feat(study): 이론공부 홈 — 오늘의 개념·진도·회독 SR (Stage S)

개념문서(가스기사 289) 소비 표면 개선 1단계. /study 허브를 데일리 랜딩으로.
- 마이그 381 study_concept_progress (개념 SR, sr_schedule 공용, documents FK 없음=락 회피)
- concept_curriculum 서비스 + /api/study (curriculum·today-concepts·concepts/{id}/read)
- read 상태 정본 = document_reads (is_read 컬럼 아님), mark_read=회독+SR 입고
- 문제풀이 표면 무접촉·additive

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

app/api/documents.py

@@ -28,7 +28,7 @@ from sqlalchemy.ext.asyncio import AsyncSession
 from starlette.requests import ClientDisconnect
 
 from ai.client import AIClient, _load_prompt, parse_json_response
-from core.auth import get_current_user
+from core.auth import get_current_user, get_egress_class
 from core.config import settings
 from core.database import async_session, get_session
 from core.utils import file_hash
@@ -742,11 +742,31 @@ async def get_document(
     doc_id: int,
     user: Annotated[User, Depends(get_current_user)],
     session: Annotated[AsyncSession, Depends(get_session)],
+    egress_class: Annotated[str, Depends(get_egress_class)],
 ):
-    """문서 단건 조회. 본문(extracted_text)·canonical markdown 동봉."""
+    """문서 단건 조회. 본문(extracted_text)·canonical markdown 동봉.
+
+    cloud egress(갭2): egress=cloud 토큰(예: Claude/MCP)은 search 와 동일한 cloud-eligibility
+    게이트를 통과한 문서만 열람 가능 — id 직접 fetch 로 비공개/인프라/개인/restricted 문서를
+    우회 열람하는 경로를 차단한다. 부적격은 404(존재 자체 비노출). local 토큰=게이트 미발동(무회귀).
+    """
+    from sqlalchemy import text as sql_text
+    from services.search.retrieval_service import cloud_eligible_doc_sql
+
     doc = await session.get(Document, doc_id)
     if not doc or doc.deleted_at is not None:
         raise HTTPException(status_code=404, detail="문서를 찾을 수 없습니다")
+    if egress_class == "cloud":
+        eligible = (
+            await session.execute(
+                sql_text(
+                    "SELECT 1 FROM documents WHERE id = :doc_id AND deleted_at IS NULL"
+                    + cloud_eligible_doc_sql("")
+                ).bindparams(doc_id=doc_id)
+            )
+        ).first()
+        if eligible is None:
+            raise HTTPException(status_code=404, detail="문서를 찾을 수 없습니다")
     return DocumentDetailResponse.model_validate(doc)
 
 
@@ -1028,6 +1048,19 @@ async def get_document_image_raw(
             DocumentImage.image_key == image_key,
         )
     )
+    if img is None:
+        # clause-KB: 절-문서는 부모 표준 이미지를 공유(md_content=부모 슬라이스) → parent_id 폴백.
+        from sqlalchemy import text as sql_text
+        _par = (await session.execute(
+            sql_text("SELECT parent_id FROM documents WHERE id = :id").bindparams(id=doc_id)
+        )).scalar()
+        if _par is not None:
+            img = await session.scalar(
+                select(DocumentImage).where(
+                    DocumentImage.document_id == _par,
+                    DocumentImage.image_key == image_key,
+                )
+            )
     if img is None:
         raise HTTPException(status_code=404, detail="이미지를 찾을 수 없습니다")
 
@@ -1801,3 +1834,305 @@ async def analyze_document(
             error_code=error_code,
             source=source,
         )
+
+
+# ─── ASME 절-지식베이스: 유기적 책 네비 (clause-KB, doc_kind='clause' 자식 문서 기반) ───
+class ClauseTocItem(BaseModel):
+    id: int
+    clause_code: str | None = None
+    clause_part: str | None = None
+    clause_order: int | None = None
+    title: str | None = None
+
+
+class ClauseBookResponse(BaseModel):
+    parent_id: int
+    parent_title: str | None = None
+    clauses: list[ClauseTocItem]
+
+
+@router.get("/{doc_id}/clauses", response_model=ClauseBookResponse)
+async def get_document_clauses(
+    doc_id: int,
+    user: Annotated[User, Depends(get_current_user)],
+    session: Annotated[AsyncSession, Depends(get_session)],
+):
+    """부모 표준 doc 의 절-문서 목차(유기적 책 TOC). doc_kind='clause' 자식을 clause_order 순 반환.
+
+    절-문서는 in_corpus=false + doc_kind='clause'(검색 제외)라 일반 목록/검색엔 안 뜨지만,
+    이 책-내 네비는 부모 표준에서 자식 절로 진입하는 전용 경로다(ASME 2025판=한 권의 책).
+    """
+    from sqlalchemy import text as sql_text
+
+    parent = await session.get(Document, doc_id)
+    if not parent or parent.deleted_at is not None:
+        raise HTTPException(status_code=404, detail="문서를 찾을 수 없습니다")
+    rows = (
+        await session.execute(
+            sql_text(
+                """
+                SELECT id, clause_code, clause_part, clause_order, title
+                FROM documents
+                WHERE parent_id = :pid AND doc_kind = 'clause' AND deleted_at IS NULL
+                ORDER BY clause_order
+                """
+            ).bindparams(pid=doc_id)
+        )
+    ).mappings().all()
+    return ClauseBookResponse(
+        parent_id=doc_id,
+        parent_title=parent.title,
+        clauses=[ClauseTocItem(**dict(r)) for r in rows],
+    )
+
+
+class BacklinkRef(BaseModel):
+    code: str
+    doc_id: int | None = None   # 해소된 절-문서(같은 부모) — dangling 이면 None
+    title: str | None = None
+    anchor: str | None = None
+    ctx: str | None = None
+
+
+class BacklinksResponse(BaseModel):
+    doc_id: int
+    clause_code: str | None = None
+    parent_id: int | None = None
+    prev: ClauseTocItem | None = None
+    next: ClauseTocItem | None = None
+    forward: list[BacklinkRef]   # 이 절이 참조하는 절들
+    back: list[BacklinkRef]      # 이 절을 참조하는 절들
+
+
+@router.get("/{doc_id}/backlinks", response_model=BacklinksResponse)
+async def get_document_backlinks(
+    doc_id: int,
+    user: Annotated[User, Depends(get_current_user)],
+    session: Annotated[AsyncSession, Depends(get_session)],
+):
+    """절-문서의 양방향 백링크 + 같은 부모 내 이전/다음 절(유기적 책 흐름)."""
+    from sqlalchemy import text as sql_text
+
+    doc = await session.get(Document, doc_id)
+    if not doc or doc.deleted_at is not None:
+        raise HTTPException(status_code=404, detail="문서를 찾을 수 없습니다")
+
+    _meta = (await session.execute(sql_text(
+        "SELECT parent_id, clause_code, clause_order FROM documents WHERE id = :id"
+    ).bindparams(id=doc_id))).mappings().first()
+    _parent_id = _meta["parent_id"] if _meta else None
+    _clause_code = _meta["clause_code"] if _meta else None
+    _clause_order = _meta["clause_order"] if _meta else None
+    forward = (
+        await session.execute(
+            sql_text(
+                """
+                SELECT cl.dst_code AS code, cl.dst_doc_id AS doc_id, cl.anchor, cl.ctx, d.title
+                FROM clause_links cl
+                LEFT JOIN documents d ON d.id = cl.dst_doc_id
+                WHERE cl.src_doc_id = :id
+                ORDER BY cl.char_off NULLS LAST
+                LIMIT 300
+                """
+            ).bindparams(id=doc_id)
+        )
+    ).mappings().all()
+    back = (
+        await session.execute(
+            sql_text(
+                """
+                SELECT s.clause_code AS code, cl.src_doc_id AS doc_id, s.title, cl.ctx
+                FROM clause_links cl
+                JOIN documents s ON s.id = cl.src_doc_id
+                WHERE cl.dst_doc_id = :id
+                ORDER BY s.clause_order NULLS LAST
+                LIMIT 300
+                """
+            ).bindparams(id=doc_id)
+        )
+    ).mappings().all()
+
+    prev = nxt = None
+    if _parent_id is not None and _clause_order is not None:
+        prow = (
+            await session.execute(
+                sql_text(
+                    """
+                    SELECT id, clause_code, clause_part, clause_order, title FROM documents
+                    WHERE parent_id = :pid AND doc_kind='clause' AND deleted_at IS NULL
+                      AND clause_order < :ord
+                    ORDER BY clause_order DESC LIMIT 1
+                    """
+                ).bindparams(pid=_parent_id, ord=_clause_order)
+            )
+        ).mappings().first()
+        nrow = (
+            await session.execute(
+                sql_text(
+                    """
+                    SELECT id, clause_code, clause_part, clause_order, title FROM documents
+                    WHERE parent_id = :pid AND doc_kind='clause' AND deleted_at IS NULL
+                      AND clause_order > :ord
+                    ORDER BY clause_order ASC LIMIT 1
+                    """
+                ).bindparams(pid=_parent_id, ord=_clause_order)
+            )
+        ).mappings().first()
+        prev = ClauseTocItem(**dict(prow)) if prow else None
+        nxt = ClauseTocItem(**dict(nrow)) if nrow else None
+
+    return BacklinksResponse(
+        doc_id=doc_id,
+        clause_code=_clause_code,
+        parent_id=_parent_id,
+        prev=prev,
+        next=nxt,
+        forward=[BacklinkRef(**dict(r)) for r in forward],
+        back=[BacklinkRef(**dict(r)) for r in back],
+    )
+
+
+# ─── 관련 문서 (유사도, on-demand pgvector KNN — 저부하·무저장) ───
+class RelatedItem(BaseModel):
+    id: int
+    title: str | None = None
+    ai_domain: str | None = None
+    material_type: str | None = None
+    year: int | None = None
+    sim: float | None = None
+
+
+class RelatedResponse(BaseModel):
+    doc_id: int
+    related: list[RelatedItem]
+
+
+@router.get("/{doc_id}/related", response_model=RelatedResponse)
+async def get_related_documents(
+    doc_id: int,
+    user: Annotated[User, Depends(get_current_user)],
+    session: Annotated[AsyncSession, Depends(get_session)],
+    limit: int = 8,
+    same_type: bool = True,
+):
+    """문서-레벨 임베딩 코사인 최근접 = '관련 문서'. on-demand(저장/배치 없음).
+
+    인용그래프가 부적합한 코퍼스(업계 기술기사=인용망 부재)의 대안 연결 레이어.
+    same_type=true면 같은 material_type 내, false면 전 코퍼스. doc_kind='clause'(절-문서)는 제외.
+    """
+    from sqlalchemy import text as sql_text
+
+    lim = max(1, min(limit, 30))
+    type_clause = "AND d.material_type = src.material_type" if same_type else ""
+    rows = (
+        await session.execute(
+            sql_text(
+                f"""
+                WITH src AS (
+                    SELECT embedding, material_type FROM documents WHERE id = :id
+                )
+                SELECT d.id, d.title, d.ai_domain, d.material_type, d.facet_year AS year,
+                       round((1 - (d.embedding <=> (SELECT embedding FROM src)))::numeric, 3) AS sim
+                FROM documents d, src
+                WHERE d.doc_kind = 'standard' AND d.deleted_at IS NULL
+                  AND d.id <> :id AND d.embedding IS NOT NULL
+                  AND (SELECT embedding FROM src) IS NOT NULL
+                  {type_clause}
+                ORDER BY d.embedding <=> (SELECT embedding FROM src)
+                LIMIT :lim
+                """
+            ).bindparams(id=doc_id, lim=lim)
+        )
+    ).mappings().all()
+    return RelatedResponse(
+        doc_id=doc_id,
+        related=[RelatedItem(**{k: r[k] for k in ("id", "title", "ai_domain", "material_type", "year")}, sim=float(r["sim"]) if r["sim"] is not None else None) for r in rows],
+    )
+
+
+# ─── 절 공부도구 (노트/형광펜/암기카드) — clause_study ───
+class StudyItem(BaseModel):
+    id: int
+    kind: str
+    payload: dict = {}
+    created_at: datetime | None = None
+
+
+class StudyListResponse(BaseModel):
+    doc_id: int
+    items: list[StudyItem]
+
+
+class StudyCreate(BaseModel):
+    kind: str           # note | highlight | card
+    payload: dict = {}
+
+
+def _parse_payload(p):
+    import json
+    if isinstance(p, str):
+        try:
+            return json.loads(p)
+        except Exception:
+            return {}
+    return p or {}
+
+
+@router.get("/{doc_id}/study", response_model=StudyListResponse)
+async def list_study(
+    doc_id: int,
+    user: Annotated[User, Depends(get_current_user)],
+    session: Annotated[AsyncSession, Depends(get_session)],
+):
+    """절-문서의 공부도구 항목(노트/형광펜/암기카드) 목록."""
+    from sqlalchemy import text as sql_text
+    rows = (
+        await session.execute(
+            sql_text("SELECT id, kind, payload, created_at FROM clause_study "
+                     "WHERE doc_id = :id ORDER BY created_at DESC").bindparams(id=doc_id)
+        )
+    ).mappings().all()
+    return StudyListResponse(
+        doc_id=doc_id,
+        items=[StudyItem(id=r["id"], kind=r["kind"], payload=_parse_payload(r["payload"]),
+                         created_at=r["created_at"]) for r in rows],
+    )
+
+
+@router.post("/{doc_id}/study", response_model=StudyItem, status_code=201)
+async def add_study(
+    doc_id: int,
+    body: StudyCreate,
+    user: Annotated[User, Depends(get_current_user)],
+    session: Annotated[AsyncSession, Depends(get_session)],
+):
+    """노트/형광펜/암기카드 1건 추가."""
+    import json
+    from sqlalchemy import text as sql_text
+    if body.kind not in ("note", "highlight", "card"):
+        raise HTTPException(status_code=400, detail="kind 는 note/highlight/card")
+    row = (
+        await session.execute(
+            sql_text("INSERT INTO clause_study(doc_id, kind, payload) "
+                     "VALUES (:d, :k, cast(:p AS jsonb)) RETURNING id, kind, payload, created_at")
+            .bindparams(d=doc_id, k=body.kind, p=json.dumps(body.payload, ensure_ascii=False))
+        )
+    ).mappings().first()
+    await session.commit()
+    return StudyItem(id=row["id"], kind=row["kind"], payload=_parse_payload(row["payload"]),
+                     created_at=row["created_at"])
+
+
+@router.delete("/{doc_id}/study/{study_id}", status_code=204)
+async def delete_study(
+    doc_id: int,
+    study_id: int,
+    user: Annotated[User, Depends(get_current_user)],
+    session: Annotated[AsyncSession, Depends(get_session)],
+):
+    from sqlalchemy import text as sql_text
+    await session.execute(
+        sql_text("DELETE FROM clause_study WHERE id = :s AND doc_id = :d")
+        .bindparams(s=study_id, d=doc_id)
+    )
+    await session.commit()

app/api/ingest_study.py

@@ -23,6 +23,7 @@ from datetime import datetime, timezone
 from fastapi import APIRouter, Depends, Header, HTTPException
 from pydantic import BaseModel
 from sqlalchemy import select
+from sqlalchemy.exc import IntegrityError
 from sqlalchemy.ext.asyncio import AsyncSession
 
 from core.config import settings
@@ -66,6 +67,22 @@ class IngestBody(BaseModel):
     attempts: list[IngestAttempt]
 
 
+def _already_ingested(rows) -> dict:
+    """이미 적재된 세션들의 캐시 요약(멱등 응답). 최초 멱등체크 + 동시경합 흡수 양쪽에서 사용."""
+    return {
+        "status": "already_ingested",
+        "sessions": [
+            {
+                "topic_id": s.study_topic_id,
+                "correct": s.correct_count,
+                "wrong": s.wrong_count,
+                "unsure": s.unsure_count,
+            }
+            for s in rows
+        ],
+    }
+
+
 def _parse_answered_at(s: str | None, now: datetime) -> datetime:
     if not s:
         return now
@@ -98,18 +115,7 @@ async def ingest_attempts(
         )
     ).scalars().all()
     if existing:
-        return {
-            "status": "already_ingested",
-            "sessions": [
-                {
-                    "topic_id": s.study_topic_id,
-                    "correct": s.correct_count,
-                    "wrong": s.wrong_count,
-                    "unsure": s.unsure_count,
-                }
-                for s in existing
-            ],
-        }
+        return _already_ingested(existing)
 
     # pub_id → source_id(내부 질문 id) 해소. deleted tombstone 제외.
     pub_ids = list({a.question_pub_id for a in body.attempts})
@@ -156,73 +162,92 @@ async def ingest_attempts(
     if not by_topic:
         raise HTTPException(status_code=404, detail="해소된 attempt 없음")
 
-    summaries = []
-    for topic_id, items in by_topic.items():
-        qids = [q.id for (_, q) in items]
-        qs = StudyQuizSession(
-            user_id=user_id,
-            study_topic_id=topic_id,
-            question_ids=qids,
-            subject_distribution={},
-            status="done",
-            cursor=len(qids),
-            source="viewer",
-            client_session_uuid=body.client_session_uuid,
-            finished_at=now,
-            created_at=now,
-            updated_at=now,
-        )
-        session.add(qs)
-        await session.flush()  # qs.id
+    try:
+        summaries = []
+        for topic_id, items in by_topic.items():
+            qids = [q.id for (_, q) in items]
+            qs = StudyQuizSession(
+                user_id=user_id,
+                study_topic_id=topic_id,
+                question_ids=qids,
+                subject_distribution={},
+                status="done",
+                cursor=len(qids),
+                source="viewer",
+                client_session_uuid=body.client_session_uuid,
+                finished_at=now,
+                created_at=now,
+                updated_at=now,
+            )
+            session.add(qs)
+            await session.flush()  # qs.id
 
-        c = w = u = 0
-        for a, q in items:
-            try:
-                sel, is_corr, outcome = derive_outcome(a.selected_choice, a.is_unsure, q.correct_choice)
-            except ValueError:
-                skipped.append(a.question_pub_id)  # 선택 없고 unsure 아님 = 무효 → skip
-                continue
-            if outcome == "correct":
-                c += 1
-            elif outcome == "wrong":
-                w += 1
-            elif outcome == "unsure":
-                u += 1
-            session.add(
-                StudyQuestionAttempt(
-                    user_id=user_id,
-                    study_question_id=q.id,
-                    study_topic_id=topic_id,
-                    selected_choice=sel,
-                    correct_choice=q.correct_choice,
-                    is_correct=is_corr,
-                    outcome=outcome,
-                    quiz_session_id=qs.id,
-                    answered_at=_parse_answered_at(a.answered_at, now),
+            c = w = u = 0
+            for a, q in items:
+                try:
+                    sel, is_corr, outcome = derive_outcome(a.selected_choice, a.is_unsure, q.correct_choice)
+                except ValueError:
+                    skipped.append(a.question_pub_id)  # 선택 없고 unsure 아님 = 무효 → skip
+                    continue
+                if outcome == "correct":
+                    c += 1
+                elif outcome == "wrong":
+                    w += 1
+                elif outcome == "unsure":
+                    u += 1
+                session.add(
+                    StudyQuestionAttempt(
+                        user_id=user_id,
+                        study_question_id=q.id,
+                        study_topic_id=topic_id,
+                        selected_choice=sel,
+                        correct_choice=q.correct_choice,
+                        is_correct=is_corr,
+                        outcome=outcome,
+                        quiz_session_id=qs.id,
+                        answered_at=_parse_answered_at(a.answered_at, now),
+                    )
+                )
+            qs.correct_count, qs.wrong_count, qs.unsure_count = c, w, u
+            await session.flush()
+
+            # finalize 무수정 재생(progress/SR/pattern + 4-A/4-B enqueue). 그 후 멱등 마커.
+            summary = await finalize_session(
+                session, user_id=user_id, study_topic_id=topic_id, quiz_session_id=qs.id
+            )
+            qs.finalized_at = now
+            summaries.append(
+                {
+                    "topic_id": topic_id,
+                    "quiz_session_id": qs.id,
+                    "correct": summary.correct,
+                    "wrong": summary.wrong,
+                    "unsure": summary.unsure,
+                    "newly_correct": summary.newly_correct,
+                    "relapsed": summary.relapsed,
+                    "recovered": summary.recovered,
+                }
+            )
+
+        await session.commit()
+    except IntegrityError:
+        # 동시 같은 client_session_uuid 경합 — 상대가 먼저 commit → (client_session_uuid,
+        # study_topic_id) uq(mig376) 위반. 데이터는 안전(원자 1-tx 전체 롤백 → SR 이중 advance
+        # 없음). 승자 결과로 graceful 수렴(500 대신 already_ingested). uuid 경합이 아닌 진짜
+        # 무결성 오류면 재조회가 비어 → re-raise 로 표면화.
+        await session.rollback()
+        winner = (
+            await session.execute(
+                select(StudyQuizSession).where(
+                    StudyQuizSession.client_session_uuid == body.client_session_uuid
                 )
             )
-        qs.correct_count, qs.wrong_count, qs.unsure_count = c, w, u
-        await session.flush()
+        ).scalars().all()
+        if not winner:
+            raise
+        logger.info("study_ingest uuid=%s 동시경합 흡수 → already_ingested", body.client_session_uuid)
+        return _already_ingested(winner)
 
