@walwal-harness/cli 7.1.0 → 7.1.2

package/HR-Resource/ceo/SKILL.md

@@ -34,3 +34,14 @@ You are the only direct conversation channel with the Owner.
 - Before CTO/CDO/OPS allocate runnable services, agree with the Owner on a `{xx}000` base port and write it to project `.env` as `HARNESS_BASE_PORT={xx}000`. Mentioning the value in `ceo.md` is not sufficient.
 - After writing `.env`, verify with `grep '^HARNESS_BASE_PORT=' .env` before routing service work.
 - For service monitoring, collect the Owner's server mapping first: local PC, Docker, VM, AWS/cloud, host, port, health path, log path, and contact/source.
+
+## Worktree Isolation Failure — No git Repo
+
+When an Agent spawn fails with a worktree or git error (e.g., "Cannot create agent worktree: not in a git repository"), this is an **isolation constraint**, not a hiring failure. The correct response is:
+
+1. **Do not skip `harness-hiring`.** Run `harness-hiring` as normal to register the worker in `hr-roster.json`.
+2. **Do not replace hired workers with inline "You are a…" prompts.** That is impersonation, not hiring.
+3. **Spawn the hired worker as a plain `Agent()` call** (no `subagent_type`, no worktree) and set the prompt to read the worker's SKILL.md from `.harness/shared/HR-Resource/{worker-name}/SKILL.md` before executing the task.
+4. **Report the isolation constraint to the Owner** in the final summary: state that worktree isolation was unavailable and workers ran without isolation.
+
+The worktree error only affects **isolation**. `harness-hiring`, `harness-resource-manager`, and `hr-roster.json` registration are independent of git and must always run.

