框架: A2收口——三域20表按主仓DDL原样建成(V1/V3/V5全文+V9V10V14V26V30触表ALTER摘录)+example_*四表(嵌入边表vector1024+HNSW);db skill交付(psycopg直连/卡片打印/文件事务);表映射与暂缓登记落档;README拍板台账①-④全拍+基座状态刷新

.claude/skills/db/SKILL.md

@@ -0,0 +1,38 @@
+---
+name: db
+description: muse-example 实验库的唯一数据库通道——查询/DML/DDL/SQL 文件应用全走 scripts/db.py（psycopg 直连），SELECT 结果卡片式打印即审查面。主会话与智能体要读写 PG 一律经此，不得裸连或散写一次性脚本。
+---
+
+# db —— muse-example 唯一数据库通道
+
+对应 muse API 面：数据访问层。连接事实与凭据见 [`db/连接信息.md`](../../../db/连接信息.md)（DSN 已锁死在脚本内，只连 `muse-example`）。
+
+## 用法（仓库根目录执行，python 一律用 `.venv/bin/python`）
+
+```bash
+# 查询：卡片式打印（默认最多 50 行、长值截 160 字）
+.venv/bin/python .claude/skills/db/scripts/db.py query "SELECT id,title FROM muse_content_work"
+.venv/bin/python .claude/skills/db/scripts/db.py query "SELECT ..." --json      # JSON 数组输出（给脚本消费）
+.venv/bin/python .claude/skills/db/scripts/db.py query "SELECT ..." --full      # 长值不截断
+.venv/bin/python .claude/skills/db/scripts/db.py query "SELECT ..." --max 200   # 放宽行数
+
+# 单条写操作（INSERT/UPDATE/DELETE/DDL）：报影响行数
+.venv/bin/python .claude/skills/db/scripts/db.py exec "UPDATE ... WHERE ..."
+
+# SQL 文件应用：整文件一个事务，失败全回滚
+.venv/bin/python .claude/skills/db/scripts/db.py apply db/ddl/91-example实验私货.sql
+
+# 表清单+活行数（deleted=FALSE 计数，无 deleted 列的表计全行）
+.venv/bin/python .claude/skills/db/scripts/db.py tables
+```
+
+## 红线
+
+- **只连 `muse-example`**：DSN 硬编码锁库；严禁改造脚本去碰共享 PG 上的 muse_local / muse_slice_live / *_test。
+- 软删约定照主仓：删除=UPDATE `deleted=TRUE`，不物理删（example_* 表同样遵守）。
+- 批量导入/嵌入等专用写路径由 import/embed skill 封装（内部同走 psycopg 直连），本 skill 承担通用查改与 DDL 应用。
+- 建表/改表先落 `db/ddl/` 文件再 `apply`，不敲一次性 DDL——文件即审计。
+
+## 元数据驱动
+
+库内 `muse_meta_schema` / `muse_meta_field` / `muse_meta_visibility_policy` 是 23 型结构本体的权威行（A3 起）；拆书与抽取的字段合同一律 `query` 读库取，不再读 `meta/schemas/` YAML（YAML 退为设计稿与种子来源）。

.claude/skills/db/scripts/db.py

@@ -0,0 +1,106 @@
+#!/usr/bin/env python3
+"""muse-example 唯一数据库通道（db skill 脚本层）。
+
+- DSN 锁死 muse-example：严禁触碰共享 PG 上其他库（muse_local / muse_slice_live / *_test）。
+- query 卡片式打印=审查面；exec 报影响行数；apply 整文件一个事务失败全回滚。
+- 失败原样抛错不静默（公约）。
+"""
+import json
+import sys
+
+import click
+import psycopg
+
+# 连接事实与凭据来源：db/连接信息.md（内网 Tailscale 段，凭据明文入仓为既定政策）
+DSN = "postgresql://root:f6710e2d0294eb1c10e26a805a64bc54@100.64.0.8:5433/muse-example"
+TRUNC = 160  # 卡片模式长值截断阈值（字符）
+
+
+def _fmt(value, full: bool) -> str:
+    """卡片值格式化：NULL 显示 ∅；长值默认截断并标注总长。"""
+    if value is None:
+        return "∅"
+    s = str(value)
+    if not full and len(s) > TRUNC:
+        return s[:TRUNC] + f"…(共{len(s)}字)"
+    return s
+
+
+@click.group()
+def cli():
+    """muse-example 数据库通道"""
+
+
+@cli.command()
+@click.argument("sql_text")
+@click.option("--json", "as_json", is_flag=True, help="按 JSON 数组输出（给脚本消费）")
+@click.option("--full", is_flag=True, help="长值不截断")
+@click.option("--max", "max_rows", default=50, show_default=True, help="最多打印行数")
+def query(sql_text: str, as_json: bool, full: bool, max_rows: int):
+    """执行 SELECT，卡片式打印（审查面）。"""
+    with psycopg.connect(DSN) as conn:
+        cur = conn.execute(sql_text)
+        if cur.description is None:
+            raise click.ClickException("该语句无结果集——写操作请用 exec")
+        cols = [d.name for d in cur.description]
+        rows = cur.fetchall()
+    if as_json:
+        out = [dict(zip(cols, r)) for r in rows[:max_rows]]
+        click.echo(json.dumps(out, ensure_ascii=False, default=str, indent=1))
+    else:
+        width = max((len(c) for c in cols), default=0)
+        for i, r in enumerate(rows[:max_rows], 1):
+            click.echo(f"── 行 {i} " + "─" * 30)
+            for c, v in zip(cols, r):
+                click.echo(f"{c:<{width}} │ {_fmt(v, full)}")
+    if len(rows) > max_rows:
+        click.echo(f"…（共 {len(rows)} 行，仅示 {max_rows} 行，--max 放宽）")
+    else:
+        click.echo(f"（共 {len(rows)} 行）")
+
+
+@cli.command("exec")
+@click.argument("sql_text")
+def exec_(sql_text: str):
+    """执行单条写操作（INSERT/UPDATE/DELETE/DDL），报影响行数。"""
+    with psycopg.connect(DSN) as conn:
+        cur = conn.execute(sql_text)
+        conn.commit()
+        n = cur.rowcount
+    click.echo(f"OK（影响 {n} 行）" if n >= 0 else "OK")
+
+
+@cli.command()
+@click.argument("file", type=click.Path(exists=True))
+def apply(file: str):
+    """应用 SQL 文件：整文件一个事务（无参数走 simple protocol 支持多语句），失败全回滚。"""
+    with open(file, encoding="utf-8") as f:
+        sql = f.read()
+    with psycopg.connect(DSN) as conn:
+        conn.execute(sql)
+        conn.commit()
+    click.echo(f"已应用: {file}")
+
+
+@cli.command()
+def tables():
+    """表清单 + 活行数（有 deleted 列的表计 deleted=FALSE，其余计全行）。"""
+    with psycopg.connect(DSN) as conn:
+        names = [r[0] for r in conn.execute(
+            "SELECT tablename FROM pg_tables WHERE schemaname='public' ORDER BY tablename")]
+        for t in names:
+            has_deleted = conn.execute(
+                "SELECT 1 FROM information_schema.columns WHERE table_name=%s AND column_name='deleted'",
+                (t,)).fetchone()
+            where = " WHERE deleted = FALSE" if has_deleted else ""
+            n = conn.execute(f'SELECT count(*) FROM "{t}"{where}').fetchone()[0]
+            click.echo(f"{t:<45} {n}")
+
+
+if __name__ == "__main__":
+    try:
+        cli()
+    except psycopg.Error as e:
+        # 数据库错误原样透出（可追踪、不静默）
+        click.echo(f"[db错误] {type(e).__name__}: {e}", err=True)
+        sys.exit(1)

README.md

@@ -52,7 +52,7 @@ agent-example/
 | read-context | 统一读取器：读装配绑定→按用途+aiContext 三级裁剪→组装六分区上下文包→裁剪清单回显 | 统一创作数据读取器（专题-06 §7） | 现有（文件版）；C4 升 PG 检索版（内部调 search） |
 | confirm | 唯一确认通道：确认=commit、丢弃=restore，仅创始人触发；PG 版加知识行草稿→已确认翻转 | 双轨提交 / 候选三决策入口 | 现有 |
 | eval | 质量收敛环：n=5、每轮只动一个变量、达标固化 `golden/` | 质量域 | 现有 |
-| db | `muse-example` 唯一数据库通道：ssh+容器 psql 封装（凭据按 `db/连接信息.md`），查询/DDL/DML 全走这，结果卡片式打印（=审查面） | 数据访问层 | A2 交付 |
+| db | `muse-example` 唯一数据库通道：psycopg 直连封装（凭据按 `db/连接信息.md`），查询/DDL/DML/SQL 文件应用全走这，结果卡片式打印（=审查面） | 数据访问层 | 已交付（2026-07-13） |
 | import | 导入解析：txt→回目正则静态分章→作品/章行入库（调 db） | 导入解析旅程 | B1 交付；C8 用户旧稿复用 |
 | embed | New-API 嵌入封装：Qwen3-Embedding-8B、`dimensions:1024`、`--noproxy`、批量+失败重试 | AI 网关（嵌入） | B2 交付 |
 | search | 向量检索：意图→embed→pgvector 相似度召回→授权过滤（仅已确认+已绑定）+字段裁剪→带相似度分的结果集 | 知识检索 | B3 交付 |
@@ -218,8 +218,8 @@ flowchart LR
 ## 九、当前位置（2026-07-10 快照）
 
 - 执行主线=§七闭环序，知识域（A/B）先行；创作域（C）后置，待 B5 过门后按 C 线走。
-- **拍板台账**：①已拍 A——B2 首轮公共面只拆 5 型范式，双层型二轮凭脚手架实据定；细纲字数比例已拍入 outline schema。**待拍**：②参考书书单——封神演义因创始人未读降级为分章正则测试样本，换创始人熟读的 2–3 本（已定框架=每本全本导入分章+首轮拆 30–50 章；原文只入私有库，公共面只有脱敏范式）；③A2 口径（主仓表原样不改列+私货全进 `example_*`）；④整序确认。齐后从 A1 起跑。
+- **拍板台账（①–④已全拍）**：①B2 首轮公共面只拆 5 型范式，双层型二轮凭脚手架实据定；细纲字数比例已拍入 outline schema。②已拍（2026-07-10）——参考书=`../小说清单/` 7 本**全部导入**私有库；首轮拆书 4 本：超神机械师/机战无限/深空之影/机动风暴（每本 30–50 章；原文只入私有库，公共面只有脱敏范式）。③已拍——主仓表原样不改列+私货全进 `example_*`。④已拍——整序确认，A1 已起跑（B2 跑批前有成本确认闸）。
 - **能力就位**：skill 16 只（基建 7：read-context/confirm/eval 现有+db/import/embed/search 按 A2–B3 交付；功能 9 全建成，对齐官方规范）× agent 5（已瘦身为身份段）；prompt 三段式（身份段×功能 skill×L0）；上下文=四层+章级基线包共享前缀+按 agent 差异矩阵（read-context skill 为准）；拆书=workflow 编排+逐章 agent 有界任务（parse-book skill 为准）。
-- **基座**：本机 psycopg 直连与 psql 18.4 均验通；`muse-example` 库与 vector 插件**尚未创建**（A1 第一步）；凭据与配方在 `db/连接信息.md`；`.venv` 已建（requirements.txt）。
+- **基座**：A1/A2 已收口（2026-07-13）——`muse-example` 库+pgvector 0.8.5 验通（容器重建恢复命令见连接信息），24 表建成（主仓一致 20 + `example_*` 4，映射与暂缓登记见 `db/表映射.md`），db skill 已交付；凭据与配方在 `db/连接信息.md`；`.venv` 丢失时按 requirements.txt 重建即用。
 - **待审资产已清零（2026-07-10 创始人清理）**：文件版首轮的 13 张范式卡、参考书档案 extracted 态、《焚忆》全部文件资产已清除，工作区（干净 clone）无任何待审存量；首轮拆书的字段设计发现已蒸馏入 `meta/schemas/`（已提交），不受影响。
 - 进度惯例：看 git log 与本节拍板台账，不设过程状态文档。

db/ddl/90-主仓ALTER摘录-V9V10V14V30.sql

@@ -0,0 +1,39 @@
+-- 主仓 ALTER 摘录（只取触达 V1/V3/V5 三个 init 文件所建 20 表的语句，一字不改）
+-- 口径（创始人拍板③ 2026-07-10）：主仓表原样不改列；本文件各段出处逐一注明。
+-- 未摘录说明：V9/V10/V14 等文件中"新建表"部分（planning_section、meta 治理 8 表、
+-- knowledge RAGFlow 14 表、各 outbox 表）实验现阶段不用，暂缓建表，登记见 db/表映射.md。
+
+-- ── 出自 sql/muse/V9__extend_content_real_api_schema.sql ──
+ALTER TABLE muse_content_export_task
+    ADD COLUMN IF NOT EXISTS download_expires_at TIMESTAMP,
+    ADD COLUMN IF NOT EXISTS completed_at TIMESTAMP,
+    ADD COLUMN IF NOT EXISTS progress INT NOT NULL DEFAULT 0,
+    ADD COLUMN IF NOT EXISTS source_snapshot JSONB;
+
+-- ── 出自 sql/muse/V10__extend_meta_real_api_schema.sql ──
+CREATE UNIQUE INDEX uk_muse_meta_schema_version_active_schema_id
+    ON muse_meta_schema_version(tenant_id, schema_id)
+    WHERE active_flag = TRUE AND deleted = FALSE;
+
+ALTER TABLE muse_meta_protection_node
+    ADD COLUMN IF NOT EXISTS node_type VARCHAR(64) NOT NULL DEFAULT 'system',
+    ADD COLUMN IF NOT EXISTS irreplaceable_reason TEXT,
+    ADD COLUMN IF NOT EXISTS permission_requirement JSONB,
+    ADD COLUMN IF NOT EXISTS audit_requirement JSONB,
+    ADD COLUMN IF NOT EXISTS shadow_to_canonical_boundary JSONB,
+    ADD COLUMN IF NOT EXISTS observability_metrics JSONB;
+
+-- ── 出自 sql/muse/V14__extend_knowledge_ragflow_real_api_schema.sql ──
+ALTER TABLE muse_knowledge_draft
+    ALTER COLUMN source_snapshot_id TYPE VARCHAR(128) USING source_snapshot_id::text;
+ALTER TABLE muse_knowledge_draft
+    ALTER COLUMN authorization_snapshot_id TYPE VARCHAR(128) USING authorization_snapshot_id::text;
+ALTER TABLE muse_knowledge_binding
+    ALTER COLUMN source_snapshot_id TYPE VARCHAR(128) USING source_snapshot_id::text;
+ALTER TABLE muse_knowledge_binding
+    ALTER COLUMN authorization_snapshot_id TYPE VARCHAR(128) USING authorization_snapshot_id::text;
+
+-- ── 出自 sql/muse/V30__widen_authorization_snapshot_id_to_varchar.sql ──
+-- （同文件另有 muse_ai_suggestion 一条，该表实验未建，不摘）
+ALTER TABLE muse_content_block_source_attribution
+    ALTER COLUMN authorization_snapshot_id TYPE VARCHAR(128) USING authorization_snapshot_id::text;

db/ddl/91-example实验私货.sql

@@ -0,0 +1,96 @@
+-- 实验私货表（口径③：主仓表原样不改列，实验扩展全进 example_* 前缀）
+-- 公共列/触发器沿用主仓惯例（creator/create_time/updater/update_time/deleted/tenant_id + updated_at 触发器）；
+-- tenant_id/creator 按拍板默认系统主账号=1（写入层给值，列默认仍照主仓惯例）。
+-- 软引用无外键（同主仓约定：引用完整性由应用/skill 层保证）。
+
+-- ══ example_reference_work：参考书档案（23 型之 reference_work 的库内实例载体）══
+-- 一本参考书 = muse_content_work 一行（原文私有库） + 本表一行（档案与拆书进度）。
+CREATE TABLE example_reference_work (
+    id BIGINT GENERATED ALWAYS AS IDENTITY PRIMARY KEY,
+    work_id BIGINT NOT NULL,                          -- → muse_content_work.id（软引用）
+    author VARCHAR(100),                              -- 作者
+    source_file VARCHAR(500),                         -- 原始 txt 文件名
+    declared_chapter_count INT,                       -- 原文件头部声明章数（无则 NULL）
+    imported_chapter_count INT NOT NULL DEFAULT 0,    -- 实际导入章数
+    char_count BIGINT NOT NULL DEFAULT 0,             -- 全文字符数
+    parse_scope JSONB,                                -- 首轮拆书范围，如 {"from":1,"to":50}
+    parse_status VARCHAR(20) NOT NULL DEFAULT 'pending', -- pending/parsing/parsed
+    notes TEXT,                                       -- 清洗记录/章题格式特点等备注
+    creator VARCHAR(64) NOT NULL DEFAULT '',
+    create_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    updater VARCHAR(64) NOT NULL DEFAULT '',
+    update_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    deleted BOOLEAN NOT NULL DEFAULT FALSE,
+    tenant_id BIGINT NOT NULL DEFAULT 0,
+    CONSTRAINT uk_example_reference_work UNIQUE (tenant_id, work_id)
+);
+CREATE TRIGGER trg_example_reference_work_updated_at BEFORE UPDATE ON example_reference_work FOR EACH ROW EXECUTE FUNCTION update_updated_at_column();
+
+-- ══ example_parse_task：拆书任务状态机（workflow 断点续跑与幂等，状态全在库）══
+-- 一章一行；两阶段各自状态：scaffold=细纲逆推+实体脚手架，pattern=范式拆解。
+CREATE TABLE example_parse_task (
+    id BIGINT GENERATED ALWAYS AS IDENTITY PRIMARY KEY,
+    work_id BIGINT NOT NULL,                          -- → muse_content_work.id
+    chapter_id BIGINT NOT NULL,                       -- → muse_content_chapter.id
+    scaffold_status VARCHAR(20) NOT NULL DEFAULT 'pending', -- pending/running/done/failed
+    pattern_status VARCHAR(20) NOT NULL DEFAULT 'pending',  -- pending/running/done/failed
+    attempt_count INT NOT NULL DEFAULT 0,             -- 累计尝试次数（含重试）
+    agent_model VARCHAR(50),                          -- 执行模型标记（审计用）
+    error_message TEXT,                               -- 最近一次失败原因
+    creator VARCHAR(64) NOT NULL DEFAULT '',
+    create_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    updater VARCHAR(64) NOT NULL DEFAULT '',
+    update_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    deleted BOOLEAN NOT NULL DEFAULT FALSE,
+    tenant_id BIGINT NOT NULL DEFAULT 0,
+    CONSTRAINT uk_example_parse_task UNIQUE (tenant_id, work_id, chapter_id),
+    CONSTRAINT chk_example_parse_task_scaffold CHECK (scaffold_status IN ('pending','running','done','failed')),
+    CONSTRAINT chk_example_parse_task_pattern CHECK (pattern_status IN ('pending','running','done','failed'))
+);
+CREATE INDEX idx_example_parse_task_work ON example_parse_task(tenant_id, work_id, scaffold_status, pattern_status);
+CREATE TRIGGER trg_example_parse_task_updated_at BEFORE UPDATE ON example_parse_task FOR EACH ROW EXECUTE FUNCTION update_updated_at_column();
+
+-- ══ example_parse_scaffold：逐章脚手架（parse-book 2b 第一步产物）══
+-- 细纲字数比例按 outline schema 拍板值（章细纲=原文 3–5%）；实体清单供判重与二轮双层型决策。
+CREATE TABLE example_parse_scaffold (
+    id BIGINT GENERATED ALWAYS AS IDENTITY PRIMARY KEY,
+    work_id BIGINT NOT NULL,                          -- → muse_content_work.id
+    chapter_id BIGINT NOT NULL,                       -- → muse_content_chapter.id
+    outline_text TEXT,                                -- 逆推细纲（对原文 3–5% 字数）
+    entities JSONB,                                   -- 实体清单 [{type,name,brief}]
+    creator VARCHAR(64) NOT NULL DEFAULT '',
+    create_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    updater VARCHAR(64) NOT NULL DEFAULT '',
+    update_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    deleted BOOLEAN NOT NULL DEFAULT FALSE,
+    tenant_id BIGINT NOT NULL DEFAULT 0,
+    CONSTRAINT uk_example_parse_scaffold UNIQUE (tenant_id, chapter_id)
+);
+CREATE INDEX idx_example_parse_scaffold_work ON example_parse_scaffold(tenant_id, work_id);
+CREATE TRIGGER trg_example_parse_scaffold_updated_at BEFORE UPDATE ON example_parse_scaffold FOR EACH ROW EXECUTE FUNCTION update_updated_at_column();
+
+-- ══ example_knowledge_embedding：知识行嵌入边表（vector(1024)，Qwen3-Embedding-8B）══
+-- 挂 draft（草稿期）或 entity（确认后），至少其一非空；content_hash 幂等判重（同文同模型不重嵌）。
+CREATE TABLE example_knowledge_embedding (
+    id BIGINT GENERATED ALWAYS AS IDENTITY PRIMARY KEY,
+    draft_id BIGINT,                                  -- → muse_knowledge_draft.id（软引用）
+    entity_id BIGINT,                                 -- → muse_knowledge_entity.id（软引用）
+    content_hash CHAR(64) NOT NULL,                   -- 嵌入文本 sha256（幂等键）
+    embed_text TEXT NOT NULL,                         -- 被嵌入文本（名称+摘要+正文拼接）
+    model VARCHAR(100) NOT NULL DEFAULT 'Qwen/Qwen3-Embedding-8B',
+    dimensions INT NOT NULL DEFAULT 1024,
+    embedding vector(1024) NOT NULL,
+    creator VARCHAR(64) NOT NULL DEFAULT '',
+    create_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    updater VARCHAR(64) NOT NULL DEFAULT '',
+    update_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    deleted BOOLEAN NOT NULL DEFAULT FALSE,
+    tenant_id BIGINT NOT NULL DEFAULT 0,
+    CONSTRAINT uk_example_knowledge_embedding UNIQUE (tenant_id, content_hash, model),
+    CONSTRAINT chk_example_knowledge_embedding_ref CHECK (draft_id IS NOT NULL OR entity_id IS NOT NULL)
+);
+CREATE INDEX idx_example_knowledge_embedding_draft ON example_knowledge_embedding(tenant_id, draft_id) WHERE draft_id IS NOT NULL;
+CREATE INDEX idx_example_knowledge_embedding_entity ON example_knowledge_embedding(tenant_id, entity_id) WHERE entity_id IS NOT NULL;
+-- 余弦相似度检索索引（B3 search skill 用 <=> 召回；数据量小，HNSW 默认参数即可）
+CREATE INDEX idx_example_knowledge_embedding_hnsw ON example_knowledge_embedding USING hnsw (embedding vector_cosine_ops);
+CREATE TRIGGER trg_example_knowledge_embedding_updated_at BEFORE UPDATE ON example_knowledge_embedding FOR EACH ROW EXECUTE FUNCTION update_updated_at_column();

db/ddl/V1__init_content_schema.sql

@@ -0,0 +1,180 @@
+-- Muse Content 模块初始 Schema（PostgreSQL 16）
+-- 影响范围：作品、章节、Block、来源归因、导入导出任务。
+-- 说明：业务表不使用数据库外键，引用完整性由应用层在权限、租户和 owner 校验后保证。
+
+CREATE OR REPLACE FUNCTION update_updated_at_column()
+RETURNS TRIGGER AS $$
+BEGIN
+    NEW.update_time = CURRENT_TIMESTAMP;
+    RETURN NEW;
+END;
+$$ LANGUAGE plpgsql;
+
+CREATE TABLE muse_content_work (
+    id BIGINT GENERATED ALWAYS AS IDENTITY PRIMARY KEY,
+    owner_user_id BIGINT NOT NULL,
+    title VARCHAR(200) NOT NULL,
+    description TEXT,
+    genre VARCHAR(50),
+    cover_image_url TEXT,
+    summary TEXT,
+    status VARCHAR(20) NOT NULL DEFAULT 'writing',
+    work_schema_id BIGINT,
+    import_status VARCHAR(20),
+    parse_status VARCHAR(20),
+    word_count INT NOT NULL DEFAULT 0,
+    chapter_count INT NOT NULL DEFAULT 0,
+    command_id VARCHAR(120),
+    revision INT NOT NULL DEFAULT 1,
+    creator VARCHAR(64) NOT NULL DEFAULT '',
+    create_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    updater VARCHAR(64) NOT NULL DEFAULT '',
+    update_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    deleted BOOLEAN NOT NULL DEFAULT FALSE,
+    tenant_id BIGINT NOT NULL DEFAULT 0
+);
+
+CREATE INDEX idx_muse_content_work_owner ON muse_content_work(tenant_id, owner_user_id, update_time);
+CREATE INDEX idx_muse_content_work_status ON muse_content_work(tenant_id, status);
+CREATE UNIQUE INDEX uk_muse_content_work_command ON muse_content_work(tenant_id, command_id) WHERE command_id IS NOT NULL;
+CREATE TRIGGER trg_muse_content_work_updated_at BEFORE UPDATE ON muse_content_work FOR EACH ROW EXECUTE FUNCTION update_updated_at_column();
+
+CREATE TABLE muse_content_chapter (
+    id BIGINT GENERATED ALWAYS AS IDENTITY PRIMARY KEY,
+    work_id BIGINT NOT NULL,
+    title VARCHAR(200) NOT NULL,
+    order_no INT NOT NULL,
+    status VARCHAR(20) NOT NULL DEFAULT 'draft',
+    goal_snapshot JSONB,
+    outline_snapshot JSONB,
+    parse_review_status VARCHAR(20),
+    command_id VARCHAR(120),
+    revision INT NOT NULL DEFAULT 1,
+    creator VARCHAR(64) NOT NULL DEFAULT '',
+    create_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    updater VARCHAR(64) NOT NULL DEFAULT '',
+    update_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    deleted BOOLEAN NOT NULL DEFAULT FALSE,
+    tenant_id BIGINT NOT NULL DEFAULT 0,
+    CONSTRAINT uk_muse_content_chapter_work_order UNIQUE (tenant_id, work_id, order_no)
+);
+
+CREATE INDEX idx_muse_content_chapter_work ON muse_content_chapter(tenant_id, work_id);
+CREATE UNIQUE INDEX uk_muse_content_chapter_command ON muse_content_chapter(tenant_id, command_id) WHERE command_id IS NOT NULL;
+CREATE TRIGGER trg_muse_content_chapter_updated_at BEFORE UPDATE ON muse_content_chapter FOR EACH ROW EXECUTE FUNCTION update_updated_at_column();
+
+CREATE TABLE muse_content_block (
+    id BIGINT GENERATED ALWAYS AS IDENTITY PRIMARY KEY,
+    work_id BIGINT NOT NULL,
+    chapter_id BIGINT NOT NULL,
+    order_no INT NOT NULL,
+    block_type VARCHAR(30) NOT NULL DEFAULT 'scene',
+    title VARCHAR(500),
+    content_doc JSONB,
+    content_text TEXT,
+    revision INT NOT NULL DEFAULT 1,
+    word_count INT NOT NULL DEFAULT 0,
+    command_id VARCHAR(120),
+    creator VARCHAR(64) NOT NULL DEFAULT '',
+    create_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    updater VARCHAR(64) NOT NULL DEFAULT '',
+    update_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    deleted BOOLEAN NOT NULL DEFAULT FALSE,
+    tenant_id BIGINT NOT NULL DEFAULT 0,
+    CONSTRAINT uk_muse_content_block_chapter_order UNIQUE (tenant_id, chapter_id, order_no)
+);
+
+CREATE INDEX idx_muse_content_block_work ON muse_content_block(tenant_id, work_id);
+CREATE INDEX idx_muse_content_block_chapter ON muse_content_block(tenant_id, chapter_id);
+CREATE UNIQUE INDEX uk_muse_content_block_command ON muse_content_block(tenant_id, command_id) WHERE command_id IS NOT NULL;
+CREATE TRIGGER trg_muse_content_block_updated_at BEFORE UPDATE ON muse_content_block FOR EACH ROW EXECUTE FUNCTION update_updated_at_column();
+
+CREATE TABLE muse_content_block_source_attribution (
+    id BIGINT GENERATED ALWAYS AS IDENTITY PRIMARY KEY,
+    work_id BIGINT NOT NULL,
+    block_id BIGINT NOT NULL,
+    revision INT NOT NULL,
+    source_type VARCHAR(50) NOT NULL,
+    source_object_id VARCHAR(100),
+    source_version INT,
+    lineage_payload JSONB,
+    authorization_snapshot_id BIGINT,
+    source_status VARCHAR(30),
+    license_restriction_snapshot JSONB,
+    creator VARCHAR(64) NOT NULL DEFAULT '',
+    create_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    updater VARCHAR(64) NOT NULL DEFAULT '',
+    update_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    deleted BOOLEAN NOT NULL DEFAULT FALSE,
+    tenant_id BIGINT NOT NULL DEFAULT 0,
+    CONSTRAINT uk_muse_content_block_src_revision UNIQUE (tenant_id, block_id, revision)
+);
+
+CREATE INDEX idx_muse_content_block_src_block ON muse_content_block_source_attribution(tenant_id, block_id);
+CREATE TRIGGER trg_muse_content_block_src_updated_at BEFORE UPDATE ON muse_content_block_source_attribution FOR EACH ROW EXECUTE FUNCTION update_updated_at_column();
+
+CREATE TABLE muse_content_import_task (
+    id BIGINT GENERATED ALWAYS AS IDENTITY PRIMARY KEY,
+    work_id BIGINT NOT NULL,
+    owner_user_id BIGINT NOT NULL,
+    source_type VARCHAR(50) NOT NULL,
+    source_snapshot JSONB,
+    status VARCHAR(30) NOT NULL DEFAULT 'queued',
+    error_code VARCHAR(80),
+    error_message TEXT,
+    command_id VARCHAR(120) NOT NULL,
+    creator VARCHAR(64) NOT NULL DEFAULT '',
+    create_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    updater VARCHAR(64) NOT NULL DEFAULT '',
+    update_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    deleted BOOLEAN NOT NULL DEFAULT FALSE,
+    tenant_id BIGINT NOT NULL DEFAULT 0,
+    CONSTRAINT uk_muse_content_import_command UNIQUE (tenant_id, command_id)
+);
+
+CREATE INDEX idx_muse_content_import_work ON muse_content_import_task(tenant_id, work_id, status);
+CREATE TRIGGER trg_muse_content_import_updated_at BEFORE UPDATE ON muse_content_import_task FOR EACH ROW EXECUTE FUNCTION update_updated_at_column();
+
+CREATE TABLE muse_content_export_task (
+    id BIGINT GENERATED ALWAYS AS IDENTITY PRIMARY KEY,
+    work_id BIGINT NOT NULL,
+    owner_user_id BIGINT NOT NULL,
+    export_format VARCHAR(30) NOT NULL,
+    status VARCHAR(30) NOT NULL DEFAULT 'queued',
+    credential_id VARCHAR(120),
+    package_ref TEXT,
+    error_code VARCHAR(80),
+    error_message TEXT,
+    command_id VARCHAR(120) NOT NULL,
+    creator VARCHAR(64) NOT NULL DEFAULT '',
+    create_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    updater VARCHAR(64) NOT NULL DEFAULT '',
+    update_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    deleted BOOLEAN NOT NULL DEFAULT FALSE,
+    tenant_id BIGINT NOT NULL DEFAULT 0,
+    CONSTRAINT uk_muse_content_export_command UNIQUE (tenant_id, command_id)
+);
+
+CREATE INDEX idx_muse_content_export_work ON muse_content_export_task(tenant_id, work_id, status);
+CREATE TRIGGER trg_muse_content_export_updated_at BEFORE UPDATE ON muse_content_export_task FOR EACH ROW EXECUTE FUNCTION update_updated_at_column();
+
+CREATE TABLE muse_content_command_log (
+    id BIGINT GENERATED ALWAYS AS IDENTITY PRIMARY KEY,
+    command_id VARCHAR(120) NOT NULL,
+    command_type VARCHAR(80) NOT NULL,
+    owner_user_id BIGINT NOT NULL,
+    target_type VARCHAR(50),
+    target_id BIGINT,
+    request_hash VARCHAR(128),
+    result_snapshot JSONB,
+    creator VARCHAR(64) NOT NULL DEFAULT '',
+    create_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    updater VARCHAR(64) NOT NULL DEFAULT '',
+    update_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    deleted BOOLEAN NOT NULL DEFAULT FALSE,
+    tenant_id BIGINT NOT NULL DEFAULT 0,
+    CONSTRAINT uk_muse_content_command UNIQUE (tenant_id, command_id)
+);
+
+CREATE INDEX idx_muse_content_command_owner ON muse_content_command_log(tenant_id, owner_user_id, create_time);
+CREATE TRIGGER trg_muse_content_command_updated_at BEFORE UPDATE ON muse_content_command_log FOR EACH ROW EXECUTE FUNCTION update_updated_at_column();

db/ddl/V26__fix_content_softdelete_order_uk_partial.sql

@@ -0,0 +1,28 @@
+-- 修复 content 章节/Block 软删后 order_no 唯一约束碰撞 schema bug。
+--
+-- 背景:muse_content_chapter / muse_content_block 继承 yudao 软删(deleted 列 + @TableLogic,
+-- 删除是 UPDATE deleted=1、行仍在表)。但 uk_muse_content_chapter_work_order(tenant_id, work_id, order_no)
+-- 与 uk_muse_content_block_chapter_order(tenant_id, chapter_id, order_no) 是完整唯一约束、不含 deleted 条件,
+-- 软删行仍占 order_no 名额;而 createChapter/createBlock 用 orderNo = selectCount(active)+1(只算未删行),
+-- 删章节/Block 后再建同序号 → 与软删行碰撞 → PostgreSQL duplicate key → 500 系统异常。
+-- (反假绿:studio 前端补齐 chapter create 契约字段后真后端建章才暴露此 bug;见 docs/mvp/进度总账.md。)
+--
+-- 修法:改为 partial unique index(WHERE deleted = FALSE),软删行不占名额、active 行唯一性不变,
+-- 对齐同库 uk_muse_content_chapter_command 等既有 partial index 惯例。
+-- 安全性:order_no uk 不被任何 ON CONFLICT 引用(content 的 ON CONFLICT 仅在 command/outbox 表、用完整 uk),
+-- 故改 partial 不影响 insert/ON CONFLICT 仲裁(区别于 V22 须保留完整索引的 command_id 场景)。
+-- 幂等:DROP CONSTRAINT/INDEX IF EXISTS 后重建,对已应用 V1 的库与全新库均安全。
+
+-- 章节:DROP 原完整唯一约束,改建 partial unique index(同名)
+ALTER TABLE muse_content_chapter DROP CONSTRAINT IF EXISTS uk_muse_content_chapter_work_order;
+DROP INDEX IF EXISTS uk_muse_content_chapter_work_order;
+CREATE UNIQUE INDEX uk_muse_content_chapter_work_order
+    ON muse_content_chapter (tenant_id, work_id, order_no)
+    WHERE deleted = FALSE;
+
+-- Block:同理(createBlock 同样用 selectCountByChapterId(active)+1,删后重建同序号会碰撞)
+ALTER TABLE muse_content_block DROP CONSTRAINT IF EXISTS uk_muse_content_block_chapter_order;
+DROP INDEX IF EXISTS uk_muse_content_block_chapter_order;
+CREATE UNIQUE INDEX uk_muse_content_block_chapter_order
+    ON muse_content_block (tenant_id, chapter_id, order_no)
+    WHERE deleted = FALSE;

db/ddl/V3__init_meta_schema.sql

@@ -0,0 +1,133 @@
+-- Muse Meta 模块初始 Schema（PostgreSQL 16）
+-- 影响范围：MetaSchema、字段、可见性策略、保护节点、功能链路。
+
+CREATE TABLE muse_meta_schema (
+    id BIGINT GENERATED ALWAYS AS IDENTITY PRIMARY KEY,
+    schema_key VARCHAR(100) NOT NULL,
+    domain VARCHAR(20) NOT NULL,
+    scope VARCHAR(20) NOT NULL,
+    target_type VARCHAR(50),
+    display_name VARCHAR(200) NOT NULL,
+    active_version_id BIGINT,
+    effective_scope JSONB,
+    projection_version INT NOT NULL DEFAULT 1,
+    status VARCHAR(20) NOT NULL DEFAULT 'active',
+    creator VARCHAR(64) NOT NULL DEFAULT '',
+    create_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    updater VARCHAR(64) NOT NULL DEFAULT '',
+    update_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    deleted BOOLEAN NOT NULL DEFAULT FALSE,
+    tenant_id BIGINT NOT NULL DEFAULT 0,
+    CONSTRAINT uk_muse_meta_schema_key UNIQUE (tenant_id, domain, scope, target_type, schema_key)
+);
+
+CREATE TRIGGER trg_muse_meta_schema_updated_at BEFORE UPDATE ON muse_meta_schema FOR EACH ROW EXECUTE FUNCTION update_updated_at_column();
+
+CREATE TABLE muse_meta_schema_version (
+    id BIGINT GENERATED ALWAYS AS IDENTITY PRIMARY KEY,
+    schema_id BIGINT NOT NULL,
+    schema_key VARCHAR(100) NOT NULL,
+    version_no INT NOT NULL,
+    status VARCHAR(20) NOT NULL DEFAULT 'draft',
+    active_flag BOOLEAN NOT NULL DEFAULT FALSE,
+    effective_scope JSONB,
+    field_contract_snapshot JSONB,
+    impact_preview_snapshot JSONB,
+    rollback_from_version_id BIGINT,
+    rollback_to_version_id BIGINT,
+    published_by VARCHAR(64),
+    published_at TIMESTAMP,
+    activated_by VARCHAR(64),
+    activated_at TIMESTAMP,
+    change_note VARCHAR(500),
+    command_id VARCHAR(120),
+    creator VARCHAR(64) NOT NULL DEFAULT '',
+    create_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    updater VARCHAR(64) NOT NULL DEFAULT '',
+    update_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    deleted BOOLEAN NOT NULL DEFAULT FALSE,
+    tenant_id BIGINT NOT NULL DEFAULT 0,
+    CONSTRAINT uk_muse_meta_schema_version_key UNIQUE (tenant_id, schema_key, version_no)
+);
+
+CREATE INDEX idx_muse_meta_schema_ver_sid ON muse_meta_schema_version(tenant_id, schema_id);
+CREATE UNIQUE INDEX uk_muse_meta_schema_ver_command ON muse_meta_schema_version(tenant_id, command_id) WHERE command_id IS NOT NULL;
+CREATE TRIGGER trg_muse_meta_schema_version_updated_at BEFORE UPDATE ON muse_meta_schema_version FOR EACH ROW EXECUTE FUNCTION update_updated_at_column();
+
+CREATE TABLE muse_meta_field (
+    id BIGINT GENERATED ALWAYS AS IDENTITY PRIMARY KEY,
+    schema_version_id BIGINT NOT NULL,
+    field_key VARCHAR(100) NOT NULL,
+    display_name VARCHAR(200) NOT NULL,
+    field_type VARCHAR(20) NOT NULL,
+    is_required BOOLEAN NOT NULL DEFAULT FALSE,
+    default_value JSONB,
+    enum_values JSONB,
+    validation_rules JSONB,
+    sort_order INT NOT NULL DEFAULT 0,
+    creator VARCHAR(64) NOT NULL DEFAULT '',
+    create_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    updater VARCHAR(64) NOT NULL DEFAULT '',
+    update_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    deleted BOOLEAN NOT NULL DEFAULT FALSE,
+    tenant_id BIGINT NOT NULL DEFAULT 0
+);
+
+CREATE INDEX idx_muse_meta_field_ver ON muse_meta_field(tenant_id, schema_version_id);
+CREATE TRIGGER trg_muse_meta_field_updated_at BEFORE UPDATE ON muse_meta_field FOR EACH ROW EXECUTE FUNCTION update_updated_at_column();
+
+CREATE TABLE muse_meta_visibility_policy (
+    id BIGINT GENERATED ALWAYS AS IDENTITY PRIMARY KEY,
+    schema_version_id BIGINT NOT NULL,
+    ui_visible BOOLEAN NOT NULL DEFAULT TRUE,
+    ai_context BOOLEAN NOT NULL DEFAULT FALSE,
+    user_editable BOOLEAN NOT NULL DEFAULT TRUE,
+    user_searchable BOOLEAN NOT NULL DEFAULT FALSE,
+    exportable BOOLEAN NOT NULL DEFAULT FALSE,
+    policy_snapshot JSONB,
+    creator VARCHAR(64) NOT NULL DEFAULT '',
+    create_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    updater VARCHAR(64) NOT NULL DEFAULT '',
+    update_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    deleted BOOLEAN NOT NULL DEFAULT FALSE,
+    tenant_id BIGINT NOT NULL DEFAULT 0
+);
+
+CREATE INDEX idx_muse_meta_vis_policy_ver ON muse_meta_visibility_policy(tenant_id, schema_version_id);
+CREATE TRIGGER trg_muse_meta_vis_policy_updated_at BEFORE UPDATE ON muse_meta_visibility_policy FOR EACH ROW EXECUTE FUNCTION update_updated_at_column();
+
+CREATE TABLE muse_meta_protection_node (
+    id BIGINT GENERATED ALWAYS AS IDENTITY PRIMARY KEY,
+    node_key VARCHAR(100) NOT NULL,
+    display_name VARCHAR(200) NOT NULL,
+    owner_module VARCHAR(50) NOT NULL,
+    replaceable BOOLEAN NOT NULL DEFAULT FALSE,
+    policy_snapshot JSONB,
+    status VARCHAR(20) NOT NULL DEFAULT 'active',
+    creator VARCHAR(64) NOT NULL DEFAULT '',
+    create_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    updater VARCHAR(64) NOT NULL DEFAULT '',
+    update_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    deleted BOOLEAN NOT NULL DEFAULT FALSE,
+    tenant_id BIGINT NOT NULL DEFAULT 0,
+    CONSTRAINT uk_muse_meta_protection_node UNIQUE (tenant_id, node_key)
+);
+
+CREATE TRIGGER trg_muse_meta_protection_node_updated_at BEFORE UPDATE ON muse_meta_protection_node FOR EACH ROW EXECUTE FUNCTION update_updated_at_column();
+
+CREATE TABLE muse_meta_function_chain (
+    id BIGINT GENERATED ALWAYS AS IDENTITY PRIMARY KEY,
+    chain_key VARCHAR(100) NOT NULL,
+    active_version VARCHAR(50),
+    chain_snapshot JSONB,
+    status VARCHAR(20) NOT NULL DEFAULT 'active',
+    creator VARCHAR(64) NOT NULL DEFAULT '',
+    create_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    updater VARCHAR(64) NOT NULL DEFAULT '',
+    update_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    deleted BOOLEAN NOT NULL DEFAULT FALSE,
+    tenant_id BIGINT NOT NULL DEFAULT 0,
+    CONSTRAINT uk_muse_meta_function_chain UNIQUE (tenant_id, chain_key)
+);
+
+CREATE TRIGGER trg_muse_meta_function_chain_updated_at BEFORE UPDATE ON muse_meta_function_chain FOR EACH ROW EXECUTE FUNCTION update_updated_at_column();

db/ddl/V5__init_knowledge_schema.sql

@@ -0,0 +1,206 @@
+-- Muse Knowledge 模块初始 Schema（PostgreSQL 16）
+-- 影响范围：用户/全局知识库、资料版本、实体关系、草稿、作品绑定。
+
+CREATE TABLE muse_knowledge_base (
+    id BIGINT GENERATED ALWAYS AS IDENTITY PRIMARY KEY,
+    name VARCHAR(200) NOT NULL,
+    description VARCHAR(500),
+    kb_type VARCHAR(20) NOT NULL DEFAULT 'user',
+    owner_user_id BIGINT,
+    source_market_asset_id BIGINT,
+    status VARCHAR(20) NOT NULL DEFAULT 'active',
+    active_version INT NOT NULL DEFAULT 1,
+    visibility_policy JSONB,
+    license_snapshot_id BIGINT,
+    command_id VARCHAR(120),
+    revision INT NOT NULL DEFAULT 1,
+    creator VARCHAR(64) NOT NULL DEFAULT '',
+    create_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    updater VARCHAR(64) NOT NULL DEFAULT '',
+    update_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    deleted BOOLEAN NOT NULL DEFAULT FALSE,
+    tenant_id BIGINT NOT NULL DEFAULT 0
+);
+
+CREATE INDEX idx_muse_kb_owner_status ON muse_knowledge_base(tenant_id, owner_user_id, status);
+CREATE INDEX idx_muse_kb_type ON muse_knowledge_base(tenant_id, kb_type);
+CREATE UNIQUE INDEX uk_muse_kb_command ON muse_knowledge_base(tenant_id, command_id) WHERE command_id IS NOT NULL;
+CREATE TRIGGER trg_muse_knowledge_base_updated_at BEFORE UPDATE ON muse_knowledge_base FOR EACH ROW EXECUTE FUNCTION update_updated_at_column();
+
+CREATE TABLE muse_knowledge_document (
+    id BIGINT GENERATED ALWAYS AS IDENTITY PRIMARY KEY,
+    kb_id BIGINT NOT NULL,
+    title VARCHAR(200) NOT NULL,
+    file_name VARCHAR(500),
+    file_size BIGINT,
+    mime_type VARCHAR(100),
+    file_hash VARCHAR(100),
+    storage_ref VARCHAR(500),
+    scan_status VARCHAR(20) NOT NULL DEFAULT 'pending',
+    parse_status VARCHAR(20),
+    index_status VARCHAR(20),
+    retention_policy JSONB,
+    author VARCHAR(200),
+    source_url VARCHAR(1000),
+    tags JSONB,
+    command_id VARCHAR(120),
+    creator VARCHAR(64) NOT NULL DEFAULT '',
+    create_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    updater VARCHAR(64) NOT NULL DEFAULT '',
+    update_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    deleted BOOLEAN NOT NULL DEFAULT FALSE,
+    tenant_id BIGINT NOT NULL DEFAULT 0
+);
+
+CREATE INDEX idx_muse_kb_doc_kb ON muse_knowledge_document(tenant_id, kb_id, scan_status);
+CREATE INDEX idx_muse_kb_doc_hash ON muse_knowledge_document(tenant_id, file_hash);
+CREATE UNIQUE INDEX uk_muse_kb_doc_command ON muse_knowledge_document(tenant_id, command_id) WHERE command_id IS NOT NULL;
+CREATE TRIGGER trg_muse_knowledge_document_updated_at BEFORE UPDATE ON muse_knowledge_document FOR EACH ROW EXECUTE FUNCTION update_updated_at_column();
+
+CREATE TABLE muse_knowledge_document_version (
+    id BIGINT GENERATED ALWAYS AS IDENTITY PRIMARY KEY,
+    document_id BIGINT NOT NULL,
+    version INT NOT NULL,
+    file_hash VARCHAR(100) NOT NULL,
+    storage_ref VARCHAR(500),
+    content_hash VARCHAR(100),
+    source_snapshot_id BIGINT,
+    scan_status VARCHAR(20) NOT NULL DEFAULT 'pending',
+    parse_status VARCHAR(20) NOT NULL DEFAULT 'pending',
+    processing_status VARCHAR(20) NOT NULL DEFAULT 'pending',
+    previous_version_id BIGINT,
+    change_summary JSONB,
+    change_reason VARCHAR(500),
+    command_id VARCHAR(120),
+    creator VARCHAR(64) NOT NULL DEFAULT '',
+    create_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    updater VARCHAR(64) NOT NULL DEFAULT '',
+    update_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    deleted BOOLEAN NOT NULL DEFAULT FALSE,
+    tenant_id BIGINT NOT NULL DEFAULT 0,
+    CONSTRAINT uk_muse_kb_doc_ver UNIQUE (tenant_id, document_id, version)
+);
+
+CREATE INDEX idx_muse_kb_doc_ver_doc ON muse_knowledge_document_version(tenant_id, document_id);
+CREATE TRIGGER trg_muse_kb_doc_ver_updated_at BEFORE UPDATE ON muse_knowledge_document_version FOR EACH ROW EXECUTE FUNCTION update_updated_at_column();
+
+CREATE TABLE muse_knowledge_entity (
+    id BIGINT GENERATED ALWAYS AS IDENTITY PRIMARY KEY,
+    work_id BIGINT NOT NULL,
+    kb_id BIGINT,
+    entity_type VARCHAR(50) NOT NULL,
+    normalized_name VARCHAR(200) NOT NULL,
+    scope VARCHAR(50) NOT NULL DEFAULT 'global',
+    description TEXT,
+    attributes JSONB,
+    source_snapshot_id BIGINT,
+    authorization_snapshot_id BIGINT,
+    source_status VARCHAR(30) NOT NULL DEFAULT 'active',
+    source_action_policy VARCHAR(30) NOT NULL DEFAULT 'allowed',
+    source_event_version INT,
+    lineage_payload JSONB,
+    status VARCHAR(20) NOT NULL DEFAULT 'active',
+    confidence DOUBLE PRECISION DEFAULT 1.0,
+    source_type VARCHAR(50),
+    source_id BIGINT,
+    command_id VARCHAR(120),
+    revision INT NOT NULL DEFAULT 1,
+    creator VARCHAR(64) NOT NULL DEFAULT '',
+    create_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    updater VARCHAR(64) NOT NULL DEFAULT '',
+    update_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    deleted BOOLEAN NOT NULL DEFAULT FALSE,
+    tenant_id BIGINT NOT NULL DEFAULT 0,
+    CONSTRAINT uk_muse_knowledge_entity UNIQUE (tenant_id, work_id, entity_type, normalized_name, scope)
+);
+
+CREATE INDEX idx_muse_knowledge_entity_work_type ON muse_knowledge_entity(tenant_id, work_id, entity_type);
+CREATE INDEX idx_muse_knowledge_entity_kb ON muse_knowledge_entity(tenant_id, kb_id);
+CREATE UNIQUE INDEX uk_muse_knowledge_entity_command ON muse_knowledge_entity(tenant_id, command_id) WHERE command_id IS NOT NULL;
+CREATE TRIGGER trg_muse_knowledge_entity_updated_at BEFORE UPDATE ON muse_knowledge_entity FOR EACH ROW EXECUTE FUNCTION update_updated_at_column();
+
+CREATE TABLE muse_knowledge_relation (
+    id BIGINT GENERATED ALWAYS AS IDENTITY PRIMARY KEY,
+    work_id BIGINT NOT NULL,
+    source_entity_id BIGINT NOT NULL,
+    target_entity_id BIGINT NOT NULL,
+    relation_type VARCHAR(50) NOT NULL,
+    description VARCHAR(500),
+    attributes JSONB,
+    source_status VARCHAR(30) NOT NULL DEFAULT 'active',
+    source_action_policy VARCHAR(30) NOT NULL DEFAULT 'allowed',
+    command_id VARCHAR(120),
+    revision INT NOT NULL DEFAULT 1,
+    creator VARCHAR(64) NOT NULL DEFAULT '',
+    create_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    updater VARCHAR(64) NOT NULL DEFAULT '',
+    update_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    deleted BOOLEAN NOT NULL DEFAULT FALSE,
+    tenant_id BIGINT NOT NULL DEFAULT 0
+);
+
+CREATE INDEX idx_muse_knowledge_rel_work ON muse_knowledge_relation(tenant_id, work_id);
+CREATE INDEX idx_muse_knowledge_rel_src ON muse_knowledge_relation(tenant_id, source_entity_id);
+CREATE INDEX idx_muse_knowledge_rel_tgt ON muse_knowledge_relation(tenant_id, target_entity_id);
+CREATE UNIQUE INDEX uk_muse_knowledge_rel_command ON muse_knowledge_relation(tenant_id, command_id) WHERE command_id IS NOT NULL;
+CREATE TRIGGER trg_muse_knowledge_relation_updated_at BEFORE UPDATE ON muse_knowledge_relation FOR EACH ROW EXECUTE FUNCTION update_updated_at_column();
+
+CREATE TABLE muse_knowledge_draft (
+    id BIGINT GENERATED ALWAYS AS IDENTITY PRIMARY KEY,
+    work_id BIGINT NOT NULL,
+    entity_id BIGINT,
+    draft_type VARCHAR(30) NOT NULL DEFAULT 'entity',
+    target_object_id BIGINT,
+    proposed_changes JSONB,
+    draft_payload JSONB,
+    current_canonical_snapshot JSONB,
+    source_snapshot_id BIGINT,
+    authorization_snapshot_id BIGINT,
+    risk_marker_snapshot JSONB,
+    source_status VARCHAR(30),
+    source_action_policy VARCHAR(30),
+    status VARCHAR(20) NOT NULL DEFAULT 'pending',
+    confidence DOUBLE PRECISION,
+    source_type VARCHAR(50),
+    source_id BIGINT,
+    command_id VARCHAR(120),
+    revision INT NOT NULL DEFAULT 1,
+    creator VARCHAR(64) NOT NULL DEFAULT '',
+    create_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    updater VARCHAR(64) NOT NULL DEFAULT '',
+    update_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    deleted BOOLEAN NOT NULL DEFAULT FALSE,
+    tenant_id BIGINT NOT NULL DEFAULT 0
+);
+
+CREATE INDEX idx_muse_knowledge_draft_work_status ON muse_knowledge_draft(tenant_id, work_id, status, create_time);
+CREATE INDEX idx_muse_knowledge_draft_entity ON muse_knowledge_draft(tenant_id, entity_id);
+CREATE UNIQUE INDEX uk_muse_knowledge_draft_command ON muse_knowledge_draft(tenant_id, command_id) WHERE command_id IS NOT NULL;
+CREATE TRIGGER trg_muse_knowledge_draft_updated_at BEFORE UPDATE ON muse_knowledge_draft FOR EACH ROW EXECUTE FUNCTION update_updated_at_column();
+
+CREATE TABLE muse_knowledge_binding (
+    id BIGINT GENERATED ALWAYS AS IDENTITY PRIMARY KEY,
+    work_id BIGINT NOT NULL,
+    kb_id BIGINT NOT NULL,
+    binding_type VARCHAR(50) NOT NULL DEFAULT 'read',
+    binding_scope VARCHAR(50) NOT NULL DEFAULT 'read',
+    binding_status VARCHAR(30) NOT NULL DEFAULT 'active',
+    source_snapshot_id BIGINT,
+    authorization_snapshot_id BIGINT,
+    source_version INT,
+    target_version INT,
+    command_id VARCHAR(120),
+    revision INT NOT NULL DEFAULT 1,
+    creator VARCHAR(64) NOT NULL DEFAULT '',
+    create_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    updater VARCHAR(64) NOT NULL DEFAULT '',
+    update_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    deleted BOOLEAN NOT NULL DEFAULT FALSE,
+    tenant_id BIGINT NOT NULL DEFAULT 0,
+    CONSTRAINT uk_muse_knowledge_binding_work_kb UNIQUE (tenant_id, work_id, kb_id)
+);
+
+CREATE INDEX idx_muse_knowledge_binding_work ON muse_knowledge_binding(tenant_id, work_id);
+CREATE INDEX idx_muse_knowledge_binding_kb ON muse_knowledge_binding(tenant_id, kb_id);
+CREATE UNIQUE INDEX uk_muse_knowledge_binding_command ON muse_knowledge_binding(tenant_id, command_id) WHERE command_id IS NOT NULL;
+CREATE TRIGGER trg_muse_knowledge_binding_updated_at BEFORE UPDATE ON muse_knowledge_binding FOR EACH ROW EXECUTE FUNCTION update_updated_at_column();