-        # finalize 무수정 재생(progress/SR/pattern + 4-A/4-B enqueue). 그 후 멱등 마커.
-        summary = await finalize_session(
-            session, user_id=user_id, study_topic_id=topic_id, quiz_session_id=qs.id
-        )
-        qs.finalized_at = now
-        summaries.append(
-            {
-                "topic_id": topic_id,
-                "quiz_session_id": qs.id,
-                "correct": summary.correct,
-                "wrong": summary.wrong,
-                "unsure": summary.unsure,
-                "newly_correct": summary.newly_correct,
-                "relapsed": summary.relapsed,
-                "recovered": summary.recovered,
-            }
-        )
-
-    await session.commit()
     logger.info(
         "study_ingest uuid=%s user=%s sessions=%s skipped=%s",
         body.client_session_uuid, user_id, len(summaries), len(skipped),

app/api/search.py

@@ -15,7 +15,7 @@ from fastapi.responses import JSONResponse
 from pydantic import BaseModel
 from sqlalchemy.ext.asyncio import AsyncSession
 
-from core.auth import get_current_user
+from core.auth import get_current_user, get_egress_class
 from core.database import get_session
 from core.utils import setup_logger
 from models.user import User
@@ -139,6 +139,7 @@ def _build_search_debug(pr: PipelineResult) -> SearchDebug:
 async def search(
     q: str,
     user: Annotated[User, Depends(get_current_user)],
+    egress_class: Annotated[str, Depends(get_egress_class)],
     session: Annotated[AsyncSession, Depends(get_session)],
     background_tasks: BackgroundTasks,
     mode: str = Query("hybrid", pattern="^(fts|trgm|vector|hybrid)$"),
@@ -211,6 +212,8 @@ async def search(
         None, description="안전 자료실 C-1: 관할 필터 (KR/US/EU/JP/GB/INT)"),
     year_from: int | None = Query(None, ge=1900, le=2100, description="published_date 연도 하한 (NULL=created_at fallback)"),
     year_to: int | None = Query(None, ge=1900, le=2100, description="published_date 연도 상한"),
+    domain_bucket: str | None = Query(None, description="377: domain_bucket 스코프 CSV (Safety,Engineering,Law,Philosophy,Programming,General,News). domain_bucket = ANY"),
+    exclude_bucket: str | None = Query(None, description="377: domain_bucket 제외 CSV (예: News). 지식질의 시 News 기본제외용"),
     facets: bool = Query(False, description="안전 자료실 C-1 후속: top-K 결과 분류 축 분포(material_type/jurisdiction/version_status)를 응답 facets 에 집계. 미지정=계산/노출 0"),
 ):
     """문서 검색 — FTS + ILIKE + 벡터 결합 (Phase 3.1 이후 run_search wrapper)"""
@@ -221,6 +224,9 @@ async def search(
             jurisdiction=jurisdiction,
             year_from=year_from,
             year_to=year_to,
+            domain_buckets=[b.strip() for b in domain_bucket.split(",") if b.strip()] if domain_bucket else None,
+            exclude_buckets=[b.strip() for b in exclude_bucket.split(",") if b.strip()] if exclude_bucket else None,
+            cloud_egress=(egress_class == "cloud"),
         )
         pr = await run_search(
             session,

app/api/study_concepts.py

@@ -0,0 +1,54 @@
+"""study_concepts API — 이론공부 홈(오늘의 개념 · 진도 · 회독 SR). prefix = /api/study.
+
+문제풀이 표면 무접촉. 개념문서(가스기사 태그) 읽기 집계 + 회독 SR write 만. 단일 토픽(가스기사=4).
+경로: GET /curriculum · GET /today-concepts · POST /concepts/{doc_id}/read.
+"""
+
+from __future__ import annotations
+
+from typing import Annotated
+
+from fastapi import APIRouter, Depends
+from sqlalchemy.ext.asyncio import AsyncSession
+
+from core.auth import get_current_user
+from core.database import get_session
+from models.user import User
+from services.study import concept_curriculum as cc
+
+router = APIRouter()
+
+# 가스기사 단일 토픽 운영(현행). 다토픽 확장 시 쿼리 파라미터로 승격.
+DEFAULT_TOPIC_ID = 4
+
+
+@router.get("/curriculum")
+async def get_curriculum(
+    user: Annotated[User, Depends(get_current_user)],
+    session: Annotated[AsyncSession, Depends(get_session)],
+    topic_id: int = DEFAULT_TOPIC_ID,
+):
+    """과목별 회독 진도 + 개념/문항 복습 due 요약."""
+    return await cc.curriculum(session, user.id, topic_id)
+
+
+@router.get("/today-concepts")
+async def get_today_concepts(
+    user: Annotated[User, Depends(get_current_user)],
+    session: Annotated[AsyncSession, Depends(get_session)],
+    topic_id: int = DEFAULT_TOPIC_ID,
+    limit: int = 6,
+):
+    """오늘 공부할 개념(재복습 → 미독 빈출순)."""
+    return await cc.today_concepts(session, user.id, topic_id, limit)
+
+
+@router.post("/concepts/{doc_id}/read")
+async def post_concept_read(
+    doc_id: int,
+    user: Annotated[User, Depends(get_current_user)],
+    session: Annotated[AsyncSession, Depends(get_session)],
+    topic_id: int = DEFAULT_TOPIC_ID,
+):
+    """개념 회독 처리 → 회독 플래그 + SR 입고/전진."""
+    return await cc.mark_read(session, user.id, topic_id, doc_id)

app/core/auth.py

@@ -31,11 +31,11 @@ def hash_password(password: str) -> str:
     return bcrypt.hashpw(password.encode(), bcrypt.gensalt()).decode()
 
 
-def create_access_token(subject: str, expires_minutes: int | None = None) -> str:
+def create_access_token(subject: str, expires_minutes: int | None = None, egress: str = "local") -> str:
     minutes = expires_minutes if expires_minutes is not None else ACCESS_TOKEN_EXPIRE_MINUTES
     now = datetime.now(timezone.utc)
     expire = now + timedelta(minutes=minutes)
-    payload = {"sub": subject, "exp": expire, "iat": int(now.timestamp()), "type": "access"}
+    payload = {"sub": subject, "exp": expire, "iat": int(now.timestamp()), "type": "access", "egress": egress}
     return jwt.encode(payload, settings.jwt_secret, algorithm=ALGORITHM)
 
 
@@ -100,6 +100,15 @@ def verify_totp(code: str, secret: str | None = None) -> bool:
     return totp.verify(code)
 
 
+async def get_egress_class(
+    credentials: Annotated[HTTPAuthorizationCredentials, Depends(security)],
+) -> str:
+    """토큰 egress claim -> 'cloud'|'local' (갭2 cloud-egress allowlist). claim 부재=local
+    (비파괴; 기존 토큰=신뢰/로컬). 쿼리파라미터 아님 -> 호출자가 끌 수 없음(우회 차단)."""
+    payload = decode_token(credentials.credentials)
+    return (payload or {}).get("egress", "local")
+
+
 async def get_current_user(
     credentials: Annotated[HTTPAuthorizationCredentials, Depends(security)],
     session: Annotated[AsyncSession, Depends(get_session)],

app/main.py

@@ -33,6 +33,7 @@ from api.study_sessions import router as study_sessions_router
 from api.study_topics import router as study_topics_router
 from api.study_reminders import router as study_reminders_router
 from api.study_cards import router as study_cards_router
+from api.study_concepts import router as study_concepts_router
 from api.video import router as video_router
 from core.config import settings
 from core.database import async_session, engine, init_db
@@ -249,6 +250,8 @@ app.include_router(study_reminders_router, prefix="/api/study-reminders", tags=[
 app.include_router(study_cards_router, prefix="/api/study-cards", tags=["study-cards"])
 # Phase 1: 학습 진행 상태 (review-complete + review-queue). prefix=/api/study-topics 안에 정의됨.
 app.include_router(study_question_progress_router, prefix="/api", tags=["study-progress"])
+# 이론공부 홈: 오늘의 개념·진도·회독 SR (개념문서 소비 표면, 문제풀이 무접촉).
+app.include_router(study_concepts_router, prefix="/api/study", tags=["study-theory"])
 
 # TODO: Phase 5에서 추가
 # app.include_router(tasks.router, prefix="/api/tasks", tags=["tasks"])

app/models/published.py

@@ -56,5 +56,9 @@ class PublishOutbox(Base):
         DateTime(timezone=True), default=datetime.now, nullable=False
     )
     processed_at: Mapped[datetime | None] = mapped_column(DateTime(timezone=True))
+    # mig378: 행별 격리 재시도/terminal. attempts=savepoint 실패 누적, failed_at=MAX 초과 terminal
+    # (set 시 워커 select 에서 제외 → head-of-line block 방지).
+    attempts: Mapped[int] = mapped_column(SmallInteger, nullable=False, default=0)
+    failed_at: Mapped[datetime | None] = mapped_column(DateTime(timezone=True))
 
     # 미처리 부분 인덱스 idx(id) WHERE processed_at IS NULL = mig372.

app/models/study_concept_progress.py

@@ -0,0 +1,46 @@
+"""study_concept_progress — 사용자 × 개념문서 단위 간격반복(SR) 진행 (이론공부 홈).
+
+문제 SR(study_question_progress)의 개념(이론)판. '개념문서' = documents 한 건(가스기사 태그).
+회독(첫 read) → 복습 큐 진입, 이후 회독마다 sr_schedule 산술(1·3·7·14·졸업) 공용 전진.
+concept_doc_id 는 documents.id 를 가리키나 FK 미설정 — hot 테이블(documents) 락 회피(clause_study 선례).
+"""
+
+from __future__ import annotations
+
+from datetime import datetime
+
+from sqlalchemy import BigInteger, DateTime, ForeignKey, SmallInteger, UniqueConstraint
+from sqlalchemy.orm import Mapped, mapped_column
+
+from core.database import Base
+
+
+class StudyConceptProgress(Base):
+    __tablename__ = "study_concept_progress"
+    __table_args__ = (
+        UniqueConstraint(
+            "user_id", "concept_doc_id", name="uq_concept_progress_user_doc"
+        ),
+    )
+
+    id: Mapped[int] = mapped_column(BigInteger, primary_key=True)
+    user_id: Mapped[int] = mapped_column(
+        BigInteger, ForeignKey("users.id", ondelete="CASCADE"), nullable=False
+    )
+    study_topic_id: Mapped[int] = mapped_column(
+        BigInteger, ForeignKey("study_topics.id", ondelete="CASCADE"), nullable=False
+    )
+    # documents.id 참조 — FK 없음(락 회피). 개념문서 삭제 시 고아 행은 read 집계에서 자연 제외.
+    concept_doc_id: Mapped[int] = mapped_column(BigInteger, nullable=False)
+
+    # 복습 큐 (sr_schedule 공용): stage 0~3 = 1·3·7·14일, 4 = 졸업(due_at NULL)
+    review_stage: Mapped[int | None] = mapped_column(SmallInteger)
+    due_at: Mapped[datetime | None] = mapped_column(DateTime(timezone=True))
+    last_read_at: Mapped[datetime | None] = mapped_column(DateTime(timezone=True))
+
+    created_at: Mapped[datetime] = mapped_column(
+        DateTime(timezone=True), default=datetime.now, nullable=False
+    )
+    updated_at: Mapped[datetime] = mapped_column(
+        DateTime(timezone=True), default=datetime.now, onupdate=datetime.now, nullable=False
+    )

app/services/search/retrieval_service.py

@@ -76,10 +76,15 @@ class AxisFilter:
     jurisdiction: str | None = None
     year_from: int | None = None
     year_to: int | None = None
+    domain_buckets: list[str] | None = None    # 377: domain_bucket = ANY (도메인 스코프)
+    exclude_buckets: list[str] | None = None    # 377: domain_bucket <> ALL (예: News 제외)
+    cloud_egress: bool = False    # 갭2: 클라우드 소비자 cloud-eligibility allowlist 강제(토큰 claim 유래)
 
     def active(self) -> bool:
         return bool(self.material_types or self.jurisdiction
-                    or self.year_from is not None or self.year_to is not None)
+                    or self.year_from is not None or self.year_to is not None
+                    or self.domain_buckets or self.exclude_buckets
+                    or self.cloud_egress)
 
 
 def _axis_sql(alias: str, af: "AxisFilter | None", params: dict) -> str:
@@ -104,6 +109,22 @@ def _axis_sql(alias: str, af: "AxisFilter | None", params: dict) -> str:
     if af.year_to is not None:
         cl.append(f"COALESCE({p}published_date, {p}created_at::date) <= make_date(:af_yt, 12, 31)")
         params["af_yt"] = af.year_to
+    if af.domain_buckets:
+        cl.append(f"{p}domain_bucket = ANY(:af_db)")
+        params["af_db"] = af.domain_buckets
+    if af.exclude_buckets:
+        cl.append(f"{p}domain_bucket <> ALL(:af_xdb)")
+        params["af_xdb"] = af.exclude_buckets
+    if af.cloud_egress:
+        # 갭2 클라우드 egress allowlist(default-deny). restricted 는 _license_sql 별도 차단.
+        cl.append(
+            f"({p}data_origin = 'external' OR ("
+            f"{p}data_origin = 'work' "
+            f"AND {p}domain_bucket IN ('Engineering','Safety','Law') "
+            f"AND ({p}source_channel IS NULL OR {p}source_channel::text NOT IN ('voice','chat','memo')) "
+            f"AND {p}category::text IS DISTINCT FROM 'memo' "
+            f"AND ({p}user_note IS NULL OR {p}user_note = '')))"
+        )
     return " AND " + " AND ".join(cl)
 
 
@@ -121,7 +142,21 @@ def _license_sql(alias: str) -> str:
     술어 정의 = license_filter.restricted_exclude_sql 공유(digest/briefing/study 풀이와 단일 source).
     """
     from services.search.license_filter import restricted_exclude_sql
-    return " AND " + restricted_exclude_sql(alias)
+    _p = (alias + ".") if alias else ""
+    # ASME clause-KB(379): clause docs (doc_kind='clause') = read/nav/backlink only, excluded from retrieval/digest legs.
+    return " AND " + restricted_exclude_sql(alias) + f" AND {_p}doc_kind = 'standard'"
+
+
+def cloud_eligible_doc_sql(alias: str = "") -> str:
+    """단일 문서가 cloud 소비자(예: Claude/MCP)에게 노출 가능한가 = search retrieval 과
+    동일한 egress allowlist(갭2) + license 제한(B-4) 결합 술어. fetch_document(cloud) 가
+    search 와 byte-동일 게이트를 공유하도록 단일 source([[feedback_structural_integrity_over_path_discipline]]).
+
+    cloud_egress·license leg 모두 bind 파라미터 없는 리터럴 술어라 호출측 추가 params 불요.
+    주의: _license_sql 은 소유자 단건 다운로드엔 미적용(a안)이지만, cloud 노출은 구매 전자책
+    verbatim 누출을 막아야 하므로 여기선 항상 적용 = search 와 동일(local 토큰은 이 게이트 미발동).
+    반환 ' AND (egress allowlist) AND (license)' (alias='' = 컬럼 직접 참조). default-deny."""
+    return _axis_sql(alias, AxisFilter(cloud_egress=True), {}) + _license_sql(alias)
 