package/assets/templates/config.json

@@ -1,200 +1,28 @@
 {
   "harness": {
     "name": "walwal-harness — NEXUS company harness",
-    "version": "6.1.4",
-    "description": "NEXUS-adapted company metaphor for AI-agent engineering. Owner ↔ Dispatcher(CEO) ↔ Conductor + Meeting-Manager ↔ Planner(COO/HR) + CTO + CQO + Service-Ops. Generators (BE/FE/Designer/DevOps) + Evaluators (Functional/Visual/Code-Quality/Architecture/Security) + COO Hypothesis Cell. Always-on company mode, parallel worker pool, parallel-tracks fork-join meetings, hierarchical Conventions/Gotchas. React/Next.js/Flutter FE, NestJS/Laravel BE.",
+    "version": "7.1.1",
+    "description": "v7 company harness. Owner uses /goal or /hot-fix, CEO coordinates CXX agents, CXX agents use resource-manager/hiring, and hired workers produce mission evidence under .harness/documents/{mission}/workers/.",
     "source": "https://www.anthropic.com/engineering/harness-design-long-running-apps"
   },
   "agents": {
-    "dispatcher": {
-      "role": "사용자 요청 분석 → 파이프라인 결정 (FE-ONLY / BE-ONLY / FULLSTACK)",
-      "skill": "harness-dispatcher",
-      "outputs": ["actions/pipeline.json"],
-      "always_first": true,
-      "model": "opus",
-      "thinking_mode": null
-    },
-    "brainstorming": {
-      "role": "사용자의 러프한 요구사항을 대화형으로 구체화하여 Planner가 바로 쓸 수 있는 brainstorm-spec.md로 변환.",
-      "skill": "harness-brainstorming",
-      "inputs": ["(user conversation)"],
-      "outputs": ["actions/brainstorm-spec.md"],
-      "conditional": true,
-      "trigger_conditions": {
-        "comment": "Dispatcher가 아래 조건 중 하나 이상 충족 시 사용자에게 브레인스토밍 제안",
-        "rules": [
-          "사용자 요청에 구체적 기능명/엔드포인트/화면명이 없고 추상적 목표만 있을 때",
-          "PRD/OpenAPI 등 구조화된 입력 문서가 없을 때",
-          "'뭔가 만들고 싶은데', '아이디어가 있는데' 등 탐색적 표현이 포함될 때"
-        ],
-        "skip_when": [
-          "사용자가 PRD, OpenAPI spec, 또는 구체적 feature list를 제공한 경우",
-          "사용자가 '브레인스토밍 없이', 'skip brainstorming'을 명시한 경우",
-          "이미 brainstorm-spec.md가 actions/에 존재하는 경우"
-        ]
-      },
-      "invoked_by": "dispatcher (user opt-in after suggestion)",
-      "next_on_complete": "planner",
-      "model": "opus",
-      "thinking_mode": "plan",
-      "thinking_mode_description": "Claude Code plan 모드로 실행. 구조화된 계획 수립 후 실행.",
-      "attribution": "Derived from obra/superpowers skills/brainstorming (MIT License)"
-    },
-    "planner": {
-      "role": "제품 사양 + API 계약서 + 서비스 분할 설계",
-      "skill": "harness-planner",
-      "inputs": ["actions/brainstorm-spec.md (optional)"],
-      "outputs": [
-        "actions/plan.md",
-        "actions/feature-list.json",
-        "actions/api-contract.json"
-      ],
-      "model": "opus",
-      "thinking_mode": null,
-      "thinking_mode_description": null
-    },
-    "generator-backend": {
-      "role": "NestJS MSA 서비스 구현, API 엔드포인트, DB 스키마",
-      "skill": "harness-generator-backend",
-      "inputs": ["actions/plan.md", "actions/feature-list.json", "actions/api-contract.json", "actions/sprint-contract.md"],
-      "outputs": ["code:backend", "actions/sprint-contract.md"],
-      "order": 1,
-      "model": "sonnet",
-      "thinking_mode": null,
-      "model_rationale": "코드 생성은 Sonnet이 비용 대비 효율적. 계획은 Planner가 이미 완료."
-    },
-    "generator-frontend": {
-      "role": "React/Next.js UI 컴포넌트, 상태관리, API 연동",
-      "skill": "harness-generator-frontend",
-      "inputs": ["actions/plan.md", "actions/feature-list.json", "actions/api-contract.json", "actions/sprint-contract.md"],
-      "outputs": ["code:frontend", "actions/sprint-contract.md"],
-      "order": 2,
-      "model": "sonnet",
-      "thinking_mode": null,
-      "model_rationale": "코드 생성은 Sonnet이 비용 대비 효율적. 계획은 Planner가 이미 완료."
-    },
-    "evaluator-code-quality": {
-      "role": "코드 자체의 유지보수성/아키텍처/Best Practice 검사 (BE/FE/libs 공통, 브라우저 없음)",
-      "skill": "harness-evaluator-code-quality",
-      "tools": ["Read", "Grep", "Glob", "Bash(tsc,eslint,biome,lint 스크립트 한정)"],
-      "inputs": ["actions/sprint-contract.md", "actions/feature-list.json", "actions/api-contract.json"],
-      "outputs": ["actions/evaluation-code-quality.md"],
-      "model": "opus",
-      "thinking_mode": "ultrathink",
-      "thinking_mode_description": "시니어 리뷰어 관점. 변경 파일 전수 정적 분석 + C1-C5 축 적대적 채점.",
-      "adversarial_rules": {
-        "comment": "Code-Quality Evaluator 적대적 행동 규칙 — rubber-stamping 금지",
-        "rules": [
-          "브라우저 없이 코드만으로 판정. '동작하니 PASS' 금지.",
-          "변경 파일(git diff)은 전수 검사. '시간 제약' 핑계 금지.",
-          "모든 Score 주장은 파일:라인 evidence 동반. Evidence 없는 Score = 0.",
-          "toolchain(tsc/eslint) 에러 1건 이상이면 나머지 축 채점 전 즉시 FAIL.",
-          "api-contract.json ↔ 실제 DTO/시그니처 불일치 1건 이상 = FAIL.",
-          "레이어 역방향 의존 / MSA 직접 DB 접근 / FE view→fetch 직결 1건 이상 = C1 Score 0.",
-          "unhandled promise 1건 이상 또는 비타당한 any 3건 이상 = C4 Score 0.",
-          "PASS를 주기 전에 자문하라: '내가 이 코드를 머지 승인할 수 있는가?' NO이면 FAIL.",
-          "반복 페널티: 같은 축에서 이전 스프린트 대비 악화 시 Score 최대 1."
-        ],
-        "forbidden": [
-          "'동작은 Functional 에서 확인됐으니 OK' 식의 책임 전가",
-          "'관용적 any', '사소한 중복' 같은 자기설득",
-          "'제안사항만 남기고 Pass' — 필수 개선은 FAIL",
-          "파일 수정/커밋 (이 평가자는 읽기 전용)"
-        ]
-      }
-    },
-    "evaluator-functional": {
-      "role": "Playwright로 E2E 기능 검증, API 응답/DB 상태 확인",
-      "skill": "harness-evaluator-functional",
-      "tools": ["playwright:browser_*"],
-      "inputs": ["actions/sprint-contract.md", "actions/feature-list.json", "actions/api-contract.json"],
-      "outputs": ["actions/evaluation-functional.md"],
-      "model": "opus",
-      "thinking_mode": "ultrathink",
-      "thinking_mode_description": "최고 수준의 사고 모드. 적대적 검증에 깊은 추론이 필요. 각 항목을 비판적으로 분석.",
-      "evaluation_template": "assets/templates/evaluation-functional.md.template",
-      "adversarial_rules": {
-        "comment": "Evaluator 적대적 행동 규칙 — 통과 편향 제거를 위한 강제 규칙",
-        "rules": [
-          "Generator의 '구현 완료' 주장을 절대 신뢰하지 마라. 모든 엔드포인트에 직접 요청을 보내고 응답을 확인하라.",
-          "정상 경로(happy path) 1개당 비정상 경로(unhappy path) 최소 2개를 반드시 테스트하라.",
-          "api-contract.json과 실제 응답을 필드 단위로 비교하라. 누락 필드, 타입 불일치, 추가 필드 모두 FAIL 사유.",
-          "PASS를 주기 전에 자문하라: '내가 이 코드를 직접 작성했다면 이 수준으로 PR을 올리겠는가?' NO이면 FAIL.",
-          "AC 항목 하나라도 미통과이면 전체 FAIL. 부분 통과는 인정하지 않는다.",
-          "Evidence(증거)가 없는 Score는 0점이다. '확인했다'는 주장은 증거가 아니다.",
-          "이전 Sprint에서 PASS된 기능이 하나라도 깨졌으면 신규 기능 점수와 무관하게 FAIL.",
-          "Score 3을 주려면 엣지케이스(빈 값, 초과값, 동시 요청, 특수문자)까지 테스트한 증거가 있어야 한다."
-        ],
-        "forbidden": [
-          "'전반적으로 잘 구현되었습니다' 류의 모호한 긍정 평가",
-          "'사소한 이슈이므로 PASS' — 사소함은 Evaluator가 판단할 사항이 아니다",
-          "'시간 제약상 일부만 테스트' — 전수 테스트가 불가하면 FAIL 처리",
-          "Generator에게 유리한 방향으로 기준을 해석하는 행위"
-        ]
-      }
-    },
-    "evaluator-visual": {
-      "role": "디자인 일관성, 반응형, 접근성, AI슬롭 감지",
-      "skill": "harness-evaluator-visual",
-      "tools": ["playwright:browser_take_screenshot", "playwright:browser_resize", "playwright:browser_snapshot"],
-      "inputs": ["actions/sprint-contract.md", "actions/evaluation-functional.md", "actions/feature-list.json"],
-      "outputs": ["actions/evaluation-visual.md"],
-      "model": "opus",
-      "thinking_mode": "ultrathink",
-      "thinking_mode_description": "최고 수준의 사고 모드. 시각적 결함, 접근성, AI슬롭을 비판적으로 탐지.",
-      "evaluation_template": "assets/templates/evaluation-visual.md.template",
-      "adversarial_rules": {
-        "comment": "Visual Evaluator 적대적 행동 규칙",
-        "rules": [
-          "모든 페이지를 반드시 3개 뷰포트(375px, 768px, 1280px)에서 스크린샷 촬영하라. 1개라도 누락이면 해당 페이지 Score 산정 불가.",
-          "axe-core 접근성 검사를 실제로 실행하라. Critical 또는 Serious 위반이 1건이라도 있으면 V3 = 0.",
-          "AI슬롭 패턴(무의미한 그라데이션, 과도한 그림자, 장식 아이콘 남발)을 적극 탐지하라. 2건 이상이면 V4 최대 1점.",
-          "evaluation-functional.md의 Cross-Validation Data와 교차 검증하라. API 성공인데 UI에 에러 표시 = 불일치 = FAIL 사유.",
-          "이전 Sprint 스크린샷 대비 의도치 않은 시각 변경이 있으면 점수 무관 FAIL.",
-          "PASS를 주기 전에 자문하라: '이 UI를 실사용자에게 보여줄 수 있는가?' NO이면 FAIL.",
-          "로딩/에러/빈 상태가 하나라도 미구현이면 V5 최대 1점.",
-          "Evidence(스크린샷 경로 또는 DOM 스냅샷)가 없는 Score는 0점이다."
-        ],
-        "forbidden": [
-          "'디자인이 깔끔합니다' 류의 주관적 긍정 평가",
-          "'접근성은 추후 개선' — 접근성은 기본 요건이지 부가 기능이 아니다",
-          "'모바일 뷰포트는 생략' — 3개 뷰포트 전수 검사는 필수",
-          "스크린샷 없이 '확인했다'고 주장하는 행위"
-        ]
-      }
-    }
+    "ceo": { "role": "Owner-facing mission intake, CXX orchestration, and final reports", "skill": "harness-ceo", "model": "sonnet" },
+    "coo": { "role": "Planning, research, hypothesis, mission fit", "skill": "harness-coo", "model": "sonnet" },
+    "cdo": { "role": "Branding, UI/UX, design review", "skill": "harness-cdo", "model": "sonnet" },
+    "cto": { "role": "Architecture, platform, API, implementation coordination", "skill": "harness-cto", "model": "opus" },
+    "cqo": { "role": "Quality gates, regression, archive criteria", "skill": "harness-cqo", "model": "opus" },
+    "ops": { "role": "Build/service environment monitoring and exception logging", "skill": "harness-ops", "model": "sonnet" },
+    "hiring": { "role": "Hire missing workers from HR-Resource", "skill": "harness-hiring", "model": "sonnet" },
+    "resource-manager": { "role": "Resolve hired-worker availability and invocation wording", "skill": "harness-resource-manager", "model": "haiku" },
+    "brick-office": { "role": "Company status and dashboard support", "skill": "harness-brick-office", "model": "haiku" }
   },
   "flow": {
-    "sequence": ["dispatcher", "planner", "generator-backend", "generator-frontend", "evaluator-code-quality", "evaluator-functional", "evaluator-visual"],
-    "evaluator_chain": {
-      "comment": "Generator 완료 후 실행되는 평가자 체인. 직렬 실행 + 조기 종료 — 앞단 FAIL 시 이후 평가자는 생략하고 Generator 로 리라우팅. 모두 PASS 여야 다음 스프린트로 이동.",
-      "execution": "serial_with_early_exit",
-      "order_rationale": "code-quality(정적, 저비용) → functional(동작, 중비용) → visual(렌더, 고비용). 구조가 깨진 코드에 동작 테스트를 낭비하지 않기 위함.",
-      "on_any_fail": "reroute_to_generator",
-      "on_all_pass": "advance_sprint"
-    },
-    "pipeline_selection": {
-      "comment": "Dispatcher가 pipeline.json으로 활성 에이전트를 결정.",
-      "pipelines": {
-        "FULLSTACK": ["planner", "generator-backend", "generator-frontend", "evaluator-code-quality", "evaluator-functional", "evaluator-visual"],
-        "FE-ONLY": ["planner:light", "generator-frontend", "evaluator-code-quality", "evaluator-functional", "evaluator-visual"],
-        "BE-ONLY": ["planner", "generator-backend", "evaluator-code-quality", "evaluator-functional:api-only"]
-      },
-      "evaluator_chains": {
-        "comment": "파이프라인별 평가자 체인. Dispatcher가 pipeline.json.evaluator_chain 으로 기록.",
-        "FULLSTACK": ["evaluator-code-quality", "evaluator-functional", "evaluator-visual"],
-        "FE-ONLY": ["evaluator-code-quality", "evaluator-functional", "evaluator-visual"],
-        "BE-ONLY": ["evaluator-code-quality", "evaluator-functional"]
-      }
-    },
-    "sprint_execution": {
-      "comment": "Backend 먼저 → API 안정 후 Frontend 연동 → 기능 QA → 비주얼 QA (FULLSTACK 기준)",
-      "backend_first": true,
-      "reason": "프론트가 존재하지 않는 API를 호출하는 실패 방지"
-    },
+    "sequence": ["ceo", "coo", "cdo", "cto", "cqo", "ops"],
+    "worker_gate": "CXX agents must use harness-resource-manager and harness-hiring before specialist deliverables.",
+    "worker_reports_required": ".harness/documents/{mission}/workers/{worker-name}.md",
     "retry_on_fail": true,
     "max_retries_per_sprint": 5,
-    "escalate_to_planner_after": 3,
+    "escalate_to_ceo_after": 3,
     "archive_on_sprint_complete": true,
     "pre_eval_gate": {
       "comment": "Generator → Evaluator 전환 전 결정론적 검증. 실패 시 Evaluator 세션을 열지 않고 Generator로 자동 리라우팅.",
@@ -338,9 +166,9 @@
     }
   },
   "behavior": {
-    "comment": "하네스 동작 플래그. UserPromptSubmit 훅이 이 값을 읽어 자동 라우팅을 결정한다.",
-    "auto_route_dispatcher": true,
-    "auto_route_dispatcher_description": "true 이면 사용자의 모든 프롬프트가 harness-dispatcher 경유로 분류/라우팅 된다. 사용자가 'harness skip' 등을 말하면 단일 메시지 한정으로 건너뛴다."
+    "comment": "하네스 동작 플래그. UserPromptSubmit 훅이 이 값을 읽어 v7 CEO/CXX 라우팅 안내를 결정한다.",
+    "auto_route_ceo": true,
+    "auto_route_ceo_description": "true 이면 /goal 또는 /hot-fix 이후 Owner 입력을 v7 CEO/CXX mission flow 기준으로 안내한다. 사용자가 'harness skip' 등을 말하면 단일 메시지 한정으로 건너뛴다."
   },
   "token_limit": {
     "comment": "모델 토큰 제한으로 작업이 중단됐을 때의 저비용 재개 정책. 별도 probe 호출 없이 시간 기반으로만 재개 알림을 계산한다.",
@@ -350,9 +178,9 @@
     "owner": "session-start"
   },
   "company_mode": {
-    "comment": "v6.3 — 회사모드는 항시 활성이다. 사용자 모드 선택과 tie-breaker 는 제거됐다. Conductor 는 feature-list/queue 를 기반으로 매 tick 가능한 병렬 작업을 자동 배정한다.",
+    "comment": "v7 — 회사모드는 항시 활성이다. Owner entrypoint는 /goal, /hot-fix이며 CEO가 CXX와 hired worker 흐름을 조정한다.",
     "default_mode": "company",
-    "owner": "conductor",
+    "owner": "ceo",
     "worker_concurrency": 3,
     "worker_spawn": "claude",
     "worker_spawn_description": "claude 이면 idle worker 배정 시 `claude -p` 백그라운드 프로세스를 띄운다. record 로 두면 prompt/log만 생성한다.",