db/表映射.md

@@ -0,0 +1,56 @@
+# muse-example 表 ↔ 主仓迁移来源映射（A2 审查面）
+
+> 口径（创始人拍板③ 2026-07-10）：主仓表**原样不改列**；实验私货全进 `example_*` 前缀。
+> 建表方式：`db/ddl/` 下文件经 db skill `apply`，主仓部分为 `muse-cloud/sql/muse/` 原文拷贝或逐字摘录。
+> 应用顺序：V1 → V3 → V5 → 90-ALTER摘录 → V26 → 91-example。已于 2026-07-13 应用，共 **24 张表**。
+
+## 主仓一致表（20 张）
+
+| 表 | 来源迁移 | 叠加的 ALTER | 实验角色 |
+|---|---|---|---|
+| muse_content_work | V1 | — | 参考书/未来新作的作品行（B1 导入） |
+| muse_content_chapter | V1 | V26（order 唯一键改 partial） | 分章行（B1 导入） |
+| muse_content_block | V1 | V26（同上） | 章正文载体（一章一 block，content_text） |
+| muse_content_block_source_attribution | V1 | V30（authorization_snapshot_id→varchar(128)） | 暂空置；C 线来源归因 |
+| muse_content_import_task | V1 | — | B1 导入任务审计 |
+| muse_content_export_task | V1 | V9（+4 列） | 暂空置；C9 导出 |
+| muse_content_command_log | V1 | — | 写命令幂等审计（import/confirm 用） |
+| muse_meta_schema | V3 | — | 23 型结构本体行（A3 入库） |
+| muse_meta_schema_version | V3 | V10（active 唯一索引） | 型版本（A3：每型 v1 active） |
+| muse_meta_field | V3 | — | 逐型字段合同行（A3 入库） |
+| muse_meta_visibility_policy | V3 | — | aiContext 控制项落点（A3 入库） |
+| muse_meta_protection_node | V3 | V10（+6 列） | 保护节点登记（read-context/confirm/judge） |
+| muse_meta_function_chain | V3 | — | 功能链登记（meta/chains 的库内影子，随需入库） |
+| muse_knowledge_base | V5 | — | 私有参考书库 + 公共范式库（kb 行） |
+| muse_knowledge_document | V5 | — | 参考书原文档案指针（storage_ref 指仓内路径） |
+| muse_knowledge_document_version | V5 | — | 暂空置 |
+| muse_knowledge_entity | V5 | — | **已确认**知识行（B5 确认后落此） |
+| muse_knowledge_relation | V5 | — | 知识行间关系（二轮再启用） |
+| muse_knowledge_draft | V5 | V14（两快照列→varchar(128)） | **草稿**知识行（B2 拆书产出落此） |
+| muse_knowledge_binding | V5 | V14（同上） | 作品↔库绑定（C3 起用） |
+
+## 实验私货表（4 张，`db/ddl/91-example实验私货.sql`）
+
+| 表 | 用途 |
+|---|---|
+| example_reference_work | 参考书档案（reference_work 型的库内实例载体）：作者/章数对账/拆书范围与状态 |
+| example_parse_task | 拆书任务状态机（一章一行，scaffold/pattern 两阶段状态）——workflow 断点续跑与幂等 |
+| example_parse_scaffold | 逐章脚手架：逆推细纲（3–5%）+ 实体清单 JSONB |
+| example_knowledge_embedding | 知识行嵌入边表：vector(1024)+HNSW（余弦），content_hash 幂等，挂 draft 或 entity |
+
+## 暂缓建表登记（主仓有、实验现阶段未建；需要时按原样加建）
+
+| 迁移 | 表 | 暂缓原因 |
+|---|---|---|
+| V9 | muse_content_planning_section / governance_action | C2 规划确认时加建 planning_section |
+| V10 | meta 治理 8 表（command/audit/validation/impact/gray/chain_version/chain_slot/chain_node） | 治理命令流实验不镜像；装配入库随阶段二 |
+| V14 | knowledge RAGFlow/治理 14 表 | 实验用 pgvector 自持嵌入，不接 RAGFlow |
+| V18/V21 | 两张 event_publish_outbox | 事件出箱不镜像 |
+| V25/V28/V32/V36/V37 | field_snapshot / asset_use_precheck / block_revision_snapshot / ai_generated_fulltext / task 批次列 | 对应场景（用量投影/市场/版本快照/SSE 瞬态）不在实验闭环内 |
+
+## 实验写入约定
+
+- `tenant_id=1`、`creator='1'`、`owner_user_id=1`（系统主账号）——写入层统一给值；列默认仍照主仓（0/''），不改列。
+- 公共范式的全局知识行：`work_id=0` + `scope='global'` + 挂 `kb_type='global'` 的库；参考书私有库 `kb_type='user'`。
+- 软删照主仓：`deleted=TRUE`，不物理删。
+- 双轨对应：B2 产出全落 `muse_knowledge_draft(status='pending')`；B5 管理员确认（仅创始人触发）= draft 翻 `confirmed` + 落 `muse_knowledge_entity(status='active')`；丢弃= draft 翻 `ignored`。