — those belong to resume-apply. Skill maintains project-scoped state at `$OMT_DIR/collect-jd/` (never global).

깔끔하고 전문적인 HTML 기반 발표자료를 단일 파일로 생성한다. 수직 스크롤 + scroll-snap 방식의 스크롤텔링 프레젠테이션으로, 슬라이드 라이브러리 없이 순수 HTML+CSS로 구현한다. 다크/라이트 테마, highlight.js 코드 블럭, 디자인 시스템을 지원하며 frontend-design 스킬 연동으로 화려한 비주얼도 선택 가능하다. 트리거: "make a presentation", "create slides", "build a deck", "발표자료", "프레젠테이션", "슬라이드", "제안서", "발표 만들어", "ppt", "keynote", "pitch deck", "tech talk", "발표 만들어줘".

Socratic deep interview with mathematical ambiguity gating before autonomous execution

Use for advisory design review — steelman antithesis and tradeoff tension analysis. Triggers include "design review", "plan review", "review the plan", "architectural soundness", "설계 검토", "플랜 리뷰", "아키텍처 건전성", "트레이드오프 분석".

Use when asked to analyze architecture, debug issues, identify root cause, or provide technical recommendations. Triggers include "analyze", "diagnose", "debug", "root cause", "what's wrong", "architecture review", "investigate", "아키텍처 분석", "디버깅", "원인 분석", "뭐가 문제", "조사해".

Use when committing changes to git or naming branches. Triggers include "commit", "커밋", "git commit", "finalize changes", "save to git", "commit my work", "branch name", "브랜치 이름", "what should I name this branch".

Setup and manage Oh-My-Toong HUD for Claude Code statusLine

Use when editing text to remove AI writing traces, when text sounds robotic or machine-generated, when reviewing Korean or English content for naturalness. Triggers include "humanize", "AI 흔적 제거", "사람답게", "자연스럽게 고쳐", "de-AI", "remove AI patterns"

Use when creating a PR description. Triggers include "PR 작성", "PR description", "make PR", "PR 만들어", "풀리퀘", "pull request 작성".

Use when generating interviewer-side mock-interview questions from a candidate's resume — triggers include "모의면접 질문", "면접 예상질문", "이력서로 면접 질문", "꼬리질문", "drill-down 질문", "2단계 3단계 깊이", "mock interview questions", "interview questions from resume", "follow-up depth". Use whenever a user provides a resume (PDF/markdown) and asks for interview questions or follow-ups.

Use when reviewing work plans or implementation plans before execution - catches context gaps, ambiguous requirements, missing acceptance criteria

아래 경로의 스킬 문서를 먼저 전부 읽고, 그 문서가 지시하는 절차와 형식을 그대로 따라 대상 변경에 대한 설명 문서를 작성해줘. - 스킬 본문: `/Users/toong/repos/oh-my-toong-playground/curious-hammer/skills/explain-diff/SKILL.md` - 참조 문서: 같은 디렉터리의 `references/` 아래 파일들 이 스킬은 아직 배포되지 않았으므로 Skill 도구로 호출할 수 없다. 위 파일을 직접 읽어서 그 안의 지시를 따르면 된다. 두 가지 실행 조건: 1. 상태 CLI(`explain-diff-state.ts`)는 이 측정에서 호출하지 않는다. 스텝 순서와 각 스텝의 문서 형식 계약만 따르면 된다. 2. 마지막 퀴즈 단계에는 답할 사람이 없다. `render` 스텝까지 마치고 거기서 멈춰라 — 퀴즈 문항을 문서에 넣는 것까지는 하되, 대화로 출제하고 채점하는 단계는 진행하지 않는다.

Browser automation CLI for AI agents. Use when the user needs to interact with websites, including navigating pages, filling forms, clicking buttons, taking screenshots, extracting data, testing web apps, or automating any browser task. Triggers include requests to "open a website", "fill out a form", "click a button", "take a screenshot", "scrape data from a page", "test this web app", "login to a site", "automate browser actions", or any task requiring programmatic web interaction. Also use...

Automates Apple-platform apps (iOS, tvOS, macOS), Android devices, and Amazon Vega OS TV apps in Vega Virtual Devices. Use when navigating apps, taking snapshots/screenshots where supported, driving TV remotes, tapping, typing, scrolling, extracting UI info, collecting evidence, or planning agent-device CLI commands.

Use when creating, refining, or managing requirement-stage PM issues, OR when turning a raw symptom report, bug, regression, or production incident whose root cause is not yet established into an issue. Triggers include "요구사항 티켓 만들어", "티켓 정리", "이 요구사항 이슈로", "티켓 써줘", "이 버그/증상 원인 파악해서 티켓", "장애/이슈 원인 찾아서 티켓", "이 제보 이슈화", "requirement to issue", "manage issue", "file this requirement", "diagnose this bug and file an issue", "incident to issue", "이슈로 만들어", "티켓 작성", "요구사항 이슈화".

