pjc Harness

Checks code quality, architecture and AGENTS.md conventions after spec compliance passes. Stage 2 review, called at Phase V-6. Read-only.

Explores codebase, finds files, traces symbols without polluting main context. Called by plan-feature during context gathering. Read-only, fast.

Whole-plan verification ONLY after ALL tasks complete (implement-task Phase F-7, plus PRD cross-check for Phase G). NOT for individual tasks. Read-only.

Adversarial review of plan.md before user approval. Called by plan-feature Step 8. Read-only, reports BLOCKER/MAJOR/MINOR.

Verifies a single task diff against plan.md acceptance and scope. Stage 1 review, called at implement-task Phase V-5. Read-only.

This skill should be used when the user requests adding business logic, a domain service, an application service, or any logic that should live in the Domain or Application layer of a DDD project. Triggers on phrases like "서비스 추가", "도메인 로직", "비즈니스 로직", "use case 추가", "add service", "add use case". Generates Domain interface + implementation + DI registration + unit test scaffold.

This skill should be used when the user requests adding a new screen, dialog, page, window, or any UI component that needs a ViewModel in a WinUI 3 / WPF / MAUI project using CommunityToolkit.Mvvm. Triggers on phrases like "ViewModel 추가", "새 화면", "다이얼로그 추가", "페이지 만들기", "add screen/page/dialog/window". Generates ViewModel + View skeleton with proper MVVM bindings and DI registration.

This skill should be used when starting work on a project that has no AGENTS.md file. Triggered automatically by plan-feature when AGENTS.md is missing, or manually with "/pjc:bootstrap-agents-md". Detects project stack from marker files (.csproj, package.json, pyproject.toml, etc.) and generates a minimal AGENTS.md from one of 7 templates. If stack is unknown, asks the user.

This skill should be used ONLY when the user explicitly wants to enable, disable, toggle, or check status of pjc harness hooks. Triggers REQUIRE explicit context words "harness" or "hook" combined with on/off/status intent. Examples that SHOULD trigger - "harness 끄기", "harness 켜기", "hook 끄기", "hook 상태", "harness status", "harness 토글", "plan 강제 꺼" (with "강제" indicating hook), "require-plan off", "require-plan on", "evidence 검사 꺼", "utf8 검사 꺼". Examples that should NOT trigger - "이 기능 끄는 코드 만들어줘", "기능 켜기", "버튼 끄기" (those are code changes via plan-feature, not harness toggle). When in doubt, do NOT trigger this skill.

This skill should be used when executing tasks from an approved plan.md. Triggers on phrases like "구현", "implement", "T<N> 진행", "이대로 진행", "go", "진행해". Runs FULLY AUTONOMOUS loop — processes ALL tasks (T1...Tn) without asking the user between tasks. Resuming mid-plan ("T6부터 계속") means T6 through the LAST task plus Phase F/G — not just T6. Only stops when all tasks complete or a Halt Condition fires. Never asks "Should I proceed to the next task?" between tasks. For trivial single-line edits without a plan, do NOT use this skill — Claude applies the change directly and lets hooks validate.

6 hooks across 4 events