 
 # 2단계 gate (R2-B1) — SQL string interpolation 직전 final allowlist.

app/services/study/concept_curriculum.py

@@ -0,0 +1,207 @@
+"""concept_curriculum — 이론공부 홈 재료 (오늘의 개념 · 진도 · 회독 SR).
+
+개념문서 = documents (user_tags = @library/{topic}/{과목}/... , 가스기사). is_read = 회독,
+md_content 의 ★ 개수 = 빈출 tier(★★★=3 / ★★=2 / else 1). 회독 SR = study_concept_progress
++ sr_schedule(문제 SR 공용 산술). 읽기 전용 집계 + mark_read(회독+SR 입고)만 write. LLM 0.
+
+문제풀이 표면 무접촉 — 여기서 읽는 study_question_progress 는 '문항 due 카운트'만(홈 표시용).
+"""
+
+from __future__ import annotations
+
+from datetime import datetime, timezone
+
+from sqlalchemy import func, or_, select, text
+from sqlalchemy.ext.asyncio import AsyncSession
+
+from models.document_read import DocumentRead
+from models.study_concept_progress import StudyConceptProgress
+from models.study_question_progress import StudyQuestionProgress
+from models.study_topic import StudyTopic
+from services.study.sr_schedule import advance, first_due
+
+# 개념 행 조회 — 태그로 개념문서 필터 + 회독 진행 LEFT JOIN. md_content 는 전송 안 하고
+# ★ 유무만 서버측 boolean 으로(홈이 자주 호출돼도 페이로드 최소).
+# is_read = document_reads(회독 정본, is_read 컬럼 아님) EXISTS. library unread 와 동일 기준.
+_CONCEPT_ROWS_SQL = text(
+    """
+    SELECT d.id AS doc_id,
+           d.title AS title,
+           EXISTS (
+             SELECT 1 FROM document_reads r
+             WHERE r.document_id = d.id AND r.user_id = :uid
+           ) AS is_read,
+           (d.md_content LIKE '%★★★%') AS f3,
+           (d.md_content LIKE '%★★%')  AS f2,
+           split_part(replace(d.user_tags::text, '"', ''), '/', 3) AS subject,
+           p.review_stage AS review_stage,
+           p.due_at AS due_at,
+           p.last_read_at AS last_read_at
+    FROM documents d
+    LEFT JOIN study_concept_progress p
+      ON p.concept_doc_id = d.id AND p.user_id = :uid
+    WHERE d.user_tags::text LIKE :like
+      AND d.deleted_at IS NULL
+    """
+)
+
+
+async def _topic_name(session: AsyncSession, topic_id: int) -> str | None:
+    return (
+        await session.execute(select(StudyTopic.name).where(StudyTopic.id == topic_id))
+    ).scalar_one_or_none()
+
+
+async def _concept_rows(session: AsyncSession, user_id: int, topic_name: str):
+    like = f"%@library/{topic_name}/%"
+    return (
+        await session.execute(_CONCEPT_ROWS_SQL, {"uid": user_id, "like": like})
+    ).mappings().all()
+
+
+def _freq(row) -> int:
+    if row["f3"]:
+        return 3
+    if row["f2"]:
+        return 2
+    return 1
+
+
+def _is_due(row, now: datetime) -> bool:
+    return (
+        row["due_at"] is not None
+        and row["due_at"] <= now
+        and (row["review_stage"] or 0) < 4
+    )
+
+
+def _item(row) -> dict:
+    return {
+        "doc_id": row["doc_id"],
+        "title": row["title"],
+        "subject": row["subject"],
+        "freq": _freq(row),
+        "review_stage": row["review_stage"],
+        "due_at": row["due_at"],
+    }
+
+
+async def _question_due_count(session: AsyncSession, user_id: int, topic_id: int, now: datetime) -> int:
+    """문항 복습 due (기존 study_question_progress 엔진 재사용, 홈 표시용)."""
+    return (
+        await session.execute(
+            select(func.count())
+            .select_from(StudyQuestionProgress)
+            .where(
+                StudyQuestionProgress.user_id == user_id,
+                StudyQuestionProgress.study_topic_id == topic_id,
+                StudyQuestionProgress.due_at.is_not(None),
+                StudyQuestionProgress.due_at <= now,
+                or_(
+                    StudyQuestionProgress.review_stage.is_(None),
+                    StudyQuestionProgress.review_stage < 4,
+                ),
+            )
+        )
+    ).scalar_one()
+
+
+async def curriculum(session: AsyncSession, user_id: int, topic_id: int) -> dict:
+    """과목별 회독 진도 + 개념/문항 복습 due 요약 (진도 대시보드)."""
+    name = await _topic_name(session, topic_id)
+    rows = await _concept_rows(session, user_id, name) if name else []
+    now = datetime.now(timezone.utc)
+
+    subj: dict[str, dict] = {}
+    for r in rows:
+        s = subj.setdefault(r["subject"], {"subject": r["subject"], "total": 0, "read": 0})
+        s["total"] += 1
+        if r["is_read"]:
+            s["read"] += 1
+
+    total = len(rows)
+    read = sum(1 for r in rows if r["is_read"])
+    concept_due = sum(1 for r in rows if _is_due(r, now))
+    question_due = await _question_due_count(session, user_id, topic_id, now)
+
+    return {
+        "topic_id": topic_id,
+        "topic_name": name,
+        "subjects": sorted(subj.values(), key=lambda x: x["subject"]),
+        "total": total,
+        "read": read,
+        "concept_due": concept_due,
+        "question_due": question_due,
+    }
+
+
+async def today_concepts(
+    session: AsyncSession, user_id: int, topic_id: int, limit: int = 6
+) -> dict:
+    """오늘 공부할 개념 = 재복습(SR due) 먼저 → 미독(빈출 우선). 졸업/재복습대기 제외."""
+    name = await _topic_name(session, topic_id)
+    rows = await _concept_rows(session, user_id, name) if name else []
+    now = datetime.now(timezone.utc)
+
+    due = [r for r in rows if _is_due(r, now)]
+    due.sort(key=lambda r: r["due_at"])
+
+    # 미독 & 아직 SR 큐 진입 전(due_at NULL) → 빈출 높은 순
+    unread = [r for r in rows if not r["is_read"] and r["due_at"] is None]
+    unread.sort(key=lambda r: (-_freq(r), r["subject"], r["title"]))
+
+    picked = [{**_item(r), "reason": "재복습"} for r in due]
+    picked += [{**_item(r), "reason": "신규"} for r in unread]
+
+    return {
+        "concepts": picked[:limit],
+        "due_total": len(due),
+        "unread_total": len(unread),
+    }
+
+
+async def mark_read(
+    session: AsyncSession, user_id: int, topic_id: int, doc_id: int, now: datetime | None = None
+) -> dict:
+    """개념 회독 처리 = document_reads(+1) + 회독 SR 입고/전진.
+
+    회독 정본 = document_reads(append-only), documents.is_read 컬럼 아님(library unread 와 정합).
+    첫 회독 → first_due(stage 0, 내일). 이후 회독은 'due 도래(due_at<=now)' 때만 correct 로 전진
+    (이른 재열람/다중클릭 과전진 방지). stage 4 졸업 후엔 due_at NULL 이라 전진 없음.
+    """
+    now = now or datetime.now(timezone.utc)
+
+    # 회독 로그 append (+1) — 사용자 명시 회독. 자동 아님(엔드포인트 = 명시 POST).
+    session.add(DocumentRead(user_id=user_id, document_id=doc_id, read_at=now))
+
+    prog = (
+        await session.execute(
+            select(StudyConceptProgress).where(
+                StudyConceptProgress.user_id == user_id,
+                StudyConceptProgress.concept_doc_id == doc_id,
+            )
+        )
+    ).scalar_one_or_none()
+
+    if prog is None:
+        stage, due = first_due(now)
+        prog = StudyConceptProgress(
+            user_id=user_id,
+            study_topic_id=topic_id,
+            concept_doc_id=doc_id,
+            review_stage=stage,
+            due_at=due,
+            last_read_at=now,
+        )
+        session.add(prog)
+    else:
+        # due 도래 시에만 전진 — 미래 due(재열람 이른 클릭)는 stage 불변, last_read_at 만 갱신.
+        if prog.due_at is not None and prog.due_at <= now:
+            res = advance(prog.review_stage, "correct", now)
+            if res is not None:
+                prog.review_stage, prog.due_at = res
+        prog.last_read_at = now
+
+    await session.commit()
+    await session.refresh(prog)
+    return {"ok": True, "review_stage": prog.review_stage, "due_at": prog.due_at}

app/services/study/publish_enqueue.py

@@ -68,10 +68,10 @@ async def enqueue_question_publish(session: AsyncSession, q: Any) -> None:
         await enqueue_publish(session, kind=KIND_EXPLANATION, source_id=q.id, payload=expl)
 
 
-async def backfill_publish_questions(session: AsyncSession, *, after_id: int = 0, limit: int = 200) -> int:
+async def backfill_publish_questions(session: AsyncSession, *, after_id: int = 0, limit: int = 200) -> tuple[int, int]:
     """active(미삭제) 문항을 id>after_id 부터 bounded 로 outbox 적재.
 
-    반환 = enqueue 한 문항 수(0 이면 끝). 큰 셋은 마지막 id 로 페이지 반복. caller commit.
+    반환 = (enqueue 수, 마지막 처리 id). caller 는 수==limit 면 last_id 로 다음 페이지. caller commit.
     """
     rows = (
         await session.execute(
@@ -83,7 +83,7 @@ async def backfill_publish_questions(session: AsyncSession, *, after_id: int = 0
     ).scalars().all()
     for q in rows:
         await enqueue_question_publish(session, q)
-    return len(rows)
+    return len(rows), (rows[-1].id if rows else after_id)
 
 
 async def enqueue_topic_publish(session: AsyncSession, topic: Any) -> None:
@@ -91,10 +91,10 @@ async def enqueue_topic_publish(session: AsyncSession, topic: Any) -> None:
     await enqueue_publish(session, kind=KIND_TOPIC, source_id=topic.id, payload=project_topic(topic))
 
 
-async def backfill_publish_topics(session: AsyncSession, *, after_id: int = 0, limit: int = 200) -> int:
+async def backfill_publish_topics(session: AsyncSession, *, after_id: int = 0, limit: int = 200) -> tuple[int, int]:
     """active(미삭제) 주제를 id>after_id 부터 bounded 로 outbox 적재(S-1 초기 백필).
 
-    반환 = enqueue 한 주제 수(0 이면 끝). 큰 셋은 마지막 id 로 페이지 반복. caller commit.
+    반환 = (enqueue 수, 마지막 처리 id). caller 는 수==limit 면 last_id 로 다음 페이지. caller commit.
     멱등 = 발행 워커의 (payload_hash, deleted) 디둡이 no-op 재투영 흡수(중복 enqueue 무해).
     """
     rows = (
@@ -107,7 +107,7 @@ async def backfill_publish_topics(session: AsyncSession, *, after_id: int = 0, l
     ).scalars().all()
     for t in rows:
         await enqueue_topic_publish(session, t)
-    return len(rows)
+    return len(rows), (rows[-1].id if rows else after_id)
 
 
 async def enqueue_card_publish(session: AsyncSession, card: Any) -> None:
@@ -123,10 +123,10 @@ async def enqueue_card_publish(session: AsyncSession, card: Any) -> None:
         await enqueue_publish(session, kind=KIND_CARD, source_id=card.id, payload=project_card(card))
 
 
-async def backfill_publish_cards(session: AsyncSession, *, after_id: int = 0, limit: int = 200) -> int:
+async def backfill_publish_cards(session: AsyncSession, *, after_id: int = 0, limit: int = 200) -> tuple[int, int]:
     """검수완료(needs_review=False)·미삭제 카드를 id>after_id 부터 bounded 로 outbox 적재(S-2 초기 백필).
 
-    반환 = enqueue 한 카드 수(0 이면 끝). 멱등 = 워커 (payload_hash, deleted) 디둡. caller commit.
+    반환 = (enqueue 수, 마지막 처리 id). caller 는 수==limit 면 last_id 로 다음 페이지. 멱등 = 워커 디둡. caller commit.
     """
     rows = (
         await session.execute(
@@ -142,7 +142,7 @@ async def backfill_publish_cards(session: AsyncSession, *, after_id: int = 0, li
     ).scalars().all()
     for c in rows:
         await enqueue_card_publish(session, c)
-    return len(rows)
+    return len(rows), (rows[-1].id if rows else after_id)
 
 
 async def enqueue_card_progress_publish(session: AsyncSession, progress: Any) -> None:
@@ -155,11 +155,11 @@ async def enqueue_card_progress_publish(session: AsyncSession, progress: Any) ->
     )
 
 
-async def backfill_publish_card_progress(session: AsyncSession, *, after_id: int = 0, limit: int = 200) -> int:
+async def backfill_publish_card_progress(session: AsyncSession, *, after_id: int = 0, limit: int = 200) -> tuple[int, int]:
     """모든 card progress row 를 id>after_id 부터 bounded 로 outbox 적재(S-4 초기 백필).
 
     ★필터 없음 = ALL row(due_at NULL sentinel·terminal 포함) — due-only 백필은 sentinel 누락.
-    반환 = enqueue 한 row 수(0 이면 끝). 멱등 = 워커 디둡. caller commit.
+    반환 = (enqueue 수, 마지막 처리 id). caller 는 수==limit 면 last_id 로 다음 페이지. 멱등 = 워커 디둡. caller commit.
     """
     rows = (
         await session.execute(
@@ -171,4 +171,4 @@ async def backfill_publish_card_progress(session: AsyncSession, *, after_id: int
     ).scalars().all()
     for p in rows:
         await enqueue_card_progress_publish(session, p)
-    return len(rows)
+    return len(rows), (rows[-1].id if rows else after_id)

app/workers/classify_worker.py

@@ -608,7 +608,9 @@ async def process(
             except Exception as exc:
                 if legacy_cfg is not None and is_deferrable_error(exc):
                     raise StageDeferred(f"macbook_unavailable:{type(exc).__name__}") from exc
-                logger.warning(f"[summary-fallback] id={document_id}: {exc}")
+                # ai_summary=NULL 로 완료되면 digest/briefing 이 조용히 제외 → ERROR 로 가시화
+                # (best-effort 강등 자체는 유지, 운영 추적성만 보강).
+                logger.error(f"[summary-fallback] id={document_id} ai_summary 미생성: {exc}")
 
     finally:
         await client.close()

app/workers/csb_collector.py

@@ -140,7 +140,8 @@ async def _download_pdf(url: str, dest: Path) -> int:
     if len(resp.content) > _MAX_PDF_BYTES:
         raise FeedError(f"PDF 크기 초과 ({len(resp.content)} bytes): {url}")
     dest.parent.mkdir(parents=True, exist_ok=True)
-    dest.write_bytes(resp.content)
+    # 최대 50MB PDF write 는 동기 blocking — 이벤트루프 점유 회피 to_thread (R5 동형).
+    await asyncio.to_thread(dest.write_bytes, resp.content)
     return len(resp.content)
 
 
@@ -190,9 +191,11 @@ async def _ingest_pdf(session, page_slug: str, pdf_url: str) -> bool:
 
     dest = Path(settings.nas_mount_path) / rel_path
     size = await _download_pdf(pdf_url, dest)
+    # 50MB PDF read + sha256 는 동기 blocking(I/O+CPU) — 이벤트루프 점유 회피 to_thread (R5 동형).
+    file_hash = await asyncio.to_thread(lambda: hashlib.sha256(dest.read_bytes()).hexdigest())
     doc = Document(
         file_path=rel_path,
-        file_hash=hashlib.sha256(dest.read_bytes()).hexdigest(),
+        file_hash=file_hash,
         file_format="pdf",
         file_size=size,
         file_type="immutable",

app/workers/file_watcher.py

@@ -251,104 +251,118 @@ async def watch_inbox():
     for extra_path in settings.additional_watch_targets:
         targets.append((extra_path, "library"))
 
-    async with async_session() as session:
-        # ─── Web/ 트랙 (devonagent) — DEVONthink Smart Rule 이 떨군 .html 만 진입 ───
-        if web_root.exists():
-            # rglob NFS 디렉토리 walk(blocking stat 다발)를 off-thread 로 수집 (R5).
-            for file_path in await asyncio.to_thread(lambda: list(web_root.rglob("*.html"))):
-                if not file_path.is_file() or should_skip(file_path):
-                    continue
-                rel_path = str(file_path.relative_to(nas_root))
-                added, _ = await _ingest_web_file(session, file_path, rel_path)
+    # 파일별 독립 세션+commit 으로 격리 — 한 파일 실패(예: rglob↔stat 사이 삭제로 FileNotFoundError,
+    # flush 오류)가 watch_inbox 전체를 raise·롤백해 그 사이클 등록분을 모두 잃거나, 결정적 poison
+    # 파일이 매 사이클 같은 지점에서 중단시키는 것을 차단 (news_collector/csb_collector 와 동형).
+    # ─── Web/ 트랙 (devonagent) — DEVONthink Smart Rule 이 떨군 .html 만 진입 ───
+    if web_root.exists():
+        # rglob NFS 디렉토리 walk(blocking stat 다발)를 off-thread 로 수집 (R5).
+        for file_path in await asyncio.to_thread(lambda: list(web_root.rglob("*.html"))):
+            if not file_path.is_file() or should_skip(file_path):
+                continue
+            rel_path = str(file_path.relative_to(nas_root))
+            try:
+                async with async_session() as session:
+                    added, _ = await _ingest_web_file(session, file_path, rel_path)
+                    await session.commit()
                 new_count += added
-
-        # ─── PKM 트랙 (기존 drive_sync) ─────────────────────────────────────────
-        for sub, expected_category in targets:
-            scan_root = pkm_root / sub
-            if not scan_root.exists():
+            except Exception as e:
+                logger.warning("[Web] 파일 처리 실패 skip path=%s: %s", rel_path, e)
                 continue
 
-            # 안전 자료실 A-2/B-4 — 타깃 폴더 기반 (material, jurisdiction, license)
-            target_mt, target_jur, target_license = _TARGET_AXIS.get(
-                Path(sub).name, (None, None, None)
+    # ─── PKM 트랙 (기존 drive_sync) ─────────────────────────────────────────
+    for sub, expected_category in targets:
+        scan_root = pkm_root / sub
+        if not scan_root.exists():
+            continue
+
+        # 안전 자료실 A-2/B-4 — 타깃 폴더 기반 (material, jurisdiction, license)
+        target_mt, target_jur, target_license = _TARGET_AXIS.get(
+            Path(sub).name, (None, None, None)
+        )
+
+        # NFS 디렉토리 walk(blocking) off-thread 수집 (R5).
+        for file_path in await asyncio.to_thread(lambda: list(scan_root.rglob("*"))):
+            if not file_path.is_file() or should_skip(file_path):
+                continue
+
+            category, needs_conversion, next_stage = _route_media(
+                file_path, expected_category
             )
 
-            # NFS 디렉토리 walk(blocking) off-thread 수집 (R5).
-            for file_path in await asyncio.to_thread(lambda: list(scan_root.rglob("*"))):
-                if not file_path.is_file() or should_skip(file_path):
-                    continue
+            # audio/video 폴더에 엉뚱한 확장자가 들어왔거나 Inbox 에
+            # audio/video 가 잘못 떨어진 경우 — 이 라운드에서 아예 skip
+            if category is None and next_stage is None:
+                continue
 
-                category, needs_conversion, next_stage = _route_media(
-                    file_path, expected_category
-                )
-
-                # audio/video 폴더에 엉뚱한 확장자가 들어왔거나 Inbox 에
-                # audio/video 가 잘못 떨어진 경우 — 이 라운드에서 아예 skip
-                if category is None and next_stage is None:
-                    continue
-
-                rel_path = str(file_path.relative_to(nas_root))
+            rel_path = str(file_path.relative_to(nas_root))
+            try:
                 # GB 파일 SHA-256 은 이벤트 루프를 점유 → 같은 루프의 모든 1분 주기 consumer
                 # + FastAPI 요청이 수십초~분 동시 정지. to_thread 오프로드. 스캔 루프가 이미
                 # 순차라 file_hash 는 한 번에 하나만 실행(직렬화) — 병렬 해싱 X = NFS 2.5GbE
-                # 대역폭·버퍼 메모리 blowup 방지 (R5).
+                # 대역폭·버퍼 메모리 blowup 방지 (R5). 세션 밖에서 계산(커넥션 미점유).
                 fhash = await asyncio.to_thread(file_hash, file_path)
 
-                result = await session.execute(
-                    select(Document).where(Document.file_path == rel_path)
-                )
-                existing = result.scalar_one_or_none()
-
-                if existing is None:
-                    ext = file_path.suffix.lstrip(".").lower() or "unknown"
-                    doc = Document(
-                        file_path=rel_path,
-                        file_hash=fhash,
-                        file_format=ext,
-                        file_size=file_path.stat().st_size,
-                        file_type="immutable",
-                        title=file_path.stem,
-                        source_channel="drive_sync",
-                        category=category,
-                        needs_conversion=needs_conversion,
-                        # 안전 자료실 A-2/B-4 — watch 타깃 매핑 (KGS=law/KR 등, 비대상=NULL)
-                        material_type=target_mt,
-                        jurisdiction=target_jur,
+                async with async_session() as session:
+                    result = await session.execute(
+                        select(Document).where(Document.file_path == rel_path)
                     )
-                    # B-4 — 타깃 폴더 license 주입(restricted 포함, 비대상=미주입). classify 는
-                    # material_type IS NULL 일 때만 제안 + extract_meta 미기록이라 주입 보존.
-                    if target_license:
-                        doc.extract_meta = {"license": dict(target_license)}
-                    session.add(doc)
-                    await session.flush()
+                    existing = result.scalar_one_or_none()
 
-                    if next_stage:
-                        await enqueue_stage(session, doc.id, next_stage)
-                    new_count += 1
+                    if existing is None:
+                        ext = file_path.suffix.lstrip(".").lower() or "unknown"
+                        doc = Document(
+                            file_path=rel_path,
+                            file_hash=fhash,
+                            file_format=ext,
+                            file_size=file_path.stat().st_size,
+                            file_type="immutable",
+                            title=file_path.stem,
+                            source_channel="drive_sync",
+                            category=category,
+                            needs_conversion=needs_conversion,
+                            # 안전 자료실 A-2/B-4 — watch 타깃 매핑 (KGS=law/KR 등, 비대상=NULL)
+                            material_type=target_mt,
+                            jurisdiction=target_jur,
+                        )
+                        # B-4 — 타깃 폴더 license 주입(restricted 포함, 비대상=미주입). classify 는
+                        # material_type IS NULL 일 때만 제안 + extract_meta 미기록이라 주입 보존.
+                        if target_license:
+                            doc.extract_meta = {"license": dict(target_license)}
+                        session.add(doc)
+                        await session.flush()
 