package/assets/templates/progress.json.template

@@ -2,10 +2,10 @@
   "version": 4,
   "mode": "company",
   "mode_decision": {
-    "owner": "conductor",
+    "owner": "ceo",
     "decided_at": null,
-    "rationale": "company mode is always-on",
-    "policy": "always_company_parallel"
+    "rationale": "v7 company mode is CEO-led",
+    "policy": "owner_goal_to_ceo_cxx_workers"
   },
   "project_name": "",
   "pipeline": null,
@@ -21,7 +21,7 @@
   "current_agent": null,
   "agent_status": "pending",
   "completed_agents": [],
-  "next_agent": "dispatcher",
+  "next_agent": "ceo",
   "failure": {
     "agent": null,
     "location": null,

package/bin/init.js

@@ -165,6 +165,26 @@ const V7_REMOVED_GOTCHA_FILES = [
   'generator-backend-laravel.md',
 ];
 
+const LEGACY_V7_FILE_OWNER = {
+  'conductor.md': 'ceo.md',
+  'dispatcher.md': 'ceo.md',
+  'coo-developer.md': 'coo.md',
+  'planner.md': 'coo.md',
+  'documentationer.md': 'coo.md',
+  'generator-frontend.md': 'cto.md',
+  'generator-backend.md': 'cto.md',
+  'generator-backend-laravel.md': 'cto.md',
+  'generator-devops.md': 'cto.md',
+  'evaluator-architecture.md': 'cto.md',
+  'generator-designer.md': 'cdo.md',
+  'evaluator-visual.md': 'cdo.md',
+  'evaluator-functional.md': 'cqo.md',
+  'evaluator-code-quality.md': 'cqo.md',
+  'evaluator-security.md': 'cqo.md',
+  'meeting-manager.md': 'ops.md',
+  'service-ops.md': 'ops.md',
+};
+
 function removeLegacyV7Files(dir, files) {
   for (const file of files) {
     const target = path.join(dir, file);
@@ -174,6 +194,65 @@ function removeLegacyV7Files(dir, files) {
   }
 }
 
+function extractMarkdownEntryBlocks(md, prefix) {
+  const escapedPrefix = prefix.replace(/[.*+?^${}()|[\]\\]/g, '\\$&');
+  const re = new RegExp(
+    `^#{2,3}\\s+\\[(${escapedPrefix}-[A-Z0-9_-]+)\\][\\s\\S]*?(?=^#{2,3}\\s+\\[${escapedPrefix}-|$(?![\\s\\S]))`,
+    'gm',
+  );
+  const blocks = [];
+  let match;
+  while ((match = re.exec(md)) !== null) {
+    blocks.push({ id: match[1], body: match[0].trimEnd() });
+  }
+  return blocks;
+}
+
+function migrateLegacyRoleEntries(dir, files, kind) {
+  const prefix = kind === 'gotchas' ? 'G' : 'C';
+  const migrated = [];
+  const backupDir = path.join(HARNESS_DIR, 'archive', 'pre-v7-role-migration');
+  for (const file of files) {
+    const sourcePath = path.join(dir, file);
+    if (!fs.existsSync(sourcePath)) continue;
+
+    const content = fs.readFileSync(sourcePath, 'utf8');
+    const blocks = extractMarkdownEntryBlocks(content, prefix);
+    if (!blocks.length) continue;
+
+    const targetFile = LEGACY_V7_FILE_OWNER[file] || 'shared.md';
+    const targetPath = path.join(dir, targetFile);
+    const targetOriginal = fs.existsSync(targetPath) ? fs.readFileSync(targetPath, 'utf8') : '';
+    const targetIds = new Set(
+      kind === 'gotchas'
+        ? extractGotchaEntryIds(targetOriginal)
+        : extractConventionEntryIds(targetOriginal),
+    );
+    const appendBlocks = [];
+    for (const block of blocks) {
+      if (targetIds.has(block.id)) continue;
+      appendBlocks.push([
+        block.body,
+        '',
+        `- **Migration source**: ${kind}/${file}`,
+      ].join('\n'));
+      targetIds.add(block.id);
+    }
+    if (!appendBlocks.length) continue;
+
+    ensureDir(backupDir);
+    fs.writeFileSync(path.join(backupDir, `${kind}-${file}`), content);
+    const sep = targetOriginal.trim() ? '\n\n' : '';
+    fs.writeFileSync(targetPath, targetOriginal.replace(/\s*$/, '') + sep + appendBlocks.join('\n\n') + '\n');
+    migrated.push(`${file} → ${targetFile} (${appendBlocks.length})`);
+  }
+
+  if (migrated.length) {
+    log(`${kind} legacy role entries migrated: ${migrated.join(', ')}`);
+    log(`Legacy role migration backup: ${backupDir}`);
+  }
+}
+
 // ─────────────────────────────────────────
 // First-install migration — extract Convention/Gotcha-shaped sections from
 // existing CLAUDE.md / AGENTS.md into .harness/conventions and .harness/gotchas.
@@ -337,6 +416,9 @@ function scaffoldHarness() {
   ensureDir(path.join(HARNESS_DIR, 'memories'));
   ensureDir(path.join(HARNESS_DIR, 'shared'));
 