Use when a design is already settled (intent, approach, invariants, and boundary decided — typically after deep-interview) and it must be broken into concrete implementation tasks filed as shareable child tickets in the PM tool (Linear sub-issue / Notion / Jira) for the team to track. Triggers include "작업 분해", "task로 쪼개서 티켓 만들어", "설계 끝났으니 sub-issue로 나눠줘", "이 설계 실행 단위로 만들어", "decompose the design into tasks", "break this into tickets", "file the work items". NOT for requirement-stage WHAT slic...

Systematically explore and test a mobile app on iOS/Android with agent-device to find bugs, UX issues, and other problems. Use when asked to dogfood, QA, exploratory test, find issues, bug hunt, or test this app on mobile.

Use when someone must actually understand a code change before acting on it — reviewing an unfamiliar PR, onboarding onto a subsystem through its history, or handing a large agent-authored diff to a human. Triggers include "diff 설명", "PR 설명해줘", "이 변경 이해하고 싶어", "explain this change", "코드 변경 설명 문서", "변경 퀴즈".

MUST USE for frontend/web UI/UX/visual work: building, styling, redesigning pages/components, React setup, performance audits, visual QA, taste, and polish. Routes four rulesets: design taste router and brand references; perfection for Playwright/Chromium Lighthouse/Core Web Vitals; ui-ux-db palettes/fonts/guidelines; designpowers personas/accessibility/critique/handoff; plus curl-only lazyweb real-app-screen research and the beui.dev interaction catalog. Triggers: frontend, UI, UX, design, r...

Raw mechanical interfaces fusing Swiss typographic print with military terminal aesthetics. Rigid grids, extreme type scale contrast, utilitarian color, analog degradation effects. For data-heavy dashboards, portfolios, or editorial sites that need to feel like declassified blueprints.

Overrides default LLM truncation behavior. Enforces complete code generation, bans placeholder patterns, and handles token-limit splits cleanly. Apply to any task requiring exhaustive, unabridged output.

When setting up or working on a React project, install three dev-only tools by default unless the user explicitly opts out. They make every coding agent's frontend work measurably faster and the resulting code measurably better.

Escalation skill for blocked or hard-to-reach web access — load it when a normal browse/fetch is blocked (WAF, 403, Cloudflare, JS-only render, login-gated, or a platform a generic fetcher cannot read). Tiered router: TIER 1 insane-search (headless extraction + WAF bypass via curl_cffi TLS impersonation, yt-dlp, Jina Reader, public APIs, Playwright real-Chrome fallback); TIER 2 agent-reach (platform-native readers for Chinese and social platforms: Xiaohongshu, Douyin, Weibo, Bilibili, V2EX, W...

Use when turning raw meeting notes, a transcript, or an ASR/recording dump into a structured record — separating discussion / decisions / action items, extracting owned action items with verbatim grounding, mapping related docs and tickets, and recommending a next step per action item. Triggers include 회의록 정리, 회의 정리해줘, 이 회의록에서 액션아이템 뽑아줘, 미팅 노트 정리, 회의 녹취록 정리, transcript to action items, meeting notes to action items, summarize this meeting, standup/retro/decision meeting notes.

Use when checking pin graph health. Runs lib/pins/audit to detect dangling relations, duplicates, invalid entities, stale entries, and orphans, then presents a ranked report.

Use when looking up pins by type, tags, or source. Drives lib/pins/query.ts to retrieve matching pin entries from the knowledge graph. Supersedes the legacy manual ls+frontmatter procedure.

Use when you need to record a single pin entity to the knowledge graph. Invokes lib/pins record() to validate and write a canonical .md file. Triggers on "record pin", "pin this", "save this as a pin".

Use when initializing the pins knowledge graph for the first time in a project. Guides the user through creating pins.yaml (the storage manifest). Triggers on "setup pins", "initialize pins", "create pins.yaml", "first-run pins".

Use at the end of a work session to review the WHOLE session and record entities worth pinning. This is the manual, deliberate complete-sweep review — NOT an automated nudge. Triggers on "wrap up", "wrap-up", "session wrap", "end of session", "what should I pin".

Use when asked to implement, build, fix, or create features - especially before writing any code or when scope and requirements are unclear

Use when verifying a code change through a standalone adversarial e2e cycle — drives the changed surface for real (curl/agent-browser/agent-device/bash) and attacks it across 6 coverage axes + 3 per-run checks (failure/boundary/injection/interruption/misleading-success/idempotency), owning diagnosis→fix→re-verify to green via `oracle` (diagnosis) and `sisyphus-junior` (fix) before issuing a binary APPROVE/REQUEST_CHANGES verdict.