-                elif existing.file_hash != fhash:
-                    existing.file_hash = fhash
-                    existing.file_size = file_path.stat().st_size
-                    # 기존 문서에 category/quarantine flag 가 비어있으면 보정
-                    if existing.category is None and category is not None:
-                        existing.category = category
-                    if needs_conversion and not getattr(existing, "needs_conversion", False):
-                        existing.needs_conversion = True
-                    # B-4 — 축/license 보정(B-4 이전 적재분이 재변경 시): material 미설정 시 주입,
-                    # license 부재 시에만 merge 주입(clobber 회피 — 기존 extract_meta 키 보존).
-                    if existing.material_type is None and target_mt is not None:
-                        existing.material_type = target_mt
-                        existing.jurisdiction = target_jur
-                    if target_license and not (existing.extract_meta or {}).get("license"):
-                        meta = dict(existing.extract_meta or {})
-                        meta["license"] = dict(target_license)
-                        existing.extract_meta = meta
+                        if next_stage:
+                            await enqueue_stage(session, doc.id, next_stage)
+                        await session.commit()
+                        new_count += 1
 
-                    if next_stage:
-                        await enqueue_stage(session, existing.id, next_stage)
-                    changed_count += 1
+                    elif existing.file_hash != fhash:
+                        existing.file_hash = fhash
+                        existing.file_size = file_path.stat().st_size
+                        # 기존 문서에 category/quarantine flag 가 비어있으면 보정
+                        if existing.category is None and category is not None:
+                            existing.category = category
+                        if needs_conversion and not getattr(existing, "needs_conversion", False):
+                            existing.needs_conversion = True
+                        # B-4 — 축/license 보정(B-4 이전 적재분이 재변경 시): material 미설정 시 주입,
+                        # license 부재 시에만 merge 주입(clobber 회피 — 기존 extract_meta 키 보존).
+                        if existing.material_type is None and target_mt is not None:
+                            existing.material_type = target_mt
+                            existing.jurisdiction = target_jur
+                        if target_license and not (existing.extract_meta or {}).get("license"):
+                            meta = dict(existing.extract_meta or {})
+                            meta["license"] = dict(target_license)
+                            existing.extract_meta = meta
 
-        await session.commit()
+                        if next_stage:
+                            await enqueue_stage(session, existing.id, next_stage)
+                        await session.commit()
+                        changed_count += 1
+                    # else: 무변경 → 쓰기 없음 (세션 자동 닫힘, commit 불요)
+            except Exception as e:
+                logger.warning("[PKM] 파일 처리 실패 skip path=%s: %s", rel_path, e)
+                continue
 
     if new_count or changed_count:
         logger.info(f"[Inbox+§3] 새 파일 {new_count}건, 변경 파일 {changed_count}건 등록")

app/workers/marker_worker.py

@@ -300,6 +300,11 @@ async def _process_single(
             f"[marker] transient error id={document_id} kind={type(exc).__name__}: {exc}"
         )
         raise
+    except json.JSONDecodeError as exc:
+        # 200 응답의 truncated/malformed body — 연결 흔들림 등 transient. _fail(non-retryable)
+        # 로 박지 말고 raise → queue retry (max_attempts 까지). 진짜 손상이면 재시도 후 failed.
+        logger.warning(f"[marker] malformed json body (200) id={document_id}: {exc}")
+        raise
     except Exception as exc:
         logger.exception(f"[marker] unexpected error id={document_id}: {exc}")
         await _fail(session, document_id, str(exc)[:1000])

app/workers/presegment_worker.py

@@ -497,12 +497,18 @@ async def process(document_id: int, session: AsyncSession) -> None:
         logger.warning(f"[presegment] id={document_id} file not found ({raw}) → extract")
         return
 
+    import asyncio
+
     import fitz  # PyMuPDF — extract_worker/marker_worker 와 동일 의존
 
+    def _read_toc(path: str):
+        # fitz open/get_toc 는 동기 blocking — live 스테이지라 이벤트루프(같은 루프의 1분 consumer +
+        # FastAPI 요청) 점유 회피 위해 to_thread 오프로드(거대/손상 PDF 파싱 수백 ms~초).
+        with fitz.open(path) as pdf:
+            return pdf.page_count, (pdf.get_toc(simple=True) or [])
+
     try:
-        with fitz.open(str(source)) as pdf:
-            page_count = pdf.page_count
-            toc = pdf.get_toc(simple=True) or []
+        page_count, toc = await asyncio.to_thread(_read_toc, str(source))
     except Exception as exc:
         # PDF 손상 등 — 분할 불가. 단일문서로 통과(extract 가 PyMuPDF/OCR 로 재시도하며 가시화).
         logger.warning(

app/workers/study_publish_worker.py

@@ -28,6 +28,8 @@ logger = setup_logger("study_publish_worker")
 BATCH_SIZE = 500
 # pg_advisory_xact_lock 전역 단일 라이터 키(발행 워커 전용 임의 상수, 타 advisory 락과 비충돌).
 ADVISORY_LOCK_KEY = 838201
+# 행별 격리 재시도 상한 — 초과 시 failed_at 스탬프(terminal)로 select 에서 제외.
+MAX_OUTBOX_ATTEMPTS = 5
 
 
 async def consume_publish_outbox() -> None:
@@ -46,11 +48,15 @@ async def consume_publish_outbox() -> None:
             max_rev = int(
                 (await session.execute(select(func.coalesce(func.max(Published.rev), 0)))).scalar() or 0
             )
-            # 3) 미처리 outbox 를 커밋순(id)으로.
+            # 3) 미처리 outbox 를 커밋순(id)으로. failed_at(terminal) 은 제외 — poison 행이
+            #    head-of-line 을 영구 점유하지 않게 함.
             rows = (
                 await session.execute(
                     select(PublishOutbox)
-                    .where(PublishOutbox.processed_at.is_(None))
+                    .where(
+                        PublishOutbox.processed_at.is_(None),
+                        PublishOutbox.failed_at.is_(None),
+                    )
                     .order_by(PublishOutbox.id.asc())
                     .limit(BATCH_SIZE)
                 )
@@ -60,59 +66,86 @@ async def consume_publish_outbox() -> None:
 
             now = datetime.now(timezone.utc)
             published_count = 0
+            failed_count = 0
             for ob in rows:
-                existing = (
-                    await session.execute(
-                        select(Published).where(
-                            Published.kind == ob.kind,
-                            Published.source_id == ob.source_id,
-                        )
-                    )
-                ).scalar_one_or_none()
+                try:
+                    # 행 단위 savepoint 격리 — 한 행의 예외가 배치 전체(앞 행 processed_at 포함)를
+                    # 롤백해 poison 행이 다음 사이클에 다시 최저 id 로 선택되는 무한 재선택을 차단.
+                    async with session.begin_nested():
+                        existing = (
+                            await session.execute(
+                                select(Published).where(
+                                    Published.kind == ob.kind,
+                                    Published.source_id == ob.source_id,
+                                )
+                            )
+                        ).scalar_one_or_none()
 
-                # (payload_hash, deleted) 디둡 — no-op 재투영은 rev 안 올림.
-                if (
-                    existing is not None
-                    and existing.payload_hash == ob.payload_hash
-                    and existing.deleted == ob.deleted
-                ):
-                    ob.processed_at = now
+                        # (payload_hash, deleted) 디둡 — no-op 재투영은 rev 안 올림.
+                        is_noop = (
+                            existing is not None
+                            and existing.payload_hash == ob.payload_hash
+                            and existing.deleted == ob.deleted
+                        )
+                        if is_noop:
+                            ob.processed_at = now
+                        else:
+                            new_rev = max_rev + 1
+                            if existing is None:
+                                session.add(
+                                    Published(
+                                        kind=ob.kind,
+                                        source_id=ob.source_id,
+                                        pub_id=uuid.uuid4().hex,
+                                        payload=ob.payload,
+                                        payload_hash=ob.payload_hash,
+                                        schema_version=ob.schema_version,
+                                        rev=new_rev,
+                                        deleted=ob.deleted,
+                                        created_at=now,
+                                        updated_at=now,
+                                    )
+                                )
+                            else:
+                                existing.payload = ob.payload
+                                existing.payload_hash = ob.payload_hash
+                                existing.schema_version = ob.schema_version
+                                existing.deleted = ob.deleted
+                                existing.rev = new_rev
+                                existing.updated_at = now
+                            ob.processed_at = now
+                            # 배치 내 동일 (kind, source_id) 후속 행이 직전 반영을 보도록 flush(최신 승).
+                            await session.flush()
+                except Exception as row_err:
+                    # savepoint 롤백 = 이 행의 쓰기(processed_at 포함) 취소. attempts/failed_at 만
+                    # 바깥 트랜잭션에 누적돼 최종 commit 으로 영속(영구 재선택 방지).
+                    ob.attempts = (ob.attempts or 0) + 1
+                    if ob.attempts >= MAX_OUTBOX_ATTEMPTS:
+                        ob.failed_at = now
+                        failed_count += 1
+                        logger.error(
+                            "publish_outbox_row_terminal id=%s kind=%s source_id=%s attempts=%s: %s",
+                            ob.id, ob.kind, ob.source_id, ob.attempts, row_err,
+                        )
+                    else:
+                        logger.warning(
+                            "publish_outbox_row_retry id=%s kind=%s source_id=%s attempts=%s: %s",
+                            ob.id, ob.kind, ob.source_id, ob.attempts, row_err,
+                        )
                     continue
-
-                max_rev += 1
-                if existing is None:
-                    session.add(
-                        Published(
-                            kind=ob.kind,
-                            source_id=ob.source_id,
-                            pub_id=uuid.uuid4().hex,
-                            payload=ob.payload,
-                            payload_hash=ob.payload_hash,
-                            schema_version=ob.schema_version,
-                            rev=max_rev,
-                            deleted=ob.deleted,
-                            created_at=now,
-                            updated_at=now,
-                        )
-                    )
                 else:
-                    existing.payload = ob.payload
-                    existing.payload_hash = ob.payload_hash
-                    existing.schema_version = ob.schema_version
-                    existing.deleted = ob.deleted
-                    existing.rev = max_rev
-                    existing.updated_at = now
-
-                ob.processed_at = now
-                # 배치 내 동일 (kind, source_id) 후속 행이 직전 반영을 보도록 flush(최신 승).
-                await session.flush()
-                published_count += 1
+                    # savepoint 커밋 성공 시에만 rev 카운터 전진(실패 행은 rev 미소모 → 드물게 gap,
+                    # 단일 라이터·커밋순 부여라 viewer since-rev 증분 동기 정합엔 무해).
+                    if not is_noop:
+                        max_rev = new_rev
+                        published_count += 1
 
             await session.commit()
             logger.info(
-                "publish_outbox_drained scanned=%s published=%s max_rev=%s",
+                "publish_outbox_drained scanned=%s published=%s failed=%s max_rev=%s",
                 len(rows),
                 published_count,
+                failed_count,
                 max_rev,
             )
         except Exception as e:

docker-compose.yml

@@ -3,7 +3,13 @@ services:
     image: pgvector/pgvector:pg16
     volumes:
       - pgdata:/var/lib/postgresql/data
-      - ./migrations:/docker-entrypoint-initdb.d
+      # ★ 2026-06-29 fresh-DB/DR 부팅 fix: initdb.d 마운트 제거(기존 `./migrations:/docker-entrypoint-initdb.d`).
+      # 빈 볼륨 첫 기동 시 postgres 엔트리포인트가 migrations/*.sql(001~) 을 psql autocommit 으로 실행해
+      # 스키마는 만들되 schema_migrations 스탬프는 안 남김(runner 만 생성) → fastapi init_db 가 documents
+      # 존재로 'fresh' 를 오판해 baseline(_load_baseline_if_fresh) 로드를 건너뛰고, 빈 schema_migrations
+      # 로 001 부터 재replay → `CREATE TABLE users`(IF NOT EXISTS 없음) 충돌 → 부팅 크래시(DR/신규환경).
+      # fresh-boot 은 init_db 의 baseline 적재 + migration runner 단일 경로로 일원화(설계 의도). 기존 prod
+      # 볼륨은 비어있지 않아 init scripts 가 애초에 미발동 → 무영향.
     environment:
       POSTGRES_DB: pkm
       POSTGRES_USER: pkm

docs/architecture.md

@@ -1094,7 +1094,7 @@ services:
     image: pgvector/pgvector:pg16
     volumes:
       - pgdata:/var/lib/postgresql/data
-      - ./migrations:/docker-entrypoint-initdb.d
+      # initdb.d 마운트 제거(2026-06-29): fresh-boot 은 fastapi init_db+baseline 단일 경로.
     environment:
       POSTGRES_DB: pkm
       POSTGRES_USER: pkm

docs/storage_layout.md

@@ -71,7 +71,7 @@ GPU 서버의 NFS mount (`/proc/mounts` 실측):
 
 | 컨테이너 | 마운트 | 모드 | 비고 |
 |---|---|---|---|
-| postgres | `pgdata:/var/lib/postgresql/data` + `./migrations:/docker-entrypoint-initdb.d` | rw | DB 본체 named volume |
+| postgres | `pgdata:/var/lib/postgresql/data` | rw | DB 본체 named volume (initdb.d 마운트는 2026-06-29 제거 — 아래 관찰) |
 | kordoc-service | `${NAS}/Document_Server:/documents` | **ro** | PDF/HWP parse |
 | ocr-service | `${NAS}/Document_Server:/documents` + `ocr_models:/root/.cache` | **ro** + rw | |
 | marker-service | `${NAS}/Document_Server:/documents` + `marker_models:/models` | **ro** + rw | PDF→markdown |
@@ -84,7 +84,7 @@ GPU 서버의 NFS mount (`/proc/mounts` 실측):
 **관찰**:
 - worker 컨테이너 (kordoc/ocr/marker/stt) 는 모두 NAS **read-only** 마운트 → 원본 안전.
 - fastapi 만 NAS **rw** → 업로드/preview/extracted_images 쓰기 단일 책임.
-- `./migrations` 이 postgres 의 `docker-entrypoint-initdb.d` 와 fastapi 의 `/app/migrations` 양쪽에 마운트. 단 실제 migration runner 는 fastapi `init_db()` 만 사용 (postgres init scripts 는 첫 생성 시만 실행 → 효과 X, 안전).
+- `./migrations` 은 fastapi 의 `/app/migrations` 에만 마운트. migration runner 는 fastapi `init_db()` 단일 경로. (~2026-06-29: postgres `docker-entrypoint-initdb.d` 마운트 제거. 기존엔 "첫 생성 시만 실행 → 효과 X" 로 봤으나, 빈 볼륨 첫 기동 시 postgres 가 migrations/*.sql 을 실제 실행해 스키마는 만들되 schema_migrations 스탬프를 안 남겨 → init_db 의 baseline fresh 판정을 깨고 부팅 크래시 유발. fresh-DB/DR 부팅을 init_db+baseline 단일 경로로 일원화.)
 
 ## 정책 정리
 

frontend/src/lib/components/RelatedDocs.svelte

@@ -0,0 +1,45 @@
+<script>
+  // 관련 문서 (유사도) — 문서 레벨 임베딩 KNN. 자기완결: docId 받아 /related 조회.
+  import { onMount } from 'svelte';
+  import { api } from '$lib/api';
+
+  let { documentId } = $props();
+  let items = $state([]);
+  let loaded = $state(false);
+
+  const KIND = { law: '법령', guide: '지침', paper: '논문', standard: '표준', incident: '사례' };
+
+  onMount(async () => {
+    try {
+      const r = await api(`/documents/${documentId}/related?limit=6`);
+      items = r?.related ?? [];
+    } catch (e) { /* silent */ }
+    finally { loaded = true; }
+  });
+</script>
+
+{#if items.length}
+  <div class="rel">
+    <div class="lab">관련 문서</div>
+    {#each items as it (it.id)}
+      <a class="ri" href={`/documents/${it.id}`}>
+        <span class="rt">{it.title}</span>
+        <span class="rm">
+          {#if it.material_type && KIND[it.material_type]}<span class="kind">{KIND[it.material_type]}</span>{/if}
+          <span class="rs">{Math.round((it.sim ?? 0) * 100)}</span>
+        </span>
+      </a>
+    {/each}
+  </div>
+{/if}
+
+<style>
+  .rel { background: var(--surface); border: 1px solid var(--border); border-radius: 14px; padding: 13px; }
+  .lab { font-size: 10.5px; font-weight: 700; color: var(--text-dim); letter-spacing: .4px; margin-bottom: 8px; }
+  .ri { display: flex; align-items: baseline; gap: 8px; padding: 5px 6px; border-radius: 7px; text-decoration: none; }
+  .ri:hover { background: var(--surface-hover, #ecf0e8); }
+  .rt { flex: 1; font-size: 12px; line-height: 1.4; color: var(--text); overflow: hidden; display: -webkit-box; -webkit-line-clamp: 2; -webkit-box-orient: vertical; }
+  .rm { flex-shrink: 0; display: flex; align-items: center; gap: 5px; }
+  .kind { font-size: 9px; font-weight: 700; color: var(--accent-hover, #3d7256); background: #e3efe2; border: 1px solid #cfe3cd; border-radius: 4px; padding: 0 4px; }
+  .rs { font-size: 10.5px; font-family: ui-monospace, Menlo, monospace; color: var(--faint, #9aa090); }
+</style>

frontend/src/lib/components/SlimRail.svelte

@@ -0,0 +1,33 @@
+<script>
+  // 시안 B — 글로벌 네비 슬림 아이콘 레일 (분류 사이드바 접힘 상태). 앱 토큰 사용.
+  import { page } from '$app/stores';
+  import { Home, FolderTree, Newspaper, StickyNote, Hash, GraduationCap, MessageCircle, Inbox, CalendarCheck } from 'lucide-svelte';
+
+  const items = [
+    { href: '/', icon: Home, label: '홈', exact: true },
+    { href: '/library', icon: FolderTree, label: '문서' },
+    { href: '/news', icon: Newspaper, label: '뉴스' },
+    { href: '/memos', icon: StickyNote, label: '메모' },
+    { href: '/clause', icon: Hash, label: '절' },
+    { href: '/events', icon: CalendarCheck, label: '일정' },
+    { href: '/study', icon: GraduationCap, label: '공부' },
+    { href: '/chat', icon: MessageCircle, label: '이드' },
+    { href: '/inbox', icon: Inbox, label: '편지함' },
+  ];
+  let path = $derived($page.url.pathname);
+  const active = (it) => (it.exact ? path === it.href : path.startsWith(it.href));
+</script>
+
+<nav class="flex flex-col items-center gap-1 py-2 h-full overflow-y-auto bg-sidebar">
+  {#each items as it (it.href)}
+    {@const Icon = it.icon}
+    <a
+      href={it.href}
+      title={it.label}
+      class="flex flex-col items-center justify-center gap-0.5 w-12 h-[46px] rounded-lg text-dim hover:bg-surface-hover hover:text-accent transition-colors {active(it) ? 'bg-surface-active text-accent font-semibold' : ''}"
+    >
+      <Icon size={17} strokeWidth={1.75} />
+      <span class="text-[8.5px] leading-none tracking-tight">{it.label}</span>
+    </a>
+  {/each}
+</nav>

frontend/src/routes/+layout.svelte

@@ -11,6 +11,7 @@
   import { queueOverview } from '$lib/stores/queueOverview';
   import { MACHINE_STATE_LABEL, machineChipClass } from '$lib/utils/queueDisplay';
   import Sidebar from '$lib/components/Sidebar.svelte';
+  import SlimRail from '$lib/components/SlimRail.svelte';
   import SystemStatusDot from '$lib/components/SystemStatusDot.svelte';
   import QueueDrawer from '$lib/components/QueueDrawer.svelte';
   import QuickMemoButton from '$lib/components/QuickMemoButton.svelte';
@@ -21,7 +22,7 @@
   const PUBLIC_PATHS = ['/login', '/setup', '/__styleguide'];
   const NO_CHROME_PATHS = ['/login', '/setup', '/__styleguide'];
   // /news = 풀스크린 브리핑 → 데스크탑 상시 사이드바 없음
-  const NO_SIDEBAR_PATHS = ['/news'];
+  const NO_SIDEBAR_PATHS = ['/news', '/book'];  // /book = 책 몰입(글로벌 분류 트리 숨김, 상단 네비 유지)
 