+  migrateLegacyRoleEntries(path.join(HARNESS_DIR, 'conventions'), V7_REMOVED_CONVENTION_FILES, 'conventions');
+  migrateLegacyRoleEntries(path.join(HARNESS_DIR, 'gotchas'), V7_REMOVED_GOTCHA_FILES, 'gotchas');
+
   removeLegacyV7Files(path.join(HARNESS_DIR, 'conventions'), V7_REMOVED_CONVENTION_FILES);
   removeLegacyV7Files(path.join(HARNESS_DIR, 'gotchas'), V7_REMOVED_GOTCHA_FILES);
 
@@ -362,7 +444,7 @@ function scaffoldHarness() {
   // a file that has such entries, or user learning history is lost.
   const gotchasSrc = path.join(PKG_ROOT, 'gotchas');
   if (fs.existsSync(gotchasSrc)) {
-    const ENTRY_PATTERN = /^### \[G-\d+\]/m;  // Gotcha entry heading
+    const ENTRY_PATTERN = /^### \[G-[A-Z0-9_-]+\]/m;  // Gotcha entry heading
     const CUSTOM_MARKER = '## Custom Gotchas'; // Legacy marker still supported
     const files = fs.readdirSync(gotchasSrc);
     for (const file of files) {
@@ -777,10 +859,10 @@ function installStatusline() {
 }
 
 // ─────────────────────────────────────────
-// 3d. UserPromptSubmit hook (auto dispatcher routing)
+// 3d. UserPromptSubmit hook (v7 CEO routing guidance)
 // ─────────────────────────────────────────
 function installUserPromptSubmitHook() {
-  log('Installing UserPromptSubmit hook (auto dispatcher routing)...');
+  log('Installing UserPromptSubmit hook (v7 CEO routing guidance)...');
 
   const settingsDir = path.join(PROJECT_ROOT, '.claude');
 
@@ -814,9 +896,9 @@ function installUserPromptSubmitHook() {
     });
     writeClaudeSettings(settings);
     log('UserPromptSubmit hook installed in .claude/settings.json');
-    log('  → All prompts will be routed through harness-dispatcher');
+    log('  → /goal and /hot-fix are guided through harness-ceo and CXX workflow');
     log('  → Opt-out per message: say "harness skip" or "without harness"');
-    log('  → Disable globally: set .harness/config.json behavior.auto_route_dispatcher = false');
+    log('  → Disable globally: set .harness/config.json behavior.auto_route_ceo = false');
   } else {
     log('UserPromptSubmit hook already installed');
   }
@@ -1103,7 +1185,9 @@ function detectMigrationNeeded() {
   const memoryTplPath = path.join(PKG_ROOT, 'assets', 'templates', 'memory.md');
   const flags = {
     progressV3toV4: false,
+    progressLegacyRouting: false,
     configMissingCompanyMode: false,
+    configLegacyRouting: false,
     memoryMissingSystemEntries: [],
     gotchaMissingEntries: {},   // { "<filename>": [G-IDs...] }
     conventionMissingEntries: {}, // { "<filename>": [C-IDs...] }
@@ -1164,12 +1248,22 @@ function detectMigrationNeeded() {
     try {
       const p = JSON.parse(fs.readFileSync(progressPath, 'utf8'));
       if ((p.version ?? 0) < TARGET_PROGRESS_VERSION) flags.progressV3toV4 = true;
+      if (
+        ['dispatcher', 'planner', 'generator-backend', 'generator-frontend', 'evaluator-functional', 'evaluator-visual', 'service-ops'].includes(p.next_agent) ||
+        p.mode_decision?.owner === 'conductor'
+      ) flags.progressLegacyRouting = true;
     } catch {}
   }
   if (fs.existsSync(configPath)) {
     try {
       const c = JSON.parse(fs.readFileSync(configPath, 'utf8'));
       if (!c.company_mode || c.mode_selection) flags.configMissingCompanyMode = true;
+      if (
+        c.behavior?.auto_route_dispatcher !== undefined ||
+        c.behavior?.auto_route_dispatcher_description ||
+        c.company_mode?.owner === 'conductor' ||
+        c.agents?.dispatcher?.skill === 'harness-dispatcher'
+      ) flags.configLegacyRouting = true;
     } catch {}
   }
   if (fs.existsSync(memoryPath) && fs.existsSync(memoryTplPath)) {
@@ -1245,15 +1339,21 @@ function extractConventionEntryBlock(md, id) {
 function showMigrationProposal(flags) {
   console.log('');
   console.log('╔══════════════════════════════════════════════════════════╗');
-  console.log('║  walwal-harness v6 — 자동 마이그레이션 사용 가능         ║');
+  console.log('║  walwal-harness v7 — 자동 마이그레이션 사용 가능         ║');
   console.log('╚══════════════════════════════════════════════════════════╝');
   if (flags.progressV3toV4) {
     console.log('  • progress.json: legacy mode → "company"');
     console.log('    legacy mode 값은 보존하지 않고 회사모드로 정규화됩니다.');
   }
+  if (flags.progressLegacyRouting) {
+    console.log('  • progress.json: legacy dispatcher/conductor routing → v7 CEO routing');
+  }
   if (flags.configMissingCompanyMode) {
     console.log('  • config.json: company_mode 섹션 동기화 가능');
-    console.log('    사용자 모드 선택 없이 회사모드 병렬 실행으로 고정합니다.');
+    console.log('    Owner /goal, /hot-fix → CEO/CXX/worker 흐름으로 고정합니다.');
+  }
+  if (flags.configLegacyRouting) {
+    console.log('  • config.json: legacy dispatcher/conductor wording → v7 CEO/CXX wording');
   }
   if (flags.memoryMissingSystemEntries && flags.memoryMissingSystemEntries.length) {
     console.log('  • memory.md: 시스템 entry 누락 — append 가능');
@@ -1291,17 +1391,21 @@ function runMigrate(opts = {}) {
   const dryRun = opts.dryRun || false;
   const flags = detectMigrationNeeded();
   const gotchaMissingTotal = Object.values(flags.gotchaMissingEntries || {}).reduce((n, a) => n + a.length, 0);
+  const conventionMissingTotal = Object.values(flags.conventionMissingEntries || {}).reduce((n, a) => n + a.length, 0);
   if (
     !flags.progressV3toV4 &&
+    !flags.progressLegacyRouting &&
     !flags.configMissingCompanyMode &&
+    !flags.configLegacyRouting &&
     (!flags.memoryMissingSystemEntries || flags.memoryMissingSystemEntries.length === 0) &&
     gotchaMissingTotal === 0 &&
+    conventionMissingTotal === 0 &&
     !flags.bundleVersionStale
   ) {
     console.log('');
     let pkgVer = 'unknown';
     try { pkgVer = JSON.parse(fs.readFileSync(path.join(PKG_ROOT, 'package.json'), 'utf8')).version; } catch {}
-    log(`이미 최신 — bundle v${pkgVer} 일치, progress v${TARGET_PROGRESS_VERSION}, config.company_mode, memory 시스템 entry, gotcha entry 모두 sync.`);
+    log(`이미 최신 — bundle v${pkgVer} 일치, progress v${TARGET_PROGRESS_VERSION} v7 routing, config.company_mode, memory 시스템 entry, gotcha entry 모두 sync.`);
     return;
   }
 
@@ -1314,7 +1418,7 @@ function runMigrate(opts = {}) {
 
   // 1. progress.json
   const progressPath = path.join(HARNESS_DIR, 'progress.json');
-  if (flags.progressV3toV4 && fs.existsSync(progressPath)) {
+  if ((flags.progressV3toV4 || flags.progressLegacyRouting) && fs.existsSync(progressPath)) {
     const original = fs.readFileSync(progressPath, 'utf8');
     const p = JSON.parse(original);
     const oldMode = p.mode ?? 'company';
@@ -1323,14 +1427,17 @@ function runMigrate(opts = {}) {
       version: TARGET_PROGRESS_VERSION,
       mode: 'company',
       mode_decision: {
-        owner: 'conductor',
+        owner: 'ceo',
         decided_at: null,
-        rationale: `migration — legacy mode="${oldMode}" normalized to company`,
-        policy: 'always_company_parallel',
+        rationale: `migration — legacy mode="${oldMode}" normalized to v7 CEO-led company`,
+        policy: 'owner_goal_to_ceo_cxx_workers',
         user_override: null,
       },
+      next_agent: ['dispatcher', 'planner', 'generator-backend', 'generator-frontend', 'evaluator-functional', 'evaluator-visual', 'service-ops'].includes(p.next_agent)
+        ? 'ceo'
+        : (p.next_agent ?? 'ceo'),
     };
-    log(`  progress.json: version 3 → ${TARGET_PROGRESS_VERSION}, mode "${oldMode}" → "company"`);
+    log(`  progress.json: v7 routing normalized, mode "${oldMode}" → "company", next_agent="${p.next_agent ?? 'null'}" → "${newP.next_agent}"`);
     if (!dryRun) {
       fs.writeFileSync(path.join(backupDir, 'progress.json'), original);
       fs.writeFileSync(progressPath, JSON.stringify(newP, null, 2) + '\n');
@@ -1340,14 +1447,26 @@ function runMigrate(opts = {}) {
   // 2. config.json — inject company_mode from template if missing
   const configPath = path.join(HARNESS_DIR, 'config.json');
   const tplPath = path.join(PKG_ROOT, 'assets', 'templates', 'config.json');
-  if (flags.configMissingCompanyMode && fs.existsSync(configPath) && fs.existsSync(tplPath)) {
+  if ((flags.configMissingCompanyMode || flags.configLegacyRouting) && fs.existsSync(configPath) && fs.existsSync(tplPath)) {
     const original = fs.readFileSync(configPath, 'utf8');
     const c = JSON.parse(original);
     const tpl = JSON.parse(fs.readFileSync(tplPath, 'utf8'));
     if (tpl.company_mode) {
-      c.company_mode = tpl.company_mode;
+      c.company_mode = {
+        ...tpl.company_mode,
+        comment: 'v7 — 회사모드는 CEO/CXX/worker 문서 흐름으로 운영된다.',
+        owner: 'ceo',
+      };
+      c.harness = tpl.harness || c.harness;
+      c.agents = tpl.agents || c.agents;
+      c.flow = tpl.flow || c.flow;
+      c.behavior = c.behavior || {};
+      c.behavior.auto_route_ceo = c.behavior.auto_route_ceo ?? c.behavior.auto_route_dispatcher ?? true;
+      c.behavior.auto_route_ceo_description = 'true 이면 /goal 또는 /hot-fix 이후 Owner 입력을 v7 CEO/CXX mission flow 기준으로 안내한다.';
+      delete c.behavior.auto_route_dispatcher;
+      delete c.behavior.auto_route_dispatcher_description;
       delete c.mode_selection;
-      log('  config.json: company_mode 섹션 주입 (always-on company parallel)');
+      log('  config.json: v7 CEO/CXX routing 섹션 동기화');
       if (!dryRun) {
         fs.writeFileSync(path.join(backupDir, 'config.json'), original);
         fs.writeFileSync(configPath, JSON.stringify(c, null, 2) + '\n');
@@ -1440,60 +1559,90 @@ function runMigrate(opts = {}) {
     log('Dry-run 완료 — 실제 변경 적용하려면 `npx walwal-harness migrate` 실행');
   } else {
     log(`Migration 완료. 백업: ${backupDir}`);
-    log('Conductor 가 다음 sprint 시작 시 자동으로 모드 결정합니다.');
+    log('v7 routing 기준으로 /goal 또는 /hot-fix 가 CEO/CXX/worker 흐름을 시작합니다.');
     log('회사모드는 기본값이며 사용자 override 는 사용하지 않습니다.');
   }
   console.log('');
 }
 
 // ─────────────────────────────────────────
-// Verify — 14 SKILL invariants + spawn whitelist + progress schema
+// Verify — v7 CXX/core skill invariants + progress schema
 // ─────────────────────────────────────────
 function runVerify() {
   const expectedSkills = [
-    'dispatcher', 'conductor', 'meeting-manager', 'planner',
-    'cto', 'cqo', 'service-ops',
-    'generator-backend', 'generator-frontend', 'generator-designer', 'generator-devops',
-    'evaluator-code-quality', 'evaluator-functional', 'evaluator-visual',
-    'evaluator-architecture', 'evaluator-security',
-    'coo-developer', 'documentationer',
-    'brainstorming',
+    'ceo', 'coo', 'cdo', 'cto', 'cqo', 'ops',
+    'hiring', 'resource-manager', 'brick-office',
   ];
   const requiredFrontmatter = ['name', 'description'];
 
   console.log('');
-  log('=== Verify: skill invariants + spawn whitelist + progress schema ===');
+  log('=== Verify: v7 command/skill invariants + progress schema ===');
   let pass = 0;
   let fail = 0;
   const issues = [];
 
