메뉴
Use when code exists but needs validation for stubs, wiring, and anti-patterns to confirm phase completion.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Use when Ground Truth alignment finds memory contamination and the user explicitly requests scoped cleanup with /cleanse.
Use when registering, reviewing, merging, or rebuilding HXSK glossary term definitions after glossary-detect suggests a candidate or the user invokes /define.
Use when staged changes exist requiring qlty checks, logical split detection, or HXSK phase-scoped conventional commit creation.
Use when validating code for circular imports, layer violations, or design doc logic errors before merging architecture-level changes.
Use when .hxsk/.bootstrap-version is missing (fresh install) or exists (verify/update) to initialize HExoskeleton.
Use when shell scripts need linting/formatting before execution or commit to ensure code quality.
| description | Use when code exists but needs validation for stubs, wiring, and anti-patterns to confirm phase completion. |
| name | verifier |
| trigger | 구현 검증, 완료 확인, 페이즈 검증, verify implementation, check phase completion, validate against spec, stub 탐지, stub check, anti-pattern 스캔, verify wiring, 3-level 검증, 재검증, re-verification, gap 식별, identify gaps, empirical validation, verify code substance, check TODO/FIXME, verification report 생성, verify artifacts, verify key links, 재검증 모드, re-verification mode, 인간 검증 필요, human verification needed, human needed, 가짜 구현 탐지, fake implementation check, placeholder 탐지, placeholder check, 연결성 검증, connection verification, wiring check, 증거 기반 검증, evidence based verification, 반패턴 스캔, anti-pattern scan, blocker 확인, check blockers, 검증 상태 결정, determine verification status, gaps 식별, identify gaps, 검증 템플릿 생성, generate verification template, stub-free 확인, stub-free check, TODO 스캔, TODO scan, FIXME 스캔, FIXME scan, 빈 파일 탐지, empty file check, 최소 구현 탐지, minimal implementation check, 외부 서비스 검증, external service verification, UI 검증, UI verification, 실시간 검증, real-time verification, WebSocket 검증, WebSocket check, SSE 검증, SSE check, 성능 검증, performance verification, 요구사항 커버리지, requirements coverage, 만족도 확인, satisfaction check, 검증 점수 계산, calculate verification score, VERIFICATION.md 생성, generate VERIFICATION.md, 검증 결과 저장, save verification result, 메모리 프로토콜, memory protocol, 영향 분석, impact analysis, 게이트 함수 검증, gate function verification, 신뢰성 검증, trust verification, 모든 것 검증, verify everything |
TODO/placeholder 등 스텁 코드가 존재하면 즉시 FAILED 판정grep 등 실증적 증거 없이는 VERIFIED 불가Trust nothing. Verify everything.
ls .hxsk/phases/{N}/*-VERIFICATION.md 2>/dev/null
RE-VERIFICATION MODE (previous exists with gaps): Extract must-haves + gaps → set is_re_verification = true → Skip to Step 3 (failed items: full check; passed items: quick regression).
INITIAL MODE (no previous): set is_re_verification = false, proceed Step 1.
ls .hxsk/phases/{N}/*-PLAN.md && ls .hxsk/phases/{N}/*-SUMMARY.md
grep "Phase {N}" .hxsk/ROADMAP.md
Option A: Read from PLAN frontmatter (must_haves.truths, .artifacts, .key_links).
Option B: Derive from phase goal — truths (observable behaviors), artifacts (concrete files), key_links (critical wiring: component → API → DB).
For each truth: identify artifacts → check levels (Step 4) → check wiring (Step 5) → assign ✓ VERIFIED / ✗ FAILED / ? UNCERTAIN.
test -f "{path}"grep -E "TODO|placeholder|stub" "{path}" — no stubsStub Detection Patterns → references/stub-detection.md
Check each link exists: Component→API (grep "fetch.*api/chat"), API→DB (grep "prisma\."), Form→Handler (grep -A5 "onSubmit"), State→Render (grep "messages\.map").
grep "Phase {N}" .hxsk/REQUIREMENTS.md
Status per requirement: ✓ SATISFIED / ✗ BLOCKED / ? NEEDS HUMAN.
grep -r -E "TODO|FIXME|XXX|HACK" src/**/*.ts
grep -r -E "placeholder|coming soon" src/**/*.tsx
grep -r -E "return null|return \{\}|return \[\]" src/**/*.ts
Categorize: 🛑 Blocker / ⚠️ Warning / ℹ️ Info.
Always human: visual appearance, user flow, real-time behavior (WebSocket/SSE), external services, performance feel.
passed = all truths verified, no blockers. gaps_found = any truth failed/stub/unwired/blocker. human_needed = automated pass but human items exist. Score = verified_truths / total_truths.
Structure gaps in YAML for /plan --gaps (truth, status, reason, artifacts, missing items).
Sections to include: frontmatter (phase/verified/status/score/is_re_verification/gaps), Must-Haves (Truths table, Artifacts table, Key Links table), Anti-Patterns Found, Human Verification Needed, Gaps, Verdict.
VERIFICATION.md 전체 템플릿 → references/verification-templates.md
empirical-validation — Gate Function 5단계로 완료 검증memory-protocol — 검증 결과를 메모리에 저장impact-analysis — 검증 범위 결정 시 영향 분석 참조# Stub/placeholder 패턴 탐지
Grep(pattern: "TODO|FIXME|NotImplementedError|pass$|return null|return \\{\\}", path: "src/", output_mode: "content")
# 파일 존재 확인
Glob(pattern: "src/**/*.{ts,js,py}")
# 파일 substance 확인 (빈 파일/최소 구현 탐지)
Read(file_path: "{file}") → 라인 수와 내용 확인
NO VERIFY WITHOUT EVIDENCE FIRST NO COMPLETION WITHOUT VERIFICATION FIRST NO EXISTENCE WITHOUT SUBSTANCE FIRST NO WIRING WITHOUT CONNECTION FIRST NO FINAL APPROVAL WITHOUT HUMAN VERIFICATION FIRST NO PASS WITHOUT CLEAN CODE FIRST