   // toast 의미 토큰 매핑 (A-8 B3)
   const TOAST_CLASS = {
@@ -198,8 +199,8 @@
       <!-- 메인: 데스크탑 상시 사이드바 + 콘텐츠 -->
       <div class="flex-1 min-h-0 flex">
         {#if showSidebar}
-          <aside class="hidden lg:block shrink-0 overflow-hidden transition-[width] duration-200 ease-out {sidebarCollapsed ? 'w-0 border-r-0' : 'w-sidebar border-r border-default'}">
-            <Sidebar />
+          <aside class="hidden lg:block shrink-0 overflow-hidden transition-[width] duration-200 ease-out {sidebarCollapsed ? 'w-14 border-r border-default' : 'w-sidebar border-r border-default'}">
+            {#if sidebarCollapsed}<SlimRail />{:else}<Sidebar />{/if}
           </aside>
         {/if}
         <main class="flex-1 min-w-0 overflow-auto">

frontend/src/routes/book/[id]/+page.svelte

@@ -0,0 +1,281 @@
+<script>
+  // ASME/법령 절-KB — 코드북·공부 리더 (r2). parent 표준/법령을 한 권의 책처럼.
+  // 좌 인덱스(Part/章→절/조) · 중 본문(MarkdownDoc=공식·표·이미지) · breadcrumb·이전다음·양방향 백링크.
+  import { onMount, tick } from 'svelte';
+  import { page } from '$app/stores';
+  import { goto } from '$app/navigation';
+  import { api } from '$lib/api';
+  import MarkdownDoc from '$lib/components/MarkdownDoc.svelte';
+
+  let parentId = $state(null);
+  let parentTitle = $state('');
+  let clauses = $state([]);
+  let selectedId = $state(null);
+  let clauseDoc = $state(null);
+  let links = $state(null);
+  let expanded = $state({});
+  let loading = $state(false);
+  let q = $state('');
+
+  // 공부도구 (노트/형광펜/암기카드) — clause_study
+  let studyItems = $state([]);
+  let studyOpen = $state(false);
+  let noteDraft = $state('');
+  const KLABEL = { note: '노트', highlight: '형광펜', card: '암기카드' };
+  async function loadStudy(id) {
+    try { const r = await api(`/documents/${id}/study`); studyItems = r?.items ?? []; }
+    catch { studyItems = []; }
+  }
+  async function addStudy(kind, payload) {
+    if (!selectedId) return;
+    try { await api(`/documents/${selectedId}/study`, { method: 'POST', body: JSON.stringify({ kind, payload }) }); await loadStudy(selectedId); }
+    catch (e) { console.warn(e); }
+  }
+  function selText() { return (typeof window !== 'undefined' && window.getSelection ? window.getSelection().toString() : '').trim(); }
+  function addNote() { const t = noteDraft.trim(); if (!t) return; addStudy('note', { text: t }); noteDraft = ''; }
+  function addHighlight() { const s = selText(); if (!s) { studyOpen = true; alert('본문에서 형광펜 칠할 부분을 먼저 드래그하세요'); return; } addStudy('highlight', { text: s }); studyOpen = true; }
+  function addCard() {
+    const s = selText();
+    const code = links?.clause_code ?? selMeta?.clause_code ?? '';
+    addStudy('card', { cue: `${code} ${strip(clauseDoc?.title, code)}`.trim(), fact: s || (clauseDoc?.md_content ?? clauseDoc?.extracted_text ?? '').replace(/[#*>]/g, '').slice(0, 280).trim() });
+    studyOpen = true;
+  }
+  async function delStudy(id) {
+    try { await api(`/documents/${selectedId}/study/${id}`, { method: 'DELETE' }); await loadStudy(selectedId); } catch {}
+  }
+
+  let parts = $derived.by(() => {
+    const out = [], idx = {};
+    for (const c of clauses) {
+      const p = c.clause_part || '·';
+      if (!(p in idx)) { idx[p] = out.length; out.push({ part: p, items: [] }); }
+      out[idx[p]].items.push(c);
+    }
+    return out;
+  });
+  let visibleParts = $derived.by(() => {
+    const term = q.trim().toLowerCase();
+    if (!term) return parts;
+    return parts.map(g => ({ part: g.part, items: g.items.filter(c =>
+      (c.clause_code || '').toLowerCase().includes(term) || (c.title || '').toLowerCase().includes(term)) }))
+      .filter(g => g.items.length);
+  });
+  let selMeta = $derived(clauses.find((c) => c.id === selectedId) || null);
+  const strip = (t, c) => (t || '').replace(c || '', '').replace(/^[(\s)]+|[(\s)]+$/g, '').trim();
+
+  async function loadBook() {
+    const r = await api(`/documents/${parentId}/clauses`);
+    parentTitle = r?.parent_title ?? '';
+    clauses = r?.clauses ?? [];
+    const e = {};
+    for (const c of clauses) e[c.clause_part || '·'] = true;
+    expanded = e;
+  }
+  async function loadClause(id) {
+    if (!id) return;
+    loading = true;
+    selectedId = id;
+    try {
+      const [d, l] = await Promise.all([api(`/documents/${id}`), api(`/documents/${id}/backlinks`)]);
+      clauseDoc = d; links = l;
+      loadStudy(id);
+      const sel = clauses.find((c) => c.id === id);
+      if (sel) expanded = { ...expanded, [sel.clause_part || '·']: true };
+      goto(`/book/${parentId}?c=${id}`, { replaceState: true, keepFocus: true, noScroll: true });
+      await tick(); window.scrollTo({ top: 0 });
+    } finally { loading = false; }
+  }
+  onMount(async () => {
+    parentId = Number($page.params.id);
+    await loadBook();
+    const c = Number($page.url.searchParams.get('c'));
+    await loadClause(c && clauses.find((x) => x.id === c) ? c : clauses[0]?.id);
+  });
+</script>
+
+<div class="book">
+  <!-- top bar -->
+  <div class="bar">
+    <span class="brand">절-KB</span>
+    <span class="crumbs">{parentTitle} {#if selMeta}<b class="sep">›</b> {selMeta.clause_part} <b class="sep">›</b> <b>{links?.clause_code ?? selMeta.clause_code}</b>{/if}</span>
+    <div class="search"><input placeholder="절·조 번호 또는 키워드" bind:value={q} /></div>
+    <div class="tools"><span class="tool on">읽기</span><span class="tool">형광펜</span><span class="tool">노트</span><span class="tool">암기카드</span></div>
+  </div>
+
+  <div class="main">
+    <!-- left index -->
+    <aside class="idx">
+      <a class="btitle" href={`/documents/${parentId}`}>{parentTitle || '표준'}</a>
+      <div class="bmeta">절 {clauses.length} · 한 권의 책처럼 탐색</div>
+      {#each visibleParts as g (g.part)}
+        <div class="parttab" role="button" tabindex="0" onclick={() => (expanded = { ...expanded, [g.part]: !expanded[g.part] })}>
+          <span class="bar2"></span><span class="pname">{g.part}</span><span class="ct">{g.items.length}</span>
+        </div>
+        {#if expanded[g.part] || q.trim()}
+          {#each g.items as c (c.id)}
+            <div class="ci" class:on={c.id === selectedId} role="button" tabindex="0" onclick={() => loadClause(c.id)}>
+              <span class="no">{c.clause_code}</span><span class="tt">{strip(c.title, c.clause_code)}</span>
+            </div>
+          {/each}
+        {/if}
+      {/each}
+    </aside>
+
+    <!-- reader -->
+    <section class="read">
+      <div class="col">
+        {#if clauseDoc}
+          <div class="studybar">
+            <button class="sbtn" title="선택 형광펜" onclick={addHighlight}>▰</button>
+            <button class="sbtn" class:on={studyOpen} title="노트/공부" onclick={() => (studyOpen = !studyOpen)}>✎</button>
+            <button class="sbtn" title="암기카드 추가" onclick={addCard}>＋</button>
+            {#if studyItems.length}<span class="scount">{studyItems.length}</span>{/if}
+          </div>
+          <div class="kicker"><span class="pth">{selMeta?.clause_part}</span></div>
+          <div class="h-no">{links?.clause_code ?? selMeta?.clause_code}</div>
+          <h1 class="h-title">{strip(clauseDoc.title, links?.clause_code ?? '')}</h1>
+
+          <div class="flow">
+            <button class="fl" disabled={!links?.prev} onclick={() => loadClause(links?.prev?.id)}>← {links?.prev?.clause_code ?? ''}</button>
+            <button class="fl next" disabled={!links?.next} onclick={() => loadClause(links?.next?.id)}>{links?.next?.clause_code ?? ''} →</button>
+          </div>
+
+          {#key clauseDoc.id}
+            <div class="docbody">
+              <MarkdownDoc documentId={clauseDoc.id} mdContent={clauseDoc.md_content ?? clauseDoc.extracted_text} mdStatus={null} class="prose prose-base max-w-none" />
+            </div>
+          {/key}
+
+          {#if links && (links.forward.length || links.back.length)}
+            <section class="conn">
+              {#if links.forward.length}
+                <div><h4>이 절이 참조 <span>{links.forward.length}</span></h4>
+                  <div class="chiprow">{#each links.forward as f}
+                    {#if f.doc_id}<button class="ref" onclick={() => loadClause(f.doc_id)}>{f.code}</button>
+                    {:else}<span class="ref dg" title="외부/미분해">{f.code}</span>{/if}
+                  {/each}</div></div>
+              {/if}
+              {#if links.back.length}
+                <div><h4>이 절을 참조 <span>{links.back.length}</span></h4>
+                  <div class="chiprow">{#each links.back as b}<button class="ref" onclick={() => loadClause(b.doc_id)}>{b.code}</button>{/each}</div></div>
+              {/if}
+            </section>
+          {/if}
+
+          {#if studyOpen}
+            <section class="study">
+              <div class="slab">공부 — 노트 · 형광펜 · 암기카드{#if studyItems.length} <span>{studyItems.length}</span>{/if}</div>
+              <div class="noteadd">
+                <textarea bind:value={noteDraft} placeholder="이 절에 노트…" rows="2"></textarea>
+                <button class="nbtn" onclick={addNote}>노트 저장</button>
+              </div>
+              {#if studyItems.length}
+                <ul class="slist">
+                  {#each studyItems as it (it.id)}
+                    <li class="sitem">
+                      <span class="skind k-{it.kind}">{KLABEL[it.kind] ?? it.kind}</span>
+                      <span class="stext">{it.payload?.text ?? it.payload?.cue ?? ''}</span>
+                      <button class="sdel" title="삭제" onclick={() => delStudy(it.id)}>×</button>
+                    </li>
+                  {/each}
+                </ul>
+              {:else}
+                <p class="shint">본문을 드래그한 뒤 형광펜(▰)/암기카드(＋), 또는 위에 노트를 적으세요.</p>
+              {/if}
+            </section>
+          {/if}
+
+          <div class="pager">
+            <button class="pg" disabled={!links?.prev} onclick={() => loadClause(links?.prev?.id)}>
+              <div class="d">← 이전</div><div class="t"><span class="pno">{links?.prev?.clause_code ?? '—'}</span> {strip(links?.prev?.title, links?.prev?.clause_code)}</div></button>
+            <button class="pg next" disabled={!links?.next} onclick={() => loadClause(links?.next?.id)}>
+              <div class="d">다음 →</div><div class="t"><span class="pno">{links?.next?.clause_code ?? '—'}</span> {strip(links?.next?.title, links?.next?.clause_code)}</div></button>
+          </div>
+        {:else}
+          <p class="empty">{loading ? '불러오는 중…' : '왼쪽에서 절을 선택하세요'}</p>
+        {/if}
+      </div>
+    </section>
+  </div>
+</div>
+
+<style>
+  :global(body) { background: var(--bg); }
+  .book { --paper:#fbfcf9; --serif:"Iowan Old Style","Palatino Linotype","Noto Serif KR",Georgia,serif;
+    display:flex; flex-direction:column; min-height:100vh; }
+  .bar { display:flex; align-items:center; gap:14px; height:50px; padding:0 18px; background:var(--paper); border-bottom:1px solid var(--border); }
+  .brand { font-weight:700; font-size:13.5px; color:var(--text); }
+  .crumbs { color:var(--text-dim); font-size:12.5px; overflow:hidden; text-overflow:ellipsis; white-space:nowrap; max-width:46%; }
+  .crumbs b { color:var(--text); font-weight:600; } .crumbs .sep { color:#c8d6c0; margin:0 5px; }
+  .search { margin-left:auto; }
+  .search input { width:280px; background:var(--surface); border:1px solid var(--border); border-radius:9px; padding:7px 12px; font-size:13px; color:var(--text); outline:none; }
+  .search input:focus { border-color:var(--accent); }
+  .tools { display:flex; gap:2px; }
+  .tool { font-size:12px; color:var(--text-dim); padding:6px 10px; border-radius:8px; border:1px solid transparent; cursor:pointer; }
+  .tool:hover { background:var(--surface); } .tool.on { background:#ecf0e8; border-color:var(--border); color:var(--accent-hover); font-weight:600; }
+
+  .main { display:flex; align-items:flex-start; flex:1; }
+  .idx { width:264px; flex-shrink:0; align-self:stretch; border-right:1px solid var(--border);
+    background:linear-gradient(180deg,#f6f8f3,#f1f4ec); padding:16px 10px 30px 16px; position:sticky; top:0; max-height:100vh; overflow:auto; }
+  .btitle { display:block; font-family:var(--serif); font-size:15.5px; font-weight:600; color:var(--text); text-decoration:none; line-height:1.32; }
+  .btitle:hover { text-decoration:underline; }
+  .bmeta { font-size:11px; color:#9aa090; margin:3px 0 14px; }
+  .parttab { display:flex; align-items:center; gap:8px; margin:11px 0 4px; padding:3px 4px; border-radius:6px; cursor:pointer;
+    font-size:11px; font-weight:700; letter-spacing:.5px; color:var(--text-dim); text-transform:uppercase; }
+  .parttab:hover { background:#fff; } .parttab .bar2 { width:3px; height:12px; border-radius:2px; background:var(--domain-engineering); }
+  .parttab .pname { flex:1; overflow:hidden; text-overflow:ellipsis; white-space:nowrap; } .parttab .ct { color:#9aa090; font-weight:600; letter-spacing:0; }
+  .ci { display:flex; gap:9px; align-items:baseline; padding:4px 9px; border-radius:7px; cursor:pointer; line-height:1.4; }
+  .ci .no { font-family:ui-monospace,Menlo,monospace; font-size:11px; color:var(--accent); font-weight:600; min-width:52px; white-space:nowrap; }
+  .ci .tt { font-size:12.5px; color:var(--text-dim); overflow:hidden; text-overflow:ellipsis; }
+  .ci:hover { background:#fff; }
+  .ci.on { background:#fff; box-shadow:inset 3px 0 0 var(--accent), 0 1px 2px rgba(35,41,31,.05); }
+  .ci.on .no { color:var(--accent-hover); font-weight:700; } .ci.on .tt { color:var(--text); font-weight:600; }
+
+  .read { flex:1; min-width:0; padding:34px 40px 80px; }
+  .col { max-width:680px; margin:0 auto; position:relative; }
+  .studybar { position:absolute; right:-30px; top:4px; display:flex; flex-direction:column; gap:6px; }
+  .sbtn { width:34px; height:34px; border-radius:9px; border:1px solid var(--border); background:var(--paper); color:var(--text-dim); font-size:13px; cursor:pointer; }
+  .sbtn:hover { background:var(--surface); color:var(--accent-hover); }
+  .kicker { margin-bottom:5px; } .kicker .pth { font-size:11.5px; color:#9aa090; font-weight:600; letter-spacing:.3px; }
+  .h-no { font-family:ui-monospace,Menlo,monospace; font-size:13px; color:var(--accent); font-weight:700; letter-spacing:.5px; }
+  .h-title { font-family:var(--serif); font-size:26px; line-height:1.24; font-weight:600; margin:2px 0 14px; letter-spacing:-.2px; color:var(--text); }
+  .flow { display:flex; justify-content:space-between; gap:8px; margin-bottom:18px; }
+  .flow .fl { font-size:11.5px; color:var(--text-dim); background:var(--surface); border:1px solid var(--border); border-radius:8px; padding:5px 11px; cursor:pointer; }
+  .flow .fl:hover:not(:disabled) { background:#ecf0e8; } .flow .fl:disabled { opacity:.35; cursor:default; }
+  .docbody { font-size:15.5px; }
+  .docbody :global(.prose) { color:#2a3024; line-height:1.78; }
+  .docbody :global(.prose h1), .docbody :global(.prose h2), .docbody :global(.prose h3) { font-family:var(--serif); }
+  .docbody :global(a) { color:var(--accent-hover); }
+  .conn { margin-top:34px; padding-top:18px; border-top:1px solid var(--border); display:grid; grid-template-columns:1fr 1fr; gap:22px; }
+  .conn h4 { font-size:11px; font-weight:700; color:var(--text-dim); letter-spacing:.4px; margin:0 0 9px; } .conn h4 span { color:#9aa090; font-weight:500; }
+  .chiprow { display:flex; flex-wrap:wrap; gap:5px; }
+  .ref { font-family:ui-monospace,Menlo,monospace; font-size:11.5px; font-weight:600; color:var(--accent-hover); background:#eef4ec; border:1px solid #d9e6d8; border-radius:6px; padding:2px 8px; cursor:pointer; }
+  .ref:hover { background:#e2efe0; } .ref.dg { color:#9aa090; background:var(--surface); border-color:var(--border); cursor:default; }
+  .pager { display:flex; gap:10px; margin-top:30px; }
+  .pg { flex:1; text-align:left; border:1px solid var(--border); border-radius:11px; padding:11px 14px; background:var(--paper); cursor:pointer; }
+  .pg.next { text-align:right; } .pg:hover:not(:disabled) { border-color:#cfd7c6; background:#fff; } .pg:disabled { opacity:.4; cursor:default; }
+  .pg .d { font-size:10.5px; color:#9aa090; } .pg .t { font-size:12.5px; color:var(--text-dim); font-weight:600; margin-top:1px; overflow:hidden; text-overflow:ellipsis; white-space:nowrap; }
+  .pg .pno { font-family:ui-monospace,Menlo,monospace; color:var(--accent); }
+  .empty { color:#9aa090; text-align:center; padding:80px 0; }
+  .sbtn.on { background:#ecf0e8; color:var(--accent-hover,#3d7256); border-color:var(--border); }
+  .scount { font-size:9px; font-weight:700; color:#fff; background:var(--accent,#4f8a6b); border-radius:8px; padding:1px 5px; text-align:center; }
+  .study { margin-top:24px; padding:14px; border:1px solid var(--border); border-radius:12px; background:var(--surface); }
+  .slab { font-size:11px; font-weight:700; color:var(--text-dim); letter-spacing:.3px; margin-bottom:9px; }
+  .slab span { color:var(--accent-hover,#3d7256); }
+  .noteadd { display:flex; gap:8px; align-items:flex-end; margin-bottom:10px; }
+  .noteadd textarea { flex:1; resize:vertical; border:1px solid var(--border); border-radius:8px; padding:7px 9px; font-size:12.5px; font-family:inherit; color:var(--text); background:var(--paper,#fbfcf9); outline:none; }
+  .noteadd textarea:focus { border-color:var(--accent); }
+  .nbtn { flex-shrink:0; font-size:12px; color:#fff; background:var(--accent,#4f8a6b); border:0; border-radius:8px; padding:8px 12px; cursor:pointer; }
+  .nbtn:hover { background:var(--accent-hover,#3d7256); }
+  .slist { list-style:none; margin:0; padding:0; display:flex; flex-direction:column; gap:5px; }
+  .sitem { display:flex; align-items:baseline; gap:8px; padding:6px 8px; border-radius:8px; background:var(--paper,#fbfcf9); border:1px solid var(--border); }
+  .skind { flex-shrink:0; font-size:9.5px; font-weight:700; border-radius:4px; padding:1px 6px; }
+  .k-note { color:#3d7256; background:#e3efe2; border:1px solid #cfe3cd; }
+  .k-highlight { color:#8a6306; background:#faf3e2; border:1px solid #ecdca3; }
+  .k-card { color:#1d4ed8; background:#eef4fc; border:1px solid #d7e4f7; }
+  .stext { flex:1; font-size:12px; line-height:1.5; color:var(--text); white-space:pre-wrap; word-break:break-word; }
+  .sdel { flex-shrink:0; background:none; border:0; color:var(--faint,#9aa090); cursor:pointer; font-size:14px; }
+  .sdel:hover { color:var(--error,#c0392b); }
+  .shint { font-size:11.5px; color:var(--faint,#9aa090); margin:0; }
+  @media(max-width:820px){ .idx{display:none} .read{padding:24px 18px} .conn{grid-template-columns:1fr} .studybar{position:static;flex-direction:row} .crumbs{max-width:30%} .search input{width:150px} }
+</style>