   // 1) skill files
+  const skillRoots = [
+    ['.claude', CLAUDE_SKILLS_DIR],
+    ['.codex', CODEX_SKILLS_DIR],
+  ];
   for (const s of expectedSkills) {
-    const local = path.join(CLAUDE_SKILLS_DIR, `harness-${s}`, 'SKILL.md');
-    const exists = fs.existsSync(local);
-    if (!exists) {
-      issues.push(`  ✗ skills/harness-${s}/SKILL.md MISSING`);
-      fail++;
-      continue;
+    for (const [toolName, root] of skillRoots) {
+      const local = path.join(root, `harness-${s}`, 'SKILL.md');
+      const exists = fs.existsSync(local);
+      if (!exists) {
+        issues.push(`  ✗ ${toolName}/skills/harness-${s}/SKILL.md MISSING`);
+        fail++;
+        continue;
+      }
+      const body = fs.readFileSync(local, 'utf8');
+      const fmMatch = body.match(/^---\n([\s\S]*?)\n---/);
+      if (!fmMatch) {
+        issues.push(`  ✗ ${toolName}/skills/harness-${s}: frontmatter 누락`);
+        fail++;
+        continue;
+      }
+      const missing = requiredFrontmatter.filter((k) => !new RegExp(`^${k}:`, 'm').test(fmMatch[1]));
+      if (missing.length) {
+        issues.push(`  ✗ ${toolName}/skills/harness-${s}: frontmatter [${missing.join(',')}] 누락`);
+        fail++;
+        continue;
+      }
+      pass++;
     }
-    const body = fs.readFileSync(local, 'utf8');
-    const fmMatch = body.match(/^---\n([\s\S]*?)\n---/);
-    if (!fmMatch) {
-      issues.push(`  ✗ harness-${s}: frontmatter 누락`);
-      fail++;
-      continue;
+  }
+  log(`  skills: ${pass}/${expectedSkills.length * skillRoots.length} OK`);
+
+  // 2) owner-facing commands
+  const allowedCommands = new Set(['goal.md', 'hot-fix.md']);
+  const forbiddenCommands = new Set(['ceo.md', 'coo.md', 'cdo.md', 'cto.md', 'cqo.md', 'ops.md', 'hiring.md', 'resource-manager.md', 'brick-office.md']);
+  const commandRoots = [
+    ['.claude', CLAUDE_COMMANDS_DIR],
+    ['.codex', CODEX_COMMANDS_DIR],
+  ];
+  for (const [toolName, root] of commandRoots) {
+    for (const cmd of allowedCommands) {
+      if (!fs.existsSync(path.join(root, cmd))) {
+        issues.push(`  ✗ ${toolName}/commands/${cmd} MISSING`);
+        fail++;
+      } else {
+        pass++;
+      }
     }
-    const missing = requiredFrontmatter.filter((k) => !new RegExp(`^${k}:`, 'm').test(fmMatch[1]));
-    if (missing.length) {
-      issues.push(`  ✗ harness-${s}: frontmatter [${missing.join(',')}] 누락`);
-      fail++;
-      continue;
+    if (fs.existsSync(root)) {
+      for (const cmd of fs.readdirSync(root).filter(f => f.endsWith('.md'))) {
+        if (forbiddenCommands.has(cmd)) {
+          issues.push(`  ✗ ${toolName}/commands/${cmd} forbidden in v7`);
+          fail++;
+        } else if (!allowedCommands.has(cmd)) {
+          issues.push(`  ⚠ ${toolName}/commands/${cmd} is outside walwal v7 owner-facing command set`);
+        }
+      }
     }
-    pass++;
   }
-  log(`  skills: ${pass}/${expectedSkills.length} OK`);
+  log(`  commands: /goal + /hot-fix checked`);
 
-  // 2) progress.json schema (v6 = version 4 + mode_decision)
+  // 3) progress.json schema (v6+ = version 4 + mode_decision)
   const progressPath = path.join(HARNESS_DIR, 'progress.json');
   if (fs.existsSync(progressPath)) {
     try {
@@ -1502,6 +1651,10 @@ function runVerify() {
       if ((p.version ?? 0) < 4) schemaIssues.push(`version=${p.version} (<4)`);
       if (!p.mode_decision) schemaIssues.push('mode_decision 누락');
       if (!p.dispatch) schemaIssues.push('dispatch 누락');
+      if (p.mode_decision?.owner === 'conductor') schemaIssues.push('mode_decision.owner=conductor legacy');
+      if (['dispatcher', 'planner', 'generator-backend', 'generator-frontend', 'evaluator-functional', 'evaluator-visual', 'service-ops'].includes(p.next_agent)) {
+        schemaIssues.push(`next_agent=${p.next_agent} legacy`);
+      }
       if (schemaIssues.length) {
         issues.push(`  ✗ progress.json schema: ${schemaIssues.join(', ')} → npx walwal-harness migrate`);
         fail++;
@@ -1523,6 +1676,13 @@ function runVerify() {
       if (!c.company_mode) {
         issues.push('  ✗ config.json: company_mode 누락 → npx walwal-harness migrate');
         fail++;
+      } else if (
+        c.company_mode.owner === 'conductor' ||
+        c.behavior?.auto_route_dispatcher !== undefined ||
+        c.agents?.dispatcher?.skill === 'harness-dispatcher'
+      ) {
+        issues.push('  ✗ config.json: legacy dispatcher/conductor routing → npx walwal-harness migrate');
+        fail++;
       } else {
         pass++;
         log(`  config.json: company_mode.owner=${c.company_mode.owner} OK`);
@@ -1552,6 +1712,20 @@ function runVerify() {
     issues.push('  ⚠ .claude/commands/harness-next.md 존재 — v6.0.3 부터 회사 내부 도구로 전환됨. `npx walwal-harness --force` 또는 직접 삭제 권장');
   }
 
+  // 6) installed script guidance must not claim removed dispatcher/planner flow.
+  const routeScript = path.join(PROJECT_ROOT, 'scripts', 'harness-user-prompt-submit.sh');
+  const sessionScript = path.join(PROJECT_ROOT, 'scripts', 'harness-session-start.sh');
+  for (const scriptPath of [routeScript, sessionScript]) {
+    if (!fs.existsSync(scriptPath)) continue;
+    const body = fs.readFileSync(scriptPath, 'utf8');
+    if (/harness-dispatcher|pipeline=none\/dispatcher|dispatch -> CEO meeting|planner\(COO\)|gen\/eval/.test(body)) {
+      issues.push(`  ✗ ${path.relative(PROJECT_ROOT, scriptPath)}: legacy dispatcher/planner guidance → npx walwal-harness init migrate --force`);
+      fail++;
+    } else {
+      pass++;
+    }
+  }
+
   console.log('');
   if (fail === 0) {
     log(`✓ Verify PASS — ${pass} 개 invariant 통과.`);
@@ -1618,7 +1792,9 @@ function main() {
   const gotchaMissing = Object.keys(migFlags.gotchaMissingEntries || {}).length > 0;
   const hasContentDrift =
     migFlags.progressV3toV4 ||
+    migFlags.progressLegacyRouting ||
     migFlags.configMissingCompanyMode ||
+    migFlags.configLegacyRouting ||
     (migFlags.memoryMissingSystemEntries && migFlags.memoryMissingSystemEntries.length) ||
     gotchaMissing;
   if (hasContentDrift) {
@@ -1631,7 +1807,7 @@ function main() {
     } catch {}
   }
 
-  // v6.0 — Brick Office dashboard one-liner
+  // Brick Office dashboard one-liner
   console.log('');
   console.log('╔══════════════════════════════════════════════════════════╗');
   console.log('║  Brick Office — 라이브 운영 대시보드 (선택)              ║');
@@ -1648,8 +1824,7 @@ function main() {
     log('╔═══════════════════════════════════════════════════════════╗');
     log('║  Restart Claude Code for skills & commands to activate!  ║');
     log('║                                                          ║');
-    log('║  Then say: "하네스 엔지니어링 시작"                        ║');
-    log('║  Or invoke: /harness-dispatcher                          ║');
+    log('║  Then invoke /goal or /hot-fix                           ║');
     log('║                                                          ║');
     log('║  기본은 회사모드: 병렬 · 자율 진행. 추가 입력 불필요.     ║');
     log('╚═══════════════════════════════════════════════════════════╝');
@@ -1660,6 +1835,10 @@ function main() {
     log('  3. Internal agents call hired agents/skills, not slash commands.');
   }
   console.log('');
+
+  if (subcommand === 'init' && subcommandArgs.includes('migrate')) {
+    runMigrate({ dryRun: args.includes('--dry-run') });
+  }
 }
 
 main();