Use when receiving code review feedback, before implementing suggestions, especially if feedback seems unclear or technically questionable - requires technical rigor and verification, not performative agreement or blind implementation

MUST USE this skill when ANY of these appear: (1) a JD or job posting is present — look for keywords like 채용, 채용공고, 주요업무, 자격요건, Requirements, Qualifications, 우대사항, 포지션; (2) user mentions applying to a specific company — '지원', '지원하려고', '지원 준비', '이력서 지원', 'resume apply', 'apply'; (3) user wants to tailor resume for a position — '이력서 맞춤', '이력서 준비', 'JD 이력서', 'JD 기반', 'JD 보고'; (4) user provides a JD via text, file path, or URL; (5) user mentions a company name with intent to apply (e.g. '토스 지원', ...

Use when creating, sourcing, or refining resume problem-solving material — building compound scenarios, developing problem definitions, crafting solution strategies, or iterating entries until examiner approval. Triggers on 이력서 재료/소재, 문제 상황 만들기, 시나리오 작성, compound scenario, 이력서 항목 작성

Use when you want the rich human-facing code review experience — a walkthrough with prose explanation plus mermaid architecture/sequence diagrams plus an HTML report of verified findings, not just terminal findings. Triggers include "review report", "코드리뷰 리포트", "리뷰 HTML", "다이어그램 리뷰", "리뷰 다이어그램", "walkthrough 리뷰", "PR 리뷰 리포트".

Use when the user asks to review, evaluate, check, or get feedback on their resume — even partially (self-introduction/자기소개, career/경력, problem-solving/문제해결, or any single section). Triggers on resume review/evaluation/feedback (이력서 리뷰/검토/피드백), section-specific evaluation requests, interview readiness checks, achievement line quality, AI tone audit, or _config.yml + review intent. When a JD is provided, evaluates JD fit and recommends optimal content from the candidate pool. NOT for simple _c...

Use when extracting a page range or chapter from a scanned/OCR'd book PDF (ABBYY FineReader 등) into text, markdown, or study notes (정리) — triggers include "PDF 텍스트 발췌", "스캔본 PDF 추출", "책 챕터 정리", "정리본 만들기", "PDF to markdown", "표가 깨져서 안 나와", "pdftotext", "pymupdf4llm", "marker", "OCR 깨짐", "scanned book extraction", "extract chapter from PDF", "make study notes from PDF". Use whenever a book PDF must become text/notes and tables, code, or precise values need to survive.

Use when orchestrating complex multi-step tasks requiring delegation, parallelization, or systematic completion verification - especially when tempted to do everything yourself or ask user codebase questions

Gemini CLI를 활용한 HTML 디자인 리뷰 스킬. 생성된 HTML 파일을 Gemini에게 전달하여 시각 디자인 개선 지침을 받고, 메인 세션에서 CSS/HTML을 수정한다. Gemini 불가 시 in-session fallback. 트리거: "디자인 리뷰", "gemini review", "design review", "디자인 검토", "디자인 보완", "slides review", "슬라이드 리뷰".

Use when evaluating technical claims in high-depth content section units (예: 문제 해결 / 상세 프로젝트 / 경력 기술서). Defines the 5-axis framework (A1 Technical Credibility, A2 Causal Honesty, A3 Outcome Presence & Clarity, A4 Ownership & Scope, A5 Scanability) plus 2 critical authenticity rules (R-Phys, R-Cross) used by tech-claim-examiner agent. Verb-scope inflation (previously a separate rule) is now caught by A4 integrity_suspected sub-flag (see a4-ownership-scope.md).

Use when reviewing teaser/promotion text for sharing technical blog posts. Triggers include "티저 리뷰", "포스트 공유", "copywriting review", "teaser review", "promotion text", "LinkedIn post review"

Use when reviewing, improving, or writing technical documentation in Korean. Triggers include "문서 리뷰", "테크니컬 라이팅", "기술 문서", "문서 개선", "writing review", "doc review"

Use when implementing any feature or bugfix, before writing implementation code

Autonomous multi-story objective-pursuit executor — runs objective pursuit as a per-story sequential loop. Auto-generates and bulk-approves a Story set, dispatches each confirmed story to sisyphus one at a time gated by a per-story APPROVE verdict, then runs one independent code-review over the accumulated diff at the final story.

Maximum-saturation research orchestration as ONE engine with two postures. Fans synchronous explore+librarian+browsing worker waves across codebase, web, official docs, and OSS repos; runs an EXPAND-until-convergence loop driven by leads workers return at each barrier; verifies contested claims through a separate verification pass (executed code for code-shaped claims, oracle citation re-read for the rest); on the explicit research posture, produces a cited REPORT.md (+ self-contained REPORT....