frontend/src/routes/documents/[id]/+page.svelte

@@ -16,6 +16,7 @@
   import Skeleton from '$lib/components/ui/Skeleton.svelte';
   import HandwriteCanvas from '$lib/components/HandwriteCanvas.svelte';
   import MarkdownDoc from '$lib/components/MarkdownDoc.svelte';
+  import RelatedDocs from '$lib/components/RelatedDocs.svelte';
   import { renderDocMarkdown } from '$lib/utils/docMarkdown';
   import MarkdownStatusBadge from '$lib/components/MarkdownStatusBadge.svelte';
   import NoteEditor from '$lib/components/editors/NoteEditor.svelte';
@@ -321,6 +322,7 @@
 <!-- ════ 우 슬림 레일 (시안 카드 스타일) ════ -->
 {#snippet rail()}
   <div style="display:flex;flex-direction:column;gap:11px;font-size:14px;">
+    <RelatedDocs documentId={doc.id} />
     {#if doc.ai_tldr || doc.ai_summary}
       <div style="background:#f4f7f1;border:1px solid #dde3d6;border-radius:14px;padding:13px;">
         <div style="font-size:10.5px;font-weight:700;color:#697061;letter-spacing:.4px;margin-bottom:7px;">TL;DR</div>

frontend/src/routes/study/+page.svelte

@@ -1,13 +1,52 @@
 <script>
-  // /study — 학습 hub.
-  // 주제로 보기(퀴즈·복습·통계) / 자료 학습 / 필사 세션 / 암기카드 검수.
+  // /study — 학습 hub + 데일리 랜딩('오늘의 공부' 대시보드).
+  // 상단 = 이론 홈(진도·오늘의 개념·복습 due, 재노출 트리거). 하단 = 기존 모드 진입.
   import { onMount } from 'svelte';
   import { api } from '$lib/api';
-  import { BookOpen, PenLine, GraduationCap, FolderKanban, Layers, Repeat, Flag, Inbox, Activity } from 'lucide-svelte';
+  import { addToast } from '$lib/stores/toast';
+  import { BookOpen, PenLine, GraduationCap, FolderKanban, Layers, Repeat, Flag, Inbox, Activity, CalendarCheck } from 'lucide-svelte';
 
   let cardReviewCount = $state(0);
   let questionFlagCount = $state(0);
+
+  // 오늘의 공부 (이론 홈)
+  let curriculum = $state(null);
+  let todayConcepts = $state([]);
+  let dashLoading = $state(true);
+
+  let readPct = $derived(
+    curriculum && curriculum.total ? Math.round((curriculum.read / curriculum.total) * 100) : 0
+  );
+
+  async function loadDashboard() {
+    dashLoading = true;
+    try {
+      const [cur, today] = await Promise.all([
+        api('/study/curriculum'),
+        api('/study/today-concepts?limit=6'),
+      ]);
+      curriculum = cur;
+      todayConcepts = today?.concepts ?? [];
+    } catch {
+      // 대시보드 실패해도 허브 나머지는 동작 (조용히)
+    } finally {
+      dashLoading = false;
+    }
+  }
+
+  async function markRead(doc) {
+    try {
+      await api(`/study/concepts/${doc.doc_id}/read`, { method: 'POST' });
+      todayConcepts = todayConcepts.filter((c) => c.doc_id !== doc.doc_id);
+      addToast('success', `회독: ${doc.title}`);
+      loadDashboard(); // 진도 갱신
+    } catch {
+      addToast('error', '회독 처리 실패');
+    }
+  }
+
   onMount(async () => {
+    loadDashboard();
     try {
       const r = await api('/study-cards/needs-review/count');
       cardReviewCount = r?.count ?? 0;
@@ -27,6 +66,64 @@
     <p class="text-sm text-dim mt-1">주제별 퀴즈·복습(SRS)·통계 / 학습 자료 회독 / 손글씨 필사 세션.</p>
   </header>
 
+  <!-- 오늘의 공부 (이론 홈 대시보드 = 데일리 트리거) -->
+  <section class="mb-5 rounded-lg border border-default bg-surface p-4 md:p-5">
+    <div class="flex items-center gap-2 mb-3">
+      <CalendarCheck size={18} class="text-accent" />
+      <h2 class="text-base font-semibold text-text">오늘의 공부</h2>
+      {#if curriculum}
+        <span class="ml-auto text-xs text-dim">이론 회독 <span class="text-text font-medium">{curriculum.read}</span> / {curriculum.total} ({readPct}%)</span>
+      {/if}
+    </div>
+
+    {#if dashLoading}
+      <p class="text-xs text-dim">불러오는 중…</p>
+    {:else}
+      {#if curriculum}
+        <div class="h-2 rounded-full bg-bg overflow-hidden mb-3">
+          <div class="h-full bg-accent" style="width: {readPct}%"></div>
+        </div>
+        <div class="flex flex-wrap gap-x-4 gap-y-1 mb-4 text-xs text-dim">
+          {#each curriculum.subjects as s}
+            <span>{s.subject} <span class="text-text">{s.read}/{s.total}</span></span>
+          {/each}
+        </div>
+
+        <div class="flex flex-wrap gap-2 mb-4">
+          <a
+            href="/study/topics/{curriculum.topic_id}/review-queue"
+            class="flex items-center gap-1.5 rounded border border-default px-3 py-1.5 text-xs text-dim hover:border-accent hover:text-text transition-colors"
+          >
+            <Repeat size={13} /> 문항 복습 <span class="font-semibold text-text">{curriculum.question_due}</span>
+          </a>
+          <span class="flex items-center gap-1.5 rounded border border-default px-3 py-1.5 text-xs text-dim">
+            <BookOpen size={13} /> 개념 재복습 <span class="font-semibold text-text">{curriculum.concept_due}</span>
+          </span>
+        </div>
+      {/if}
+
+      <div class="text-xs text-dim mb-2">오늘의 개념</div>
+      {#if todayConcepts.length === 0}
+        <p class="text-xs text-dim">오늘 볼 개념이 없습니다. 잘 하고 있어요.</p>
+      {:else}
+        <ul class="space-y-1.5">
+          {#each todayConcepts as c (c.doc_id)}
+            <li class="flex items-center gap-2 rounded border border-default px-3 py-2">
+              <span class="text-accent shrink-0 text-xs" title="빈출">{#each Array(c.freq) as _}★{/each}</span>
+              <a href="/documents/{c.doc_id}" class="text-sm text-text hover:text-accent truncate flex-1">{c.title}</a>
+              <span class="shrink-0 text-[10px] rounded-full px-2 py-0.5 {c.reason === '재복습' ? 'bg-accent/15 text-accent' : 'bg-surface border border-default text-dim'}">{c.reason}</span>
+              <button
+                type="button"
+                onclick={() => markRead(c)}
+                class="shrink-0 text-xs rounded border border-default px-2 py-1 text-dim hover:border-accent hover:text-accent transition-colors"
+              >읽음</button>
+            </li>
+          {/each}
+        </ul>
+      {/if}
+    {/if}
+  </section>
+
   <a
     href="/study/topics"
     class="block mb-3 p-5 rounded-lg border border-default bg-surface hover:border-accent hover:bg-accent/5 transition-colors"
@@ -126,7 +223,8 @@
   <div class="mt-6 p-4 rounded-lg border border-dashed border-default/60 text-xs text-dim">
     <div class="font-medium text-dim mb-1">예정</div>
     <ul class="list-disc list-inside space-y-0.5">
-      <li>애플워치 빠른복습 + 공부 알람(push)</li>
+      <li>개념 학습 리더 (가리고 떠올리기 · 빈출★ · 관련개념 백링크)</li>
+      <li>이론↔문제 연결 (개념별 정답률 · 약점 개념 지도)</li>
     </ul>
   </div>
 </div>

migrations/377_domain_bucket.sql

@@ -0,0 +1,23 @@
+-- 377_domain_bucket.sql
+-- ai_domain(반자유 AI 분류, 드리프트 존재)을 검색 스코프용 7버킷으로 결정적 롤업.
+-- 축: ai_domain(routing/해석 축)의 coarsening — category(UI축) 아님 (feedback_category_vs_ai_domain_axis 준수).
+-- 버킷: News / Safety / Law / Engineering / General / Philosophy / Programming.
+-- STORED generated → 신규/재분류 문서도 ai_domain 붙으면 자동 버킷. ai_domain 원본 보존(하위 검색 유지).
+-- 롤백: ALTER TABLE documents DROP COLUMN domain_bucket;
+ALTER TABLE documents ADD COLUMN IF NOT EXISTS domain_bucket text
+  GENERATED ALWAYS AS (
+    CASE
+      WHEN ai_domain LIKE 'News%' THEN 'News'
+      WHEN ai_domain = '법령' OR ai_domain LIKE 'Industrial_Safety/Legislation%' THEN 'Law'
+      WHEN ai_domain = 'Safety' OR ai_domain LIKE 'Safety/%'
+           OR ai_domain LIKE 'Industrial_Safety%'
+           OR ai_domain = 'Knowledge/Industrial_Safety' THEN 'Safety'
+      WHEN ai_domain LIKE 'Engineering%' OR ai_domain = 'Knowledge/Engineering' THEN 'Engineering'
+      WHEN ai_domain LIKE 'Philosophy%' THEN 'Philosophy'
+      WHEN ai_domain LIKE 'Programming%' THEN 'Programming'
+      ELSE 'General'
+    END
+  ) STORED;
+
+CREATE INDEX IF NOT EXISTS documents_domain_bucket_idx
+  ON documents (domain_bucket) WHERE deleted_at IS NULL;

migrations/378_publish_outbox_attempts_failed.sql

@@ -0,0 +1,9 @@
+-- 378_publish_outbox_attempts_failed.sql
+-- (번호: 멀티세션 중 prod 가 377_domain_bucket 을 선점 → 378 로 리넘버.)
+-- publish_outbox poison row head-of-line block 차단. 발행 워커가 행별 savepoint 격리 후
+-- 예외 시 attempts++ 하고 MAX 초과 시 failed_at 스탬프(terminal) → 그 행을 select 에서 제외해
+-- 후속 발행이 막히지 않게 함. 기존 미처리 행은 attempts=0 / failed_at=NULL 로 정상 재처리.
+-- (단일 ALTER = 1 statement = asyncpg prepared 호환.)
+ALTER TABLE publish_outbox
+    ADD COLUMN IF NOT EXISTS attempts SMALLINT NOT NULL DEFAULT 0,
+    ADD COLUMN IF NOT EXISTS failed_at TIMESTAMPTZ;

migrations/379_asme_clause_kb.sql

@@ -0,0 +1,37 @@
+-- 379_asme_clause_kb.sql
+-- ASME 절-지식베이스: 절 = 개별 documents 행(parent_id) + 절↔절 백링크 + 태깅 (additive, idempotent)
+-- 검색 무접촉: 절 doc 은 embedding NULL(벡터 제외) + doc_kind='clause'(retrieval doc-leg 필터로 제외).
+
+ALTER TABLE documents
+  ADD COLUMN IF NOT EXISTS parent_id    bigint REFERENCES documents(id),
+  ADD COLUMN IF NOT EXISTS doc_kind     text NOT NULL DEFAULT 'standard',
+  ADD COLUMN IF NOT EXISTS clause_code  text,
+  ADD COLUMN IF NOT EXISTS clause_part  text,
+  ADD COLUMN IF NOT EXISTS clause_order int;
+
+CREATE INDEX IF NOT EXISTS idx_documents_parent_id   ON documents(parent_id) WHERE parent_id IS NOT NULL;
+CREATE INDEX IF NOT EXISTS idx_documents_doc_kind    ON documents(doc_kind);
+CREATE INDEX IF NOT EXISTS idx_documents_clause_code ON documents(clause_code) WHERE clause_code IS NOT NULL;
+
+-- 절↔절 백링크 (dangling 허용: dst_doc_id nullable)
+CREATE TABLE IF NOT EXISTS clause_links (
+  id          bigserial PRIMARY KEY,
+  src_doc_id  bigint NOT NULL REFERENCES documents(id) ON DELETE CASCADE,
+  dst_code    text   NOT NULL,
+  dst_doc_id  bigint REFERENCES documents(id) ON DELETE SET NULL,
+  anchor      text,
+  ctx         text,
+  char_off    int
+);
+CREATE INDEX IF NOT EXISTS idx_clause_links_src     ON clause_links(src_doc_id);
+CREATE INDEX IF NOT EXISTS idx_clause_links_dst     ON clause_links(dst_doc_id) WHERE dst_doc_id IS NOT NULL;
+CREATE INDEX IF NOT EXISTS idx_clause_links_dstcode ON clause_links(dst_code);
+
+-- 태깅 (Part 자동 + 주제)
+CREATE TABLE IF NOT EXISTS document_tags (
+  doc_id    bigint NOT NULL REFERENCES documents(id) ON DELETE CASCADE,
+  tag       text   NOT NULL,
+  tag_kind  text   NOT NULL DEFAULT 'topic',
+  PRIMARY KEY (doc_id, tag)
+);
+CREATE INDEX IF NOT EXISTS idx_document_tags_tag ON document_tags(tag);

migrations/380_clause_study.sql

@@ -0,0 +1,9 @@
+-- 380_clause_study.sql — 절-문서 공부도구(노트/형광펜/암기카드) 저장. FK 없음(documents 락 회피).
+CREATE TABLE IF NOT EXISTS clause_study (
+  id         bigserial PRIMARY KEY,
+  doc_id     bigint NOT NULL,
+  kind       text   NOT NULL,            -- 'note' | 'highlight' | 'card'
+  payload    jsonb  NOT NULL DEFAULT '{}',
+  created_at timestamptz NOT NULL DEFAULT now()
+);
+CREATE INDEX IF NOT EXISTS idx_clause_study_doc ON clause_study(doc_id, kind);

migrations/381_study_concept_progress.sql

@@ -0,0 +1,16 @@
+-- 381_study_concept_progress.sql — 이론 개념(문서) 간격반복(SR) 진행. 이론공부 홈 트리거.
+-- concept_doc_id 는 documents.id 를 가리키나 FK 미설정(hot 테이블 락 회피, clause_study 380 선례).
+-- SR 산술은 study_question_progress 와 동일(sr_schedule 공용): stage 0→1→2→3(1·3·7·14일)→4 졸업.
+CREATE TABLE IF NOT EXISTS study_concept_progress (
+  id             bigserial PRIMARY KEY,
+  user_id        bigint NOT NULL REFERENCES users(id) ON DELETE CASCADE,
+  study_topic_id bigint NOT NULL REFERENCES study_topics(id) ON DELETE CASCADE,
+  concept_doc_id bigint NOT NULL,
+  review_stage   smallint,
+  due_at         timestamptz,
+  last_read_at   timestamptz,
+  created_at     timestamptz NOT NULL DEFAULT now(),
+  updated_at     timestamptz NOT NULL DEFAULT now(),
+  CONSTRAINT uq_concept_progress_user_doc UNIQUE (user_id, concept_doc_id)
+);
+CREATE INDEX IF NOT EXISTS idx_concept_progress_due ON study_concept_progress(user_id, due_at) WHERE due_at IS NOT NULL;

scripts/asme_backlinks_persist.py

@@ -0,0 +1,53 @@
+#!/usr/bin/env python3
+"""ASME clause-KB backlinks: resolve clause-id mentions in each clause doc -> clause_links.
+dst resolved to the clause doc of the same parent (top-level code); sub-code mention -> anchor;
+unresolved (cross-standard / material spec not split) -> dangling (dst_doc_id NULL).
+Idempotent per parent. Usage: python3 asme_backlinks_persist.py <parent_id> [--commit]
+"""
+import asyncio, os, re, sys
+
+MENTION_RE = re.compile(r'(?<![A-Za-z0-9])([A-Z]{1,4}-\d+(?:\.\d+)*[A-Za-z]?)(?![A-Za-z0-9])')
+def top(code): return re.match(r'^[A-Z]{1,4}-\d+', code).group(0)
+
+async def main():
+    parent = int(sys.argv[1]); commit = '--commit' in sys.argv
+    import asyncpg
+    conn = await asyncpg.connect(os.environ['DATABASE_URL'].replace('+asyncpg', ''))
+    docs = await conn.fetch("SELECT id, clause_code, md_content FROM documents "
+                            "WHERE parent_id=$1 AND doc_kind='clause' ORDER BY clause_order", parent)
+    code2id = {d['clause_code']: d['id'] for d in docs}
+    edges = []          # (src_id, dst_code, dst_doc_id, anchor, ctx, char_off)
+    resolved = dangling = 0
+    for d in docs:
+        body = d['md_content']; src_top = d['clause_code']
+        seen = set()
+        for m in MENTION_RE.finditer(body):
+            code = m.group(1); t = top(code)
+            if t == src_top: continue                 # self-reference
+            if (d['id'], code) in seen: continue      # dedup per (src,dst_code)
+            seen.add((d['id'], code))
+            dst_id = code2id.get(t)                    # resolve to same-parent clause doc
+            anchor = code.lower().replace('.', '-') if code != t else None
+            off = m.start()
+            ctx = re.sub(r'\s+', ' ', body[max(0, off-50):off+50]).strip()
+            edges.append((d['id'], code, dst_id, anchor, ctx, off))
+            if dst_id: resolved += 1
+            else: dangling += 1
+    print(f"parent={parent} clause_docs={len(docs)} edges={len(edges)} resolved={resolved} dangling={dangling}")
+    # top referenced clauses
+    from collections import Counter
+    tgt = Counter(top(e[1]) for e in edges if e[2])
+    print("most-referenced:", tgt.most_common(8))
+    if not commit:
+        print("DRY-RUN. pass --commit to persist."); await conn.close(); return
+    async with conn.transaction():
+        ids = [d['id'] for d in docs]
+        await conn.execute("DELETE FROM clause_links WHERE src_doc_id = ANY($1::bigint[])", ids)
+        await conn.executemany(
+            "INSERT INTO clause_links(src_doc_id,dst_code,dst_doc_id,anchor,ctx,char_off) "
+            "VALUES ($1,$2,$3,$4,$5,$6)", edges)
+        n = await conn.fetchval("SELECT count(*) FROM clause_links WHERE src_doc_id = ANY($1::bigint[])", ids)
+        print(f"COMMITTED: {n} clause_links for parent {parent}")
+    await conn.close()
+
+asyncio.run(main())

scripts/asme_clause_persist.py

@@ -0,0 +1,118 @@
+#!/usr/bin/env python3
+"""ASME clause-KB persist (v2: over-CAP pagination). Split a parent standard into per-clause
+documents (A-granularity); over-CAP clause bodies are paginated into readable page-docs.
+Idempotent per parent. doc_kind='clause', embedding NULL (search-excluded), parent_id=<parent>.
+Usage: python3 asme_clause_persist.py <parent_id> [--commit]
+"""
+import asyncio, os, re, sys, hashlib, statistics
+
+CAP = 12000; PAGE_TOK = 11000
+EN, KO = 0.217, 0.529
+LINE_RE = re.compile(r'^([ \t#>*]{0,8})([A-Z]{2,4}-\d+(?:\.\d+)*[A-Za-z]?)(.*)$')
+MENTION_RE = re.compile(r'(?<![A-Za-z0-9])([A-Z]{1,4}-\d+(?:\.\d+)*[A-Za-z]?)(?![A-Za-z0-9])')
+EXACT_TOP = re.compile(r'^[A-Z]{2,4}-\d+$')
+TITLE_AFTER = re.compile(r'^[\s.]*[A-Z(]')
+REF_LEAD = re.compile(r'^[\s.]*(and|or|to|of|in|on|the|as|is|are|shall|through|per|see|with|'
+                      r'for|by|that|which|such|또는|및|등|의|은|는|에|을|를|과|와)\b', re.I)
+
+def tok(s):
+    ko = sum(1 for c in s if '가' <= c <= '힣'); return int((len(s)-ko)*EN + ko*KO)
+
+def clean_title(rest):
+    t = re.sub(r'<sup>ð</sup>\s*\**\d*\**\s*<sup>Þ</sup>', '', rest)
+    t = re.sub(r'ð\**\d*\**Þ', '', t)
+    t = t.replace('**', '').replace('#', '')
+    return re.sub(r'\s+', ' ', t).strip(' *:—-')
+
+def is_header(markup, rest):
+    if '#' in markup or '*' in markup: return True
+    rs = rest.strip()
+    if rs == '': return True
+    if REF_LEAD.match(rest): return False
+    if rs[0] in ',;.)': return False
+    if '가' <= rs[0] <= '힣': return False
+    if rs[0].islower(): return False
+    return bool(TITLE_AFTER.match(rs))
+
+def paginate(body):
+    """split an over-CAP body into <=MAX_PAGES line-aligned pages of ~PAGE_TOK tokens."""
+    pages, cur, ct = [], [], 0
+    for ln in body.split('\n'):
+        lt = tok(ln) + 1
+        if ct + lt > PAGE_TOK and cur:
+            pages.append('\n'.join(cur)); cur, ct = [ln], lt
+        else:
+            cur.append(ln); ct += lt
+    if cur: pages.append('\n'.join(cur))
+    return pages
+
+def build_clauses(text):
+    lines = text.split('\n'); off = []; a = 0
+    for ln in lines: off.append(a); a += len(ln) + 1
+    bounds = []; seen = set()
+    for i, ln in enumerate(lines):
+        m = LINE_RE.match(ln)
+        if not m: continue
+        markup, code, rest = m.group(1), m.group(2), m.group(3)
+        if not EXACT_TOP.match(code): continue
+        if not is_header(markup, rest): continue
+        if code in seen: continue
+        seen.add(code); bounds.append((off[i], code, clean_title(rest)))
+    raw = []
+    for idx, (start, code, title) in enumerate(bounds):
+        end = bounds[idx+1][0] if idx+1 < len(bounds) else len(text)
+        body = text[start:end]
+        part = re.match(r'^[A-Z]{2,4}', code).group(0)
+        links = sorted(set(re.match(r'^[A-Z]{1,4}-\d+', mm).group(0)
+                           for mm in MENTION_RE.findall(body)) - {code})
+        raw.append(dict(code=code, part=part, title=(code + (' ' + title if title else '')),
+                        body=body, tok=tok(body), links=links))
+    # expand over-CAP into pages; assign running clause_order
+    final, order = [], 0
+    for c in raw:
+        if c['tok'] <= CAP:
+            final.append({**c, 'order': order}); order += 1; continue
+        pages = paginate(c['body'])
+        for pi, pb in enumerate(pages):
+            code = c['code'] if pi == 0 else f"{c['code']}·p{pi+1}"
+            title = c['title'] if pi == 0 else f"{c['title']} (페이지 {pi+1}/{len(pages)})"
+            final.append(dict(code=code, part=c['part'], order=order, title=title,
+                              body=pb, tok=tok(pb), links=c['links'] if pi == 0 else []))
+            order += 1
+    return final
+
+async def main():
+    parent = int(sys.argv[1]); commit = '--commit' in sys.argv
+    import asyncpg
+    conn = await asyncpg.connect(os.environ['DATABASE_URL'].replace('+asyncpg', ''))
+    row = await conn.fetchrow("SELECT md_content, ai_domain, data_origin FROM documents WHERE id=$1", parent)
+    if not row: print(f"parent {parent} not found"); return
+    clauses = build_clauses(row['md_content'])
+    toks = [c['tok'] for c in clauses]
+    over = [c for c in clauses if c['tok'] > CAP]
+    print(f"parent={parent} clause_docs={len(clauses)} median_tok={int(statistics.median(toks))} "
+          f"max_tok={max(toks)} over_cap_remaining={len(over)}")
+    if over: print("still over-CAP:", [f"{c['code']}:{c['tok']}t" for c in over])
+    if not commit:
+        print("DRY-RUN. pass --commit to persist."); await conn.close(); return
+    async with conn.transaction():
+        deld = await conn.execute("DELETE FROM documents WHERE parent_id=$1 AND doc_kind='clause'", parent)
+        print("deleted prior:", deld)
+        for c in clauses:
+            fh = hashlib.sha256(f"{parent}:{c['code']}:{c['body']}".encode()).hexdigest()
+            cid = await conn.fetchval("""
+                INSERT INTO documents
+                  (file_format, file_hash, title, md_content, parent_id, doc_kind,
+                   clause_code, clause_part, clause_order, ai_domain, data_origin,
+                   md_status, review_status, conversion_status, preview_status)
+                VALUES ('md',$1,$2,$3,$4,'clause',$5,$6,$7,$8,$9,'success','approved','none','none')
+                RETURNING id
+            """, fh, c['title'], c['body'], parent, c['code'], c['part'], c['order'],
+                 row['ai_domain'], row['data_origin'] or 'external')
+            await conn.execute("INSERT INTO document_tags(doc_id,tag,tag_kind) VALUES ($1,$2,'part') "
+                               "ON CONFLICT DO NOTHING", cid, c['part'])
+        n = await conn.fetchval("SELECT count(*) FROM documents WHERE parent_id=$1 AND doc_kind='clause'", parent)
+        print(f"COMMITTED: {n} clause docs for parent {parent}")
+    await conn.close()
+
+asyncio.run(main())

scripts/backfill_publish_card_progress.py

@@ -31,8 +31,8 @@ from core.database import async_session
 from models.study_memo_card_progress import StudyMemoCardProgress
 from services.study.publish_enqueue import backfill_publish_card_progress
 
-# 개인 학습툴 progress row 대비 넉넉. 도달 시 가드 경보.
-PAGE = 100000
+# 페이지 배치 크기 — after_id 루프로 전량 처리(bounded tx).
+PAGE = 5000
 
 
 async def run(dry_run: bool) -> None:
@@ -50,13 +50,17 @@ async def run(dry_run: bool) -> None:
         print("[dry-run] 적재 안 함. 실제 실행은 --dry-run 제거.")
         return
 
-    async with async_session() as session:
-        n = await backfill_publish_card_progress(session, after_id=0, limit=PAGE)
-        await session.commit()
+    total = 0
+    after = 0
+    while True:
+        async with async_session() as session:
+            n, after = await backfill_publish_card_progress(session, after_id=after, limit=PAGE)
+            await session.commit()
+        total += n
+        if n < PAGE:
+            break
 
-    print(f"\n[ok] outbox 적재 {n}건 — 발행 워커가 drain(flag on 시) 하며 rev 부여.")
-    if n >= PAGE:
-        print(f"[warn] PAGE({PAGE}) 도달 — progress 가 더 있을 수 있음. after_id 페이징 추가 필요.")
+    print(f"\n[ok] outbox 적재 {total}건 — 발행 워커가 drain(flag on 시) 하며 rev 부여.")
 
 
 def main() -> None:

scripts/backfill_publish_cards.py

@@ -31,8 +31,8 @@ from core.database import async_session
 from models.study_memo_card import StudyMemoCard
 from services.study.publish_enqueue import backfill_publish_cards
 
-# 개인 학습툴 카드 수 대비 넉넉(단일 outbox 적재 tx, 워커는 BATCH_SIZE 로 drain). 도달 시 가드 경보.
-PAGE = 100000
+# 페이지 배치 크기 — after_id 루프로 전량 처리(bounded tx). 워커는 BATCH_SIZE 로 drain.
+PAGE = 5000
 
 
 async def run(dry_run: bool) -> None:
@@ -55,13 +55,17 @@ async def run(dry_run: bool) -> None:
         print("[dry-run] 적재 안 함. 실제 실행은 --dry-run 제거.")
         return
 
-    async with async_session() as session:
-        n = await backfill_publish_cards(session, after_id=0, limit=PAGE)
-        await session.commit()
+    total = 0
+    after = 0
+    while True:
+        async with async_session() as session:
+            n, after = await backfill_publish_cards(session, after_id=after, limit=PAGE)
+            await session.commit()
+        total += n
+        if n < PAGE:
+            break
 
-    print(f"\n[ok] outbox 적재 {n}건 — 발행 워커가 drain(flag on 시) 하며 rev 부여.")
-    if n >= PAGE:
-        print(f"[warn] PAGE({PAGE}) 도달 — 카드가 더 있을 수 있음. after_id 페이징 추가 필요.")
+    print(f"\n[ok] outbox 적재 {total}건 — 발행 워커가 drain(flag on 시) 하며 rev 부여.")
 
 
 def main() -> None:

scripts/backfill_publish_topics.py

@@ -37,7 +37,7 @@ from core.database import async_session
 from models.study_topic import StudyTopic
 from services.study.publish_enqueue import backfill_publish_topics
 
-# 개인 학습툴 주제 수 대비 넉넉. 도달 시 overflow 가드가 경보.
+# 페이지 배치 크기 — after_id 루프로 전량 처리(bounded tx).
 PAGE = 5000
 
 
@@ -58,13 +58,17 @@ async def run(dry_run: bool) -> None:
         print("[dry-run] 적재 안 함. 실제 실행은 --dry-run 제거.")
         return
 
-    async with async_session() as session:
-        n = await backfill_publish_topics(session, after_id=0, limit=PAGE)
-        await session.commit()
+    total = 0
+    after = 0
+    while True:
+        async with async_session() as session:
+            n, after = await backfill_publish_topics(session, after_id=after, limit=PAGE)
+            await session.commit()
+        total += n
+        if n < PAGE:
+            break
 
-    print(f"\n[ok] outbox 적재 {n}건 — 발행 워커가 drain(flag on 시) 하며 rev 부여.")
-    if n >= PAGE:
-        print(f"[warn] PAGE({PAGE}) 도달 — 주제가 더 있을 수 있음. after_id 페이징 추가 필요.")
+    print(f"\n[ok] outbox 적재 {total}건 — 발행 워커가 drain(flag on 시) 하며 rev 부여.")
 
 
 def main() -> None:

scripts/ci/boot_smoke.py

@@ -106,7 +106,7 @@ async def main() -> None:
             "SELECT count(*) FROM pg_indexes WHERE indexname='uq_attempt_session_question'"))).scalar()
         mx = (await s.execute(text("SELECT max(version) FROM schema_migrations"))).scalar()
     print(f"SCHEMA OK — max_migration={mx} documents={docs} purge_col={purge} cand_qwen={cand} attempt_uq={uq}")
-    assert docs and purge == 1 and cand == 0 and uq == 1 and mx == 361, "FAIL: 기대 스키마 상태 불일치"
+    assert docs and purge == 1 and cand == 0 and uq == 1 and mx == 378, "FAIL: 기대 스키마 상태 불일치"
 
     # ── 5) /health 직접 호출 ──────────────────────────────────────────────
     health = await main.health_check()

scripts/guide_clause_persist.py

@@ -0,0 +1,100 @@
+#!/usr/bin/env python3
+"""기술지침(KOSHA guide) 절-KB persist: 번호섹션(# 1. 목적 / ## 4.1) 단위 분해 + 제본.
+ASME/법령과 동일 clause-KB 모델(doc_kind='clause', parent_id=지침, 검색제외, /book 리더 공용).
+Usage: python3 guide_clause_persist.py <id|all> [--commit]
+"""
+import asyncio, os, re, sys, hashlib, statistics
+
+CAP = 12000; PAGE_TOK = 11000
+EN, KO = 0.217, 0.529
+# 번호섹션 헤더: '# 1. 목 적', '## 4.1 누출...'  (번호 1~3자리=연도(4자리) 배제)
+ART_RE = re.compile(r'^#{1,6}\s*(\d{1,3}(?:\.\d{1,3})*)\.?\s+(\S.*)$')
+TOP_RE = re.compile(r'^\d{1,3}$')
+# 외부 표준/법규 참조(대부분 dangling): ASME B16.5 · KS B 1501 · 규칙 제N조
+EXT_RE = re.compile(r'(ASME\s+[A-Z][0-9.]+|KS\s+[A-Z]\s*[0-9]+|ISO\s+[0-9]+|제\d+조)')
+
+def tok(s):
+    ko = sum(1 for c in s if '가' <= c <= '힣'); return int((len(s)-ko)*EN + ko*KO)
+
+def build_sections(text):
+    lines = text.split('\n'); off = []; a = 0
+    for ln in lines: off.append(a); a += len(ln) + 1
+    bounds = []; seen = set()
+    for i, ln in enumerate(lines):
+        m = ART_RE.match(ln)
+        if not m: continue
+        code, name = m.group(1), m.group(2).strip()
+        if not TOP_RE.match(code): continue       # top-level 번호섹션만 경계
+        if code in seen: continue
+        if len(name) < 1: continue
+        seen.add(code); bounds.append((off[i], code, name))
+    out = []
+    for idx, (start, code, name) in enumerate(bounds):
+        end = bounds[idx+1][0] if idx+1 < len(bounds) else len(text)
+        body = text[start:end].strip()
+        ext = sorted(set(EXT_RE.findall(body)))[:8]
+        out.append(dict(code=code, part='본문', order=0, title=f"{code}. {name}"[:120],
+                        body=body, tok=tok(body), links=[], ext=ext))
+    # over-CAP 페이지네이션 + 순번
+    final, order = [], 0
+    for c in out:
+        if c['tok'] <= CAP:
+            final.append({**c, 'order': order}); order += 1; continue
+        pages, cur, ct = [], [], 0
+        for ln in c['body'].split('\n'):
+            lt = tok(ln)+1
+            if ct+lt > PAGE_TOK and cur: pages.append('\n'.join(cur)); cur=[ln]; ct=lt
+            else: cur.append(ln); ct+=lt
+        if cur: pages.append('\n'.join(cur))
+        for pi, pb in enumerate(pages):
+            final.append(dict(code=c['code'] if pi==0 else f"{c['code']}·p{pi+1}", part='본문',
+                              order=order, title=c['title'] if pi==0 else f"{c['title']} (p{pi+1})",
+                              body=pb, tok=tok(pb), links=[], ext=[]))
+            order += 1
+    return final
+
+async def process_one(conn, gid, commit, verbose=True):
+    row = await conn.fetchrow("SELECT title, md_content, ai_domain, data_origin FROM documents WHERE id=$1", gid)
+    if not row: return ('notfound', 0)
+    if not row['md_content']: return ('nullmd', 0)
+    secs = build_sections(row['md_content'])
+    if len(secs) < 2: return ('few', len(secs))     # 섹션 2 미만 = 번호구조 아님
+    toks = [c['tok'] for c in secs]
+    if verbose:
+        print(f"guide={gid} «{(row['title'] or '')[:40]}» 섹션={len(secs)} median={int(statistics.median(toks))} max={max(toks)}")
+        print("  샘플:", [c['title'][:26] for c in secs[:7]])
+    if not commit: return ('dry', len(secs))
+    async with conn.transaction():
+        await conn.execute("DELETE FROM clause_links WHERE src_doc_id IN (SELECT id FROM documents WHERE parent_id=$1 AND doc_kind='clause')", gid)
+        await conn.execute("DELETE FROM documents WHERE parent_id=$1 AND doc_kind='clause'", gid)
+        for c in secs:
+            fh = hashlib.sha256(f"{gid}:{c['code']}:{c['body']}".encode()).hexdigest()
+            cid = await conn.fetchval("""
+                INSERT INTO documents (file_format,file_hash,title,md_content,parent_id,doc_kind,
+                  clause_code,clause_part,clause_order,ai_domain,data_origin,
+                  md_status,review_status,conversion_status,preview_status)
+                VALUES ('md',$1,$2,$3,$4,'clause',$5,$6,$7,$8,$9,'success','approved','none','none') RETURNING id
+            """, fh, c['title'], c['body'], gid, c['code'], c['part'], c['order'], row['ai_domain'], row['data_origin'] or 'external')
+            await conn.execute("INSERT INTO document_tags(doc_id,tag,tag_kind) VALUES ($1,'기술지침','kind') ON CONFLICT DO NOTHING", cid)
+        n = await conn.fetchval("SELECT count(*) FROM documents WHERE parent_id=$1 AND doc_kind='clause'", gid)
+        print(f"  COMMITTED: {n} 섹션 for guide {gid}")
+    return ('committed', len(secs))
+
+async def main():
+    import asyncpg
+    arg = sys.argv[1]; commit = '--commit' in sys.argv
+    conn = await asyncpg.connect(os.environ['DATABASE_URL'].replace('+asyncpg', ''))
+    if arg == 'all':
+        gs = await conn.fetch("SELECT id FROM documents WHERE material_type='guide' AND doc_kind='standard' "
+                              "AND deleted_at IS NULL AND md_content IS NOT NULL ORDER BY id")
+        agg = {}; tot = 0
+        for i, r in enumerate(gs):
+            st, n = await process_one(conn, r['id'], commit, verbose=False)
+            agg[st] = agg.get(st, 0)+1; tot += n if st in ('dry','committed') else 0
+            if commit and (i+1) % 40 == 0: print(f"  …{i+1}/{len(gs)} (누적섹션 {tot})")
+        print(f"BATCH {'COMMIT' if commit else 'DRY'} guides={len(gs)} status={agg} 총섹션={tot}")
+    else:
+        await process_one(conn, int(arg), commit, verbose=True)
+    await conn.close()
+
+asyncio.run(main())

scripts/law_clause_persist.py

@@ -0,0 +1,146 @@
+#!/usr/bin/env python3
+"""법령 조-KB persist: 법령을 조(條) 단위 개별 문서로 분해 + 조↔조 백링크 + 장(章) 태그.
+ASME clause-KB와 동일 모델(doc_kind='clause', parent_id=법령, embedding NULL, 검색제외).
+법령 추출 노이즈(조 앞 ### 메타 반복) 트림. Usage: python3 law_clause_persist.py <law_id> [--commit]
+"""
+import asyncio, os, re, sys, hashlib, statistics
+
+CAP = 12000; PAGE_TOK = 11000
+EN, KO = 0.217, 0.529
+# 조 헤더: '### 제3조의2(가스안전관리...) 본문'
+ART_RE = re.compile(r'^#{0,6}\s*(제\d+조(?:의\d+)?)\s*\(([^)]*)\)\s*(.*)$')
+CHAP_RE = re.compile(r'^#{1,6}\s*(제\d+장(?:의\d+)?)\s*(.*)$')         # 장 = part
+# 같은-법 조 멘션(백링크)
+MENTION_RE = re.compile(r'제\d+조(?:의\d+)?')
+# 타법 참조: 「법명」 ... 제N조
+EXTLAW_RE = re.compile(r'「([^」]+)」')
+
+def tok(s):
+    ko = sum(1 for c in s if '가' <= c <= '힣'); return int((len(s)-ko)*EN + ko*KO)
+def art_code(c): return c  # '제3조의2'
+
+def build_articles(text):
+    lines = text.split('\n'); off = []; a = 0
+    for ln in lines: off.append(a); a += len(ln) + 1
+    arts = []      # (line_idx, code, name, part)
+    cur_part = None
+    for i, ln in enumerate(lines):
+        ch = CHAP_RE.match(ln)
+        if ch and not ART_RE.match(ln):
+            cur_part = (ch.group(1) + (' ' + ch.group(2).strip() if ch.group(2).strip() else '')).strip()
+            continue
+        m = ART_RE.match(ln)
+        if m:
+            arts.append((i, m.group(1), m.group(2).strip(), cur_part))
+    # 본문 슬라이스 + 다음 조 앞 메타 노이즈 트림
+    out = []
+    for idx, (li, code, name, part) in enumerate(arts):
+        end_li = arts[idx+1][0] if idx+1 < len(arts) else len(lines)
+        body_lines = lines[li:end_li]
+        # 트림: 끝에서부터 '### {짧은 메타}' (조번호/조문/날짜/제목, [개정] 제N조 아님) 제거
+        while len(body_lines) > 1:
+            last = body_lines[-1].strip()
+            if last == '':
+                body_lines.pop(); continue
+            mh = re.match(r'^#{1,6}\s+(.*)$', last)
+            if mh:
+                c = mh.group(1).strip()
+                if not c.startswith('[') and not c.startswith('제') and (
+                        c in ('조문', 'N') or re.fullmatch(r'\d+', c) or re.fullmatch(r'\d{8}', c) or len(c) <= 30):
+                    body_lines.pop(); continue
+            break
+        body = '\n'.join(body_lines).strip()
+        links = sorted(set(MENTION_RE.findall(body)) - {code})
+        ext = sorted(set(EXTLAW_RE.findall(body)))[:6]
+        out.append(dict(code=code, part=part or '본칙', order=0,
+                        title=f"{code}({name})" if name else code,
+                        body=body, tok=tok(body), links=links, ext=ext))
+    # 페이지네이션(over-CAP) + 순번
+    final, order = [], 0
+    for c in out:
+        if c['tok'] <= CAP:
+            final.append({**c, 'order': order}); order += 1; continue
+        # 11K 토큰 라인 단위 분할
+        pages, cur, ct = [], [], 0
+        for ln in c['body'].split('\n'):
+            lt = tok(ln)+1
+            if ct+lt > PAGE_TOK and cur: pages.append('\n'.join(cur)); cur=[ln]; ct=lt
+            else: cur.append(ln); ct+=lt
+        if cur: pages.append('\n'.join(cur))
+        for pi, pb in enumerate(pages):
+            final.append(dict(code=c['code'] if pi==0 else f"{c['code']}·p{pi+1}", part=c['part'],
+                              order=order, title=c['title'] if pi==0 else f"{c['title']} (p{pi+1}/{len(pages)})",
+                              body=pb, tok=tok(pb), links=c['links'] if pi==0 else [], ext=[]))
+            order += 1
+    return final
+
+async def process_one(conn, law, commit, verbose=True):
+    row = await conn.fetchrow("SELECT title, coalesce(md_content, extracted_text) AS md_content, ai_domain, data_origin FROM documents WHERE id=$1", law)
+    if not row: return ('notfound', 0, 0)
+    if not row['md_content']: return ('nullmd', 0, 0)
+    arts = build_articles(row['md_content'])
+    if not arts: return ('noart', 0, 0)
+    toks = [c['tok'] for c in arts]
+    nlink = sum(len(c['links']) for c in arts)
+    if verbose:
+        parts = {}
+        for c in arts: parts[c['part']] = parts.get(c['part'], 0)+1
+        print(f"law={law} «{(row['title'] or '')[:34]}» 조문={len(arts)} median={int(statistics.median(toks))} "
+              f"max={max(toks)} 장={len(parts)} 백링크={nlink}")
+        print("  샘플:", [c['title'][:22] for c in arts[:6]])
+    if not commit:
+        return ('dry', len(arts), nlink)
+    async with conn.transaction():
+        await conn.execute(
+            "DELETE FROM clause_links WHERE src_doc_id IN (SELECT id FROM documents WHERE parent_id=$1 AND doc_kind='clause')", law)
+        await conn.execute("DELETE FROM documents WHERE parent_id=$1 AND doc_kind='clause'", law)
+        code2id = {}
+        for c in arts:
+            fh = hashlib.sha256(f"{law}:{c['code']}:{c['body']}".encode()).hexdigest()
+            cid = await conn.fetchval("""
+                INSERT INTO documents (file_format,file_hash,title,md_content,parent_id,doc_kind,
+                  clause_code,clause_part,clause_order,ai_domain,data_origin,
+                  md_status,review_status,conversion_status,preview_status)
+                VALUES ('md',$1,$2,$3,$4,'clause',$5,$6,$7,$8,$9,'success','approved','none','none') RETURNING id
+            """, fh, c['title'], c['body'], law, c['code'], c['part'], c['order'],
+                 row['ai_domain'], row['data_origin'] or 'external')
+            code2id[c['code']] = cid
+            await conn.execute("INSERT INTO document_tags(doc_id,tag,tag_kind) VALUES ($1,$2,'chapter') ON CONFLICT DO NOTHING", cid, c['part'])
+        # 조↔조 백링크 (같은 법 내부; 타법 참조는 dangling)
+        edges = []
+        for c in arts:
+            src = code2id[c['code']]
+            for dst in c['links']:
+                edges.append((src, dst, code2id.get(dst), None, None, None))
+        if edges:
+            await conn.executemany(
+                "INSERT INTO clause_links(src_doc_id,dst_code,dst_doc_id,anchor,ctx,char_off) VALUES ($1,$2,$3,$4,$5,$6)", edges)
+        n = await conn.fetchval("SELECT count(*) FROM documents WHERE parent_id=$1 AND doc_kind='clause'", law)
+        print(f"  COMMITTED: {n} 조문 + {len(edges)} 백링크 for law {law}")
+    return ('committed', n, len(edges))
+
+
+async def main():
+    import asyncpg
+    arg = sys.argv[1]; commit = '--commit' in sys.argv
+    conn = await asyncpg.connect(os.environ['DATABASE_URL'].replace('+asyncpg', ''))
+    if arg == 'all':
+        laws = await conn.fetch("SELECT lm.document_id AS id FROM legal_meta lm "
+                                "JOIN documents d ON d.id=lm.document_id "
+                                "WHERE lm.law_doc_kind='primary' AND lm.version_status='current' "
+                                "AND coalesce(d.md_content, d.extracted_text) IS NOT NULL "
+                                "ORDER BY lm.document_id")
+        agg = {}; tot_art = tot_link = 0; zero = []
+        for i, r in enumerate(laws):
+            st, na, nl = await process_one(conn, r['id'], commit, verbose=False)
+            agg[st] = agg.get(st, 0) + 1
+            tot_art += na; tot_link += nl
+            if st == 'noart': zero.append(r['id'])
+            if commit and (i + 1) % 30 == 0: print(f"  …{i+1}/{len(laws)} (누적 조 {tot_art})")
+        print(f"BATCH {'COMMIT' if commit else 'DRY'} laws={len(laws)} status={agg} 총조문={tot_art} 총백링크={tot_link}")
+        if zero: print(f"  0-조(추출구조 이질) {len(zero)}건: {zero[:20]}")
+    else:
+        await process_one(conn, int(arg), commit, verbose=True)
+    await conn.close()
+
+asyncio.run(main())

scripts/paper_citation_analyze.py

@@ -0,0 +1,51 @@
+#!/usr/bin/env python3
+"""논문 인용그래프 가능성 측정(read-only) — 본문 DOI로 코퍼스내 인용 엣지 추정.
+own_doi = 헤더(앞 2500자) 첫 DOI / cited = References 이후(또는 전체) DOI. owner 맵 → 엣지.
+"""
+import asyncio, os, re, sys
+
+DOI_RE = re.compile(r'10\.\d{4,9}/[^\s"<>)\]\},;]+')
+REF_RE = re.compile(r'(references|참고문헌|bibliography|reference\s*list)', re.I)
+
+def norm(d): return d.rstrip('.').lower()
+
+async def main():
+    import asyncpg
+    conn = await asyncpg.connect(os.environ['DATABASE_URL'].replace('+asyncpg', ''))
+    rows = await conn.fetch("SELECT id, title, coalesce(md_content, extracted_text) AS txt FROM documents "
+                            "WHERE material_type='paper' AND doc_kind='standard' AND deleted_at IS NULL "
+                            "AND coalesce(md_content, extracted_text) IS NOT NULL")
+    owner = {}        # doi -> paper id (헤더 DOI = 그 논문 소유)
+    cited = {}        # paper id -> set(cited doi)
+    n_own = n_refsec = 0
+    for r in rows:
+        txt = r['txt']
+        head = txt[:2500]
+        hdois = [norm(d) for d in DOI_RE.findall(head)]
+        if hdois:
+            owner.setdefault(hdois[0], r['id']); n_own += 1
+        m = REF_RE.search(txt)
+        body = txt[m.start():] if m else ''
+        if m: n_refsec += 1
+        cds = set(norm(d) for d in DOI_RE.findall(body))
+        if cds: cited[r['id']] = cds
+    # 엣지: paper -> owner(cited doi)
+    edges = []
+    for pid, cds in cited.items():
+        for d in cds:
+            o = owner.get(d)
+            if o and o != pid: edges.append((pid, o, d))
+    cited_papers = set(e[0] for e in edges)
+    target_papers = set(e[1] for e in edges)
+    print(f"papers={len(rows)} 헤더DOI보유={n_own} References보유={n_refsec} owner_map={len(owner)}")
+    print(f"인용엣지(코퍼스내)={len(edges)} 인용하는논문={len(cited_papers)} 피인용논문={len(target_papers)}")
+    # 피인용 top
+    from collections import Counter
+    top = Counter(e[1] for e in edges).most_common(6)
+    if top:
+        idmap = {r['id']: r['title'] for r in rows}
+        print("피인용 top:")
+        for pid, c in top: print(f"  {c}회 ← {(idmap.get(pid) or '')[:48]}")
+    await conn.close()
+
+asyncio.run(main())

scripts/paper_openalex_match.py

@@ -0,0 +1,39 @@
+#!/usr/bin/env python3
+"""OpenAlex 고신뢰 매치율 측정 — References 보유 논문(학술 추정) 표본."""
+import asyncio, os, re
+
+def toks(s):
+    return set(re.findall(r'[a-z0-9]+', (s or '').lower()))
+def sim(a, b):
+    ta, tb = toks(a), toks(b)
+    if not ta or not tb: return 0.0
+    return len(ta & tb) / len(ta | tb)
+
+async def main():
+    import asyncpg, httpx
+    conn = await asyncpg.connect(os.environ['DATABASE_URL'].replace('+asyncpg', ''))
+    rows = await conn.fetch("SELECT id, title FROM documents WHERE material_type='paper' "
+                            "AND doc_kind='standard' AND deleted_at IS NULL AND title IS NOT NULL "
+                            "AND coalesce(md_content,extracted_text) ~* 'references|참고문헌' "
+                            "ORDER BY id LIMIT 40")
+    hi = mid = lo = 0; hits = []
+    async with httpx.AsyncClient(timeout=20) as client:
+        for r in rows:
+            title = re.sub(r'\s+', ' ', r['title']).strip()
+            try:
+                resp = await client.get("https://api.openalex.org/works",
+                    params={"search": title[:200], "per_page": 1, "mailto": "hyun49196@gmail.com"})
+                res = (resp.json().get("results") or [])
+                if not res: lo += 1; continue
+                s = sim(title, res[0].get("title"))
+                if s >= 0.6: hi += 1; hits.append((s, title[:40], (res[0].get('title') or '')[:40], res[0].get('cited_by_count'), len(res[0].get('referenced_works') or [])))
+                elif s >= 0.4: mid += 1
+                else: lo += 1
+            except Exception: lo += 1
+    print(f"표본={len(rows)} 고신뢰(≥0.6)={hi} 중간(0.4~0.6)={mid} 저신뢰/무매치={lo}")
+    print("고신뢰 매치 샘플:")
+    for s, a, b, cb, rf in hits[:8]:
+        print(f"  sim={s:.2f} cited={cb} refs={rf} | {a} ≈ {b}")
+    await conn.close()
+
+asyncio.run(main())

scripts/paper_openalex_test.py

@@ -0,0 +1,30 @@
+#!/usr/bin/env python3
+"""OpenAlex 보강 타당성 테스트 — 소수 논문 제목으로 매칭/메타 확인 (외부 API)."""
+import asyncio, os, re
+
+async def main():
+    import asyncpg, httpx
+    conn = await asyncpg.connect(os.environ['DATABASE_URL'].replace('+asyncpg', ''))
+    rows = await conn.fetch("SELECT id, title FROM documents WHERE material_type='paper' "
+                            "AND doc_kind='standard' AND deleted_at IS NULL AND title IS NOT NULL "
+                            "AND length(title) > 15 ORDER BY id LIMIT 6")
+    async with httpx.AsyncClient(timeout=20) as client:
+        for r in rows:
+            title = re.sub(r'\s+', ' ', r['title']).strip()
+            try:
+                resp = await client.get("https://api.openalex.org/works",
+                    params={"search": title[:200], "per_page": 1, "mailto": "hyun49196@gmail.com"})
+                js = resp.json()
+                res = (js.get("results") or [])
+                if not res:
+                    print(f"[{r['id']}] NO MATCH | {title[:50]}"); continue
+                w = res[0]
+                oid = (w.get("id") or "").split("/")[-1]
+                print(f"[{r['id']}] {title[:46]}")
+                print(f"   → OA {oid} | {(w.get('title') or '')[:46]} | {w.get('publication_year')} | "
+                      f"cited_by={w.get('cited_by_count')} | refs={len(w.get('referenced_works') or [])} | doi={w.get('doi')}")
+            except Exception as e:
+                print(f"[{r['id']}] ERROR {type(e).__name__}: {e}")
+    await conn.close()
+
+asyncio.run(main())

services/mineru/server.py

@@ -59,6 +59,11 @@ MAX_IMAGES_PER_DOC = int(os.getenv("MINERU_MAX_IMAGES_PER_DOC", "200"))
 MAX_BYTES_PER_IMAGE = int(os.getenv("MINERU_MAX_BYTES_PER_IMAGE", str(10 * 1024 * 1024)))
 MAX_PAGES_HARD = int(os.getenv("MINERU_MAX_PAGES_HARD", "200"))  # 1-shot max_pages 안전장치
 
+# self-timeout — 변환/워밍이 vLLM 행으로 _engine_lock 을 영구 점유해 서비스가 wedge 되는 것을 차단.
+# (클라이언트 marker_worker 는 300s 로 포기하나 서버측 inflight 는 자동 취소 안 됨 → 서버 자체 상한 필요.)
+PARSE_TIMEOUT_S = float(os.getenv("MINERU_PARSE_TIMEOUT_S", "600"))
+WARMUP_TIMEOUT_S = float(os.getenv("MINERU_WARMUP_TIMEOUT_S", "1200"))  # 최초 모델 다운로드(~2.4GB) 여유
+
 _PRELOAD = os.getenv("MINERU_PRELOAD", "1") != "0"
 
 # ---- 엔진 상태 ---------------------------------------------------------------
@@ -68,6 +73,15 @@ _warmup_error: str | None = None
 _engine_lock = asyncio.Lock()
 
 
+def _is_engine_fatal(exc: BaseException) -> bool:
+    """OOM/CUDA 류 = 엔진 상태 오염 가능 → 재워밍 강제 대상(타임아웃은 호출측에서 별도 판정)."""
+    s = f"{type(exc).__name__} {exc}".lower()
+    return any(
+        k in s
+        for k in ("out of memory", "oom", "cuda", "cublas", "device-side", "illegal memory")
+    )
+
+
 async def _run_mineru(pdf_bytes: bytes, lang: str) -> tuple[str, list[dict]]:
     """슬라이스된 PDF bytes → (markdown, 이미지 dict 리스트). **async 엔진 경로.**
 
@@ -148,7 +162,7 @@ async def _ensure_warmup() -> None:
             page.insert_text((72, 72), "MinerU warmup.")
             warmup_bytes = doc.tobytes()
             doc.close()
-            await _run_mineru(warmup_bytes, MINERU_LANG)
+            await asyncio.wait_for(_run_mineru(warmup_bytes, MINERU_LANG), timeout=WARMUP_TIMEOUT_S)
             _warmup_done = True
             _warmup_error = None
             logger.info(f"[mineru-service] warmup done engine_version={_engine_version}")
@@ -274,6 +288,7 @@ def _serialize_images(images: list[dict], src_path: str) -> tuple[list[ConvertIm
 
 @app.post("/convert", response_model=ConvertResponse)
 async def convert(req: ConvertRequest):
+    global _warmup_done
     p = _resolve_path(req.file_path)
     if p is None or not p.is_file():
         raise HTTPException(404, detail={"code": "file_not_found", "message": req.file_path})
@@ -288,10 +303,18 @@ async def convert(req: ConvertRequest):
     async with _engine_lock:                # 실제 변환 직렬화(단일 GPU)
         start = time.monotonic()
         try:
-            md_text, raw_images = await _run_mineru(pdf_bytes, MINERU_LANG)
+            md_text, raw_images = await asyncio.wait_for(
+                _run_mineru(pdf_bytes, MINERU_LANG), timeout=PARSE_TIMEOUT_S
+            )
         except HTTPException:
             raise
         except Exception as exc:
+            # 타임아웃(엔진 행) 또는 OOM/CUDA 류면 엔진 오염 가능 → 다음 요청이 재워밍하도록 리셋.
+            # 재워밍까지 실패하면 _ensure_warmup 이 _warmup_error 설정 → /ready 503 → healthcheck
+            # 재시작으로 escalate(영구 degradation 차단). 일시 OOM 이면 재워밍 성공 후 정상화.
+            if isinstance(exc, (asyncio.TimeoutError, TimeoutError)) or _is_engine_fatal(exc):
+                _warmup_done = False
+                logger.error("[mineru-service] engine reset (timeout/fatal) path=%s: %s", p, exc)
             logger.exception(f"[mineru-service] conversion failed path={p}: {exc}")
             raise HTTPException(422, detail={"code": "conversion_failed",
                 "message": f"{type(exc).__name__}: {exc}"}) from exc