package/init.sh

@@ -0,0 +1,208 @@
+#!/bin/bash
+# Harness Init Script — 적응형(adaptive) 하네스 진입점 허브
+#
+# 서브커맨드:
+#   help                               — 사용법 출력
+#   check         (default · 무인자)   — 환경 확인 + scan + ref 상태 출력 (기존 동작)
+#   init          [project_root]       — 신규 프로젝트 초기화 (scan → ref placeholder → AGENTS.md → migration preview)
+#   migrate       [project_root]       — v7.1 마이그레이션 적용
+#   refresh-ref   <role> <stack>       — 기존 ref 파일 archive 백업 후 재생성 유도
+#
+# 스택별 서버 기동은 scan-result.json.tech_stack 과 .harness/ref/<role>-<stack>.md.runner 에서
+# 가져온다. NestJS/React 하드코딩은 제거되었다.
+
+set -e
+
+SUBCOMMAND="${1:-check}"
+shift || true
+
+usage() {
+  cat <<'USAGE'
+Harness Init — 적응형 하네스 진입점
+
+Usage:
+  bash init.sh help                                 — 이 메시지
+  bash init.sh check         [project_root]         — 환경 확인 + scan + ref 상태 (기본)
+  bash init.sh init          [project_root]         — scan + ref placeholder 생성 + AGENTS.md + migration dry-run
+  bash init.sh init          [project_root] --apply-migration
+                                                    — init 후 v7.1 마이그레이션까지 적용
+  bash init.sh migrate       [project_root]         — v7.1 마이그레이션 적용
+  bash init.sh migrate       [project_root] --dry-run
+                                                    — v7.1 마이그레이션 미리보기
+  bash init.sh refresh-ref   <role> <stack> [path]  — 기존 ref 파일 백업 후 재생성 안내
+
+스택 독립: 개발 서버 기동 명령은 .harness/ref/<role>-<stack>.md 의 runner.dev_command 에서 조회한다.
+USAGE
+}
+
+run_walwal_cli() {
+  local project_root="$1"
+  shift || true
+  local script_dir
+  script_dir="$(cd "$(dirname "$0")" && pwd)"
+  local cli_path="${script_dir}/bin/init.js"
+
+  if [ -f "$cli_path" ]; then
+    (cd "$project_root" && node "$cli_path" "$@")
+  else
+    (cd "$project_root" && npx walwal-harness "$@")
+  fi
+}
+
+run_migrate() {
+  local project_root="."
+  local mode=""
+
+  if [ "${1:-}" = "--dry-run" ]; then
+    mode="--dry-run"
+  else
+    project_root="${1:-.}"
+    mode="${2:-}"
+  fi
+
+  echo "=== Harness Migration: ${project_root} ==="
+  if [ "$mode" = "--dry-run" ]; then
+    run_walwal_cli "$project_root" migrate --dry-run
+  else
+    run_walwal_cli "$project_root" migrate
+  fi
+}
+
+run_check() {
+  local project_root="${1:-.}"
+  echo "=== Harness Environment Check ==="
+  echo "Working Directory: $(pwd)"
+  echo "Date: $(date)"
+  echo ""
+
+  # Git 상태
+  echo "--- Git Status ---"
+  if [ -d "${project_root}/.git" ]; then
+    git -C "$project_root" log --oneline -5 2>/dev/null || echo "No commits yet"
+    echo ""
+    git -C "$project_root" status --short
+  else
+    echo "Not a git repository — run: git init"
+  fi
+  echo ""
+
+  # Harness 상태
+  echo "--- Harness Status ---"
+  if [ -f "${project_root}/.harness/progress.json" ]; then
+    if [ -f "${project_root}/scripts/lib/harness-render-progress.sh" ]; then
+      (cd "$project_root" && source scripts/lib/harness-render-progress.sh && render_progress "." 2>/dev/null) || cat "${project_root}/.harness/progress.json"
+    else
+      cat "${project_root}/.harness/progress.json"
+    fi
+  else
+    echo "progress.json not found — run: bash init.sh init"
+  fi
+  echo ""
+
+  # scan 상태
+  echo "--- Scan Result ---"
+  if [ -f "${project_root}/.harness/actions/scan-result.json" ]; then
+    jq '{tech_stack, tech_stack_confidence, recommendation}' "${project_root}/.harness/actions/scan-result.json"
+  else
+    echo "scan-result.json not found — run: bash scripts/scan-project.sh ${project_root}"
+  fi
+  echo ""
+
+  # ref 상태
+  echo "--- Ref Docs ---"
+  if [ -d "${project_root}/.harness/ref" ]; then
+    ls -1 "${project_root}/.harness/ref/" 2>/dev/null | grep -v '^\.generated' | sed 's/^/  /' || echo "  (none)"
+  else
+    echo "  .harness/ref/ not present — run: bash init.sh init"
+  fi
+  echo ""
+
+  echo "=== Harness Ready ==="
+}
+
+run_init() {
+  local project_root="."
+  local migration_mode="--dry-run"
+  if [ "${1:-}" = "--apply-migration" ]; then
+    migration_mode="--apply-migration"
+  else
+    project_root="${1:-.}"
+    if [ "${2:-}" = "--apply-migration" ]; then
+      migration_mode="--apply-migration"
+    fi
+  fi
+  echo "=== Harness Init: ${project_root} ==="
+
+  # 1. scan
+  echo "[1/4] scan-project.sh"
+  bash "$(dirname "$0")/scripts/scan-project.sh" "$project_root"
+  echo ""
+
+  # 2. ref placeholder 생성 (감지된 스택 전체)
+  echo "[2/4] init-ref-docs.sh (placeholder)"
+  bash "$(dirname "$0")/scripts/init-ref-docs.sh" --yes --placeholder "$project_root"
+  echo ""
+
+  # 3. AGENTS.md 동적 생성
+  if [ -f "$(dirname "$0")/scripts/init-agents-md.sh" ]; then
+    echo "[3/4] init-agents-md.sh"
+    bash "$(dirname "$0")/scripts/init-agents-md.sh" "$project_root"
+  else
+    echo "[3/4] init-agents-md.sh — skip (not found)"
+  fi
+  echo ""
+
+  # 4. v7.1 migration preview/apply
+  if [ "$migration_mode" = "--apply-migration" ]; then
+    echo "[4/4] walwal-harness migrate"
+    run_walwal_cli "$project_root" migrate
+  else
+    echo "[4/4] walwal-harness migrate --dry-run"
+    run_walwal_cli "$project_root" migrate --dry-run
+  fi
+  echo ""
+
+  echo "=== Init Complete ==="
+  echo "다음 단계:"
+  echo "  1. 각 .harness/ref/<role>-<stack>.md 를 실제 컨텐츠로 채우려면:"
+  echo "     bash init.sh refresh-ref <role> <stack> $project_root"
+  echo "  2. 그 결과 프롬프트를 Claude 세션에 붙여 WebSearch + WebFetch 로 ref 작성"
+  if [ "$migration_mode" != "--apply-migration" ]; then
+    echo "  3. 마이그레이션 미리보기가 적절하면 적용:"
+    echo "     bash init.sh migrate $project_root"
+  fi
+}
+
+run_refresh_ref() {
+  local role="$1" stack="$2" project_root="${3:-.}"
+  if [ -z "$role" ] || [ -z "$stack" ]; then
+    echo "ERROR: bash init.sh refresh-ref <role> <stack> [project_root]" >&2
+    exit 1
+  fi
+
+  local ref_path="${project_root}/.harness/ref/${role}-${stack}.md"
+  local archive_dir="${project_root}/.harness/archive"
+  mkdir -p "$archive_dir"
+
+  if [ -f "$ref_path" ]; then
+    local backup="${archive_dir}/ref-${role}-${stack}-$(date +%Y%m%d%H%M%S).md"
+    mv "$ref_path" "$backup"
+    echo "archive: $backup"
+  fi
+
+  echo ""
+  bash "$(dirname "$0")/scripts/init-ref-docs.sh" --claude-prompt --stack "$stack" --role "$role" "$project_root"
+}
+
+case "$SUBCOMMAND" in
+  help|-h|--help)      usage ;;
+  check)               run_check "$@" ;;
+  init)                run_init "$@" ;;
+  migrate)             run_migrate "$@" ;;
+  refresh-ref)         run_refresh_ref "$@" ;;
+  *)
+    echo "ERROR: unknown subcommand '$SUBCOMMAND'" >&2
+    usage
+    exit 1
+    ;;
+esac

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@walwal-harness/cli",
-  "version": "7.1.0",
+  "version": "7.1.2",
   "description": "Company-style AI agent harness for Claude and Codex. Installs commands, CXX agents, skills, HR-Resource hiring pool, and project-local .harness runtime state.",
   "bin": {
     "walwal-harness": "bin/init.js"
@@ -48,6 +48,7 @@
     "assets/",
     "gotchas/",
     "conventions/",
+    "init.sh",
     "CHANGELOG.md"
   ]
 }

package/scripts/harness-session-start.sh

@@ -168,24 +168,24 @@ fi
 # init 상태: 첫 안내
 # ─────────────────────────────────────────
 if [ "$sprint_status" = "init" ]; then
-  if [ "$next_agent" != "null" ] && [ "$next_agent" != "none" ] && [ "$next_agent" != "dispatcher" ]; then
+  if [ "$next_agent" != "null" ] && [ "$next_agent" != "none" ] && [ "$next_agent" != "ceo" ]; then
     echo "# Harness Company Mode active"
     echo "# Sprint is init, but next_agent=${next_agent}; continue the autonomous loop instead of waiting for Owner."
   else
-  echo "# Harness ready — say \"하네스 엔지니어링 시작\" or /harness-dispatcher"
-  echo "# 기본 경로는 회사 루프입니다: dispatch -> CEO meeting -> COO/CTO 분배 -> gen/eval -> CQO -> service-ops -> batch meeting."
+  echo "# Harness ready — use /goal or /hot-fix"
+  echo "# 기본 경로는 v7 회사 루프입니다: CEO -> 필요한 CXX -> resource-manager/hiring -> worker reports -> CXX review -> CEO report."
   exit 0
   fi
 fi
 
 # ─────────────────────────────────────────
-# Audit lifecycle: Planner/Dispatcher 시작 시 리셋
+# Audit lifecycle: CEO 시작 시 리셋
 # ─────────────────────────────────────────
 init_audit "$PROJECT_ROOT"
-if [ "$next_agent" = "planner" ] || [ "$next_agent" = "dispatcher" ]; then
+if [ "$next_agent" = "ceo" ]; then
   # 새 사이클 — 이전 audit을 archive로 이동하고 새로 시작
   reset_audit "$PROJECT_ROOT" "$sprint_num"
-  audit_log "system" "cycle" "start" "sprint-${sprint_num}" "new plan/dispatch cycle"
+  audit_log "system" "cycle" "start" "sprint-${sprint_num}" "new CEO mission cycle"
 fi
 
 # ─────────────────────────────────────────

package/scripts/harness-user-prompt-submit.sh

@@ -14,7 +14,7 @@ SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
 # 조건 2: opt-out 플래그 확인
 AUTO_ROUTE="true"
 if command -v jq >/dev/null 2>&1; then
-  AUTO_ROUTE=$(jq -r '.behavior.auto_route_dispatcher // true' "$CWD/.harness/config.json" 2>/dev/null || echo "true")
+  AUTO_ROUTE=$(jq -r '.behavior.auto_route_ceo // .behavior.auto_route_dispatcher // true' "$CWD/.harness/config.json" 2>/dev/null || echo "true")
 fi
 if [ "$AUTO_ROUTE" != "true" ]; then exit 0; fi
 
@@ -114,13 +114,13 @@ cat <<EOF
 [harness] company | S${SPRINT_NUM} | ${PIPELINE} | agent=${CURRENT_AGENT} (${AGENT_STATUS}) | next=${NEXT_AGENT} | queue=${T_PASSED}/${T_TOTAL} passed | failed=${T_FAILED}
 ${CONTEXT_WARNING}
 ## Route
-- pipeline=none/dispatcher 미실행 → harness-dispatcher 스킬 호출
-- 기본 경로는 conductor 중심 회사 루프다: dispatch -> CEO meeting -> planner(COO) -> cto -> gen/eval -> cqo -> service-ops -> batch meeting
-- worker pool(기본 3)이 자율적으로 Gen→Eval 루프 실행 중이며 control-plane은 이 상한에 포함되지 않는다
+- pipeline=none/init → harness-ceo 스킬로 Owner goal/hot-fix를 접수하고 mission 문서를 생성
+- 기본 경로는 v7 CXX 회사 루프다: CEO -> COO/CDO/CTO/CQO/OPS -> harness-resource-manager -> harness-hiring -> hired workers -> CXX review -> CEO report
+- CXX는 직접 전문 산출물을 만들지 않고 worker report를 `.harness/documents/{mission}/workers/`에 남겨야 한다
 - Owner에게 "계속 진행", "진행할까요?", "다음 명령을 입력하세요"를 요구하지 않는다. GOAL이 있고 escalation이 아니면 즉시 next_agent/current_agent 업무를 수행한다
 - Owner 입력은 목표 변경·사고·결과 확인용이지 회사 진행을 펌프하는 신호가 아니다
-- conductor 는 Planner 직행 대신 meeting-manager / cto / cqo / service-ops 로 next_agent를 재작성할 수 있다
-- 기능 요청 → pipeline flow | 실수 지적 → gotcha flow | 메타 질문 → 짧게 응답 (skip)
+- CEO는 필요한 CXX만 호출하고, CXX는 resource-manager/hiring 없이 worker 업무를 수행하지 않는다
+- 기능 요청 → CEO mission flow | 긴급 수정 → hot-fix flow | 메타 질문 → 짧게 응답 (skip)
 - 활성 pipeline → next_agent/current_agent 컨텍스트로 계속
 - skip: "harness skip", "just answer" 등 명시 시 단일 메시지 건너뜀
 EOF

package/commands/play-harness.md

@@ -1,22 +0,0 @@
----
-description: Run the harness play build and notify OPS monitoring.
-argument-hint: "[additional play args]"
----
-
-# /play-harness
-
-Run the target project's harness play mode.
-
-Required flow:
-1. Confirm `.harness/` exists in the current project.
-2. Run `bash scripts/play.sh $ARGUMENTS`.
-3. Treat play mode and release mode as mutually exclusive. If release mode is active, stop it first with `/stop-harness` or report the conflict to the Owner.
-4. After the script returns, read the emitted OPS report path and notify the installed `harness-ops` skill/agent that play monitoring has started.
-5. Record the build result in `.harness/documents/{mission_name}/ops.md` when a mission is active.
-6. Do not route this through `/goal`; this command is an Owner-facing runtime control entrypoint.
-
-Play arguments:
-
-```
-$ARGUMENTS
-```

package/commands/release-harness.md

@@ -1,22 +0,0 @@
----
-description: Run the harness release build, start service operation, and notify OPS monitoring.
-argument-hint: "[additional release args]"
----
-
-# /release-harness
-
-Run the target project's harness release mode and begin service operation.
-
-Required flow:
-1. Confirm `.harness/` exists in the current project.
-2. Run `bash scripts/release.sh release $ARGUMENTS`.
-3. Treat play mode and release mode as mutually exclusive. If play mode is active, stop it first with `/stop-harness` or report the conflict to the Owner.
-4. After the script returns, read the emitted OPS report path and notify the installed `harness-ops` skill/agent that release monitoring has started.
-5. Record the release result in `.harness/documents/{mission_name}/ops.md` when a mission is active.
-6. Do not route this through `/goal`; this command is an Owner-facing runtime control entrypoint.
-
-Release arguments:
-
-```
-$ARGUMENTS
-```

package/commands/stop-harness.md

@@ -1,21 +0,0 @@
----
-description: Stop any harness play or release process/session.
-argument-hint: "[optional stop reason]"
----
-
-# /stop-harness
-
-Stop any active harness play or release runtime.
-
-Required flow:
-1. Confirm `.harness/` exists in the current project.
-2. Run `bash scripts/harness-runtime-stop.sh $ARGUMENTS`.
-3. Verify both play and release runtime state files are cleared or marked stopped.
-4. Notify the installed `harness-ops` skill/agent that monitoring should stop for the terminated runtime mode.
-5. Record the stop result in `.harness/documents/{mission_name}/ops.md` when a mission is active.
-
-Stop reason:
-
-```
-$ARGUMENTS
-```