diff --git a/apps/backend/src/db/migrations/103_product_catalog.sql b/apps/backend/src/db/migrations/103_product_catalog.sql
index f20ae134..803f9434 100644
--- a/apps/backend/src/db/migrations/103_product_catalog.sql
+++ b/apps/backend/src/db/migrations/103_product_catalog.sql
@@ -3,11 +3,18 @@
 -- 目标：把「平台原始 SKU / 商品名」归一到内部标准商品，后续成本、销量、
 -- 店铺日报和订单毛利都只认 canonical_product_id。
 --
+-- 设计要点（2026-10-23 重写）：
+--   1. 商品表不存「履约载体」。履约方式只由 order_items.fulfillment_mode 决定，
+--      商品的来源由 canonical_sku 前缀表达（aff: 自动发货 / tpl: 接单模板 /
+--      wo: 历史工单证据 / man: 人工），避免两处口径互相打架。
+--   2. 映射唯一键包含匹配方式与名称键，91 同店铺复用 sku_code 才能消歧。
+--   3. normalized_quantity 不在这里预填：预填会把「5 个」类 SKU 永久锁死成 1，
+--      必须由回填脚本按 SKU 名称解析后写入。
+--
 -- 约束：
 --   1. canonical_sku 是内部稳定编码，创建后不随平台改名；
---   2. 同一 provider + shop_id + platform_sku_code 只能有一条映射；
---   3. shop_id 为空串表示该来源的全局映射，具体店铺映射优先；
---   4. order_items.normalized_quantity 从 SKU 名称解析，解析失败先回填 quantity。
+--   2. shop_id 为空串表示该来源的全局映射，具体店铺映射优先；
+--   3. 任何一层匹配出现多个候选都不自动选择，交由冲突清单人工处理。
 
 CREATE TABLE IF NOT EXISTS product_catalog (
   id                BIGSERIAL PRIMARY KEY,
@@ -25,8 +32,6 @@ CREATE UNIQUE INDEX IF NOT EXISTS uq_product_catalog_canonical_sku
   ON product_catalog (canonical_sku);
 CREATE INDEX IF NOT EXISTS idx_product_catalog_status_category
   ON product_catalog (status, category, canonical_name);
-CREATE INDEX IF NOT EXISTS idx_product_catalog_name
-  ON product_catalog (canonical_name);
 
 DO $$
 BEGIN
@@ -48,16 +53,19 @@ BEGIN
 END $$;
 
 COMMENT ON TABLE product_catalog IS '标准商品/皮肤主数据：报表、成本和销量统计的唯一商品维度';
-COMMENT ON COLUMN product_catalog.canonical_sku IS '内部稳定编码；创建后不随平台改名';
+COMMENT ON COLUMN product_catalog.canonical_sku IS
+  '内部稳定编码；前缀表达来源：aff: 自动发货、tpl: 接单模板、wo: 工单证据、man: 人工';
 COMMENT ON COLUMN product_catalog.unit_name IS '计量单位：件/个/颗等，配合 normalized_quantity 使用';
 
 CREATE TABLE IF NOT EXISTS channel_product_mappings (
   id                   BIGSERIAL PRIMARY KEY,
   provider             TEXT NOT NULL,
   shop_id              TEXT NOT NULL DEFAULT '',
-  platform_sku_code    TEXT NOT NULL,
+  platform_sku_code    TEXT NOT NULL DEFAULT '',
   platform_sku_name    TEXT NOT NULL DEFAULT '',
+  platform_sku_name_key TEXT NOT NULL DEFAULT '',
   canonical_product_id BIGINT NOT NULL REFERENCES product_catalog(id) ON DELETE RESTRICT,
+  match_type           TEXT NOT NULL DEFAULT 'sku_code',
   quantity_multiplier  INTEGER NOT NULL DEFAULT 1,
   confidence           TEXT NOT NULL DEFAULT 'manual',
   source               TEXT NOT NULL DEFAULT 'manual',
@@ -67,12 +75,25 @@ CREATE TABLE IF NOT EXISTS channel_product_mappings (
   updated_at           TIMESTAMPTZ NOT NULL
 );
 
-CREATE UNIQUE INDEX IF NOT EXISTS uq_channel_product_mappings_key
-  ON channel_product_mappings (provider, shop_id, platform_sku_code);
+CREATE UNIQUE INDEX IF NOT EXISTS uq_channel_product_mappings_match_key
+  ON channel_product_mappings (
+    provider,
+    shop_id,
+    platform_sku_code,
+    match_type,
+    platform_sku_name_key
+  );
 CREATE INDEX IF NOT EXISTS idx_channel_product_mappings_product
   ON channel_product_mappings (canonical_product_id, enabled);
 CREATE INDEX IF NOT EXISTS idx_channel_product_mappings_lookup
-  ON channel_product_mappings (provider, shop_id, platform_sku_code, enabled);
+  ON channel_product_mappings (
+    provider,
+    shop_id,
+    platform_sku_code,
+    match_type,
+    platform_sku_name_key,
+    enabled
+  );
 
 DO $$
 BEGIN
@@ -91,23 +112,41 @@ BEGIN
       ADD CONSTRAINT chk_channel_product_mappings_multiplier
       CHECK (quantity_multiplier > 0);
   END IF;
+  IF NOT EXISTS (SELECT 1 FROM pg_constraint WHERE conname = 'chk_channel_product_mappings_match_type') THEN
+    ALTER TABLE channel_product_mappings
+      ADD CONSTRAINT chk_channel_product_mappings_match_type
+      CHECK (match_type IN ('sku_code', 'sku_code_name', 'sku_code_series', 'sku_name_series'));
+  END IF;
+  IF NOT EXISTS (SELECT 1 FROM pg_constraint WHERE conname = 'chk_channel_product_mappings_match_key') THEN
+    ALTER TABLE channel_product_mappings
+      ADD CONSTRAINT chk_channel_product_mappings_match_key
+      CHECK (platform_sku_code <> '' OR platform_sku_name_key <> '');
+  END IF;
 END $$;
 
 COMMENT ON TABLE channel_product_mappings IS '来源 + 店铺 + 平台 SKU -> 标准商品映射';
 COMMENT ON COLUMN channel_product_mappings.shop_id IS '空串=该来源的全局映射；具体店铺映射优先';
+COMMENT ON COLUMN channel_product_mappings.match_type IS
+  '匹配方式：sku_code / sku_code_name / sku_code_series / sku_name_series';
+COMMENT ON COLUMN channel_product_mappings.platform_sku_name_key IS
+  '归一后的 SKU 名称或数量系列键；sku_code 模式为空串';
 COMMENT ON COLUMN channel_product_mappings.quantity_multiplier IS 'SKU 名称内含数量时的换算系数，例如「5个」=5';
 
 ALTER TABLE order_items
   ADD COLUMN IF NOT EXISTS canonical_product_id BIGINT REFERENCES product_catalog(id) ON DELETE RESTRICT,
-  ADD COLUMN IF NOT EXISTS normalized_quantity INTEGER;
+  ADD COLUMN IF NOT EXISTS normalized_quantity INTEGER,
+  ADD COLUMN IF NOT EXISTS normalized_sku_name TEXT,
+  ADD COLUMN IF NOT EXISTS normalized_sku_series TEXT;
 
 CREATE INDEX IF NOT EXISTS idx_order_items_canonical_product
   ON order_items (canonical_product_id);
+CREATE INDEX IF NOT EXISTS idx_order_items_normalized_sku_lookup
+  ON order_items (normalized_sku_name, normalized_sku_series);
 
 COMMENT ON COLUMN order_items.canonical_product_id IS '归一到标准商品；由渠道映射或人工指定';
-COMMENT ON COLUMN order_items.normalized_quantity IS '归一化件数：从 SKU 名称解析并固化，禁止直接用 quantity 做销量';
-
--- 初始回填只保证不为空；后续由成本中心页面/脚本按 SKU 名称解析覆盖。
-UPDATE order_items
-SET normalized_quantity = GREATEST(COALESCE(quantity, 1), 1)
-WHERE normalized_quantity IS NULL;
+COMMENT ON COLUMN order_items.normalized_quantity IS
+  '归一化件数：从 SKU 名称解析并固化，禁止直接用 quantity 做销量；未解析时保持 NULL';
+COMMENT ON COLUMN order_items.normalized_sku_name IS
+  'NFKC、小写、去空白后的 SKU 名称，用于编码 + 名称消歧';
+COMMENT ON COLUMN order_items.normalized_sku_series IS
+  '剥离数量并归一同义字后的 SKU 名称系列，用于数量系列匹配';
diff --git a/apps/backend/src/db/migrations/104_cost_center.sql b/apps/backend/src/db/migrations/104_cost_center.sql
index b5814ce1..70d27f2f 100644
--- a/apps/backend/src/db/migrations/104_cost_center.sql
+++ b/apps/backend/src/db/migrations/104_cost_center.sql
@@ -1,50 +1,61 @@
--- 104_cost_center.sql —— 成本项目、成本版本、成本快照与成本异常。
+-- 104_cost_center.sql —— 成本项、成本版本、成本快照、成本异常与订单项收入固化。
 --
--- 目标：所有成本金额只在 cost_versions 维护；订单项/工单/人工单只保存
--- cost_snapshot_id 或快照行，历史成本永不回写。
+-- 目标：所有成本金额只在 cost_versions 维护；订单项/工单/人工单只保存快照行，
+-- 历史成本永不回写。
 --
--- 口径（2026-10-17 已确认）：
+-- 设计要点（2026-10-23 重写）：
+--   1. 成本项用 cost_model 表达「成本怎么算」，不再复用 carrier_type。
+--      cost_model ∈ { affiliate_purchase, work_order_reward, manual_actual,
+--                      platform_fee, payment_fee, after_sales_loss, other }
+--      履约方式在 order_items.fulfillment_mode，两者互不推导。
+--   2. 快照表不存 carrier_type：快照属于哪个订单项，join order_items 就有履约方式。
+--   3. 异常表同理，异常属于哪个订单项由 order_item_id 决定。
+--   4. 成本版本按业务发生时间解析（支付时间/验收时间），不是回填时的 now()。
+--   5. 匹配不到成本不得默认 0，必须进入 cost_exceptions。
+--
+-- 口径（2026-10-22 确认）：
 --   * 自动发货 = 上游采购成本；
 --   * 接单平台 = 打手报酬，按总额法计履约成本；
---   * 人工客服 = 人工支出登记，按 5102 履约成本-人工客服；
---   * 匹配不到成本不得默认 0，必须进入 cost_exceptions。
+--   * 人工客服 = 人工支出登记的实际金额；
+--   * 收入按 paid_at 归属销售日；一单一品归整单，多品按等额拆分并标 basis。
+
+-- ---------------------------------------------------------------------------
+-- 1. 成本项
+-- ---------------------------------------------------------------------------
 
 CREATE TABLE IF NOT EXISTS cost_items (
-  id              BIGSERIAL PRIMARY KEY,
-  code            TEXT NOT NULL,
-  name            TEXT NOT NULL,
-  cost_type       TEXT NOT NULL,
-  carrier_type    TEXT NOT NULL,
-  pricing_mode    TEXT NOT NULL,
-  unit_name       TEXT NOT NULL DEFAULT '件',
-  currency        TEXT NOT NULL DEFAULT 'CNY',
-  status          TEXT NOT NULL DEFAULT 'active',
-  note            TEXT NOT NULL DEFAULT '',
-  created_at      TIMESTAMPTZ NOT NULL,
-  updated_at      TIMESTAMPTZ NOT NULL
+  id           BIGSERIAL PRIMARY KEY,
+  code         TEXT NOT NULL,
+  name         TEXT NOT NULL,
+  cost_model   TEXT NOT NULL,
+  pricing_mode TEXT NOT NULL,
+  unit_name    TEXT NOT NULL DEFAULT '件',
+  currency     TEXT NOT NULL DEFAULT 'CNY',
+  status       TEXT NOT NULL DEFAULT 'active',
+  note         TEXT NOT NULL DEFAULT '',
+  created_at   TIMESTAMPTZ NOT NULL,
+  updated_at   TIMESTAMPTZ NOT NULL
 );
 
 CREATE UNIQUE INDEX IF NOT EXISTS uq_cost_items_code
   ON cost_items (code);
-CREATE INDEX IF NOT EXISTS idx_cost_items_status_carrier
-  ON cost_items (status, carrier_type, cost_type);
+CREATE INDEX IF NOT EXISTS idx_cost_items_status_model
+  ON cost_items (status, cost_model);
 
 DO $$
 BEGIN
-  IF NOT EXISTS (SELECT 1 FROM pg_constraint WHERE conname = 'chk_cost_items_cost_type') THEN
+  IF NOT EXISTS (SELECT 1 FROM pg_constraint WHERE conname = 'chk_cost_items_cost_model') THEN
     ALTER TABLE cost_items
-      ADD CONSTRAINT chk_cost_items_cost_type
-      CHECK (cost_type IN ('auto_purchase', 'work_order_reward', 'manual_service', 'platform_fee', 'payment_fee', 'after_sales_loss', 'shipping_other', 'other'));
-  END IF;
-  IF NOT EXISTS (SELECT 1 FROM pg_constraint WHERE conname = 'chk_cost_items_carrier_type') THEN
-    ALTER TABLE cost_items
-      ADD CONSTRAINT chk_cost_items_carrier_type
-      CHECK (carrier_type IN ('affiliate_dash', 'work_order', 'manual', 'shared'));
+      ADD CONSTRAINT chk_cost_items_cost_model
+      CHECK (cost_model IN (
+        'affiliate_purchase', 'work_order_reward', 'manual_actual',
+        'platform_fee', 'payment_fee', 'after_sales_loss', 'other'
+      ));
   END IF;
   IF NOT EXISTS (SELECT 1 FROM pg_constraint WHERE conname = 'chk_cost_items_pricing_mode') THEN
     ALTER TABLE cost_items
       ADD CONSTRAINT chk_cost_items_pricing_mode
-      CHECK (pricing_mode IN ('per_item', 'per_order', 'per_hour', 'percent_of_revenue', 'fixed_daily'));
+      CHECK (pricing_mode IN ('per_item', 'per_order', 'percent_of_revenue'));
   END IF;
   IF NOT EXISTS (SELECT 1 FROM pg_constraint WHERE conname = 'chk_cost_items_status') THEN
     ALTER TABLE cost_items
@@ -53,9 +64,17 @@ BEGIN
   END IF;
 END $$;
 
-COMMENT ON TABLE cost_items IS '成本项定义：类型 + 载体 + 计价方式；金额一律放 cost_versions';
-COMMENT ON COLUMN cost_items.carrier_type IS '适用执行载体；shared 表示与载体无关的通用成本';
-COMMENT ON COLUMN cost_items.pricing_mode IS 'per_item/per_order/per_hour/percent_of_revenue/fixed_daily';
+COMMENT ON TABLE cost_items IS '成本项定义：成本模型 + 计价方式；金额一律放 cost_versions';
+COMMENT ON COLUMN cost_items.cost_model IS
+  '成本模型：affiliate_purchase=上游采购 / work_order_reward=打手报酬 / manual_actual=人工实际支出 / platform_fee / payment_fee / after_sales_loss / other';
+COMMENT ON COLUMN cost_items.pricing_mode IS
+  '计价方式：per_item 按件 / per_order 按单 / percent_of_revenue 按收入比例';
+COMMENT ON COLUMN cost_items.code IS
+  '成本项编码；sys_ 前缀为系统内置项，业务侧不要删除或改编码';
+
+-- ---------------------------------------------------------------------------
+-- 2. 成本版本：改价只在这里新增版本
+-- ---------------------------------------------------------------------------
 
 CREATE TABLE IF NOT EXISTS cost_versions (
   id                        BIGSERIAL PRIMARY KEY,
@@ -99,13 +118,11 @@ DO $$
 BEGIN
   IF NOT EXISTS (SELECT 1 FROM pg_constraint WHERE conname = 'chk_cost_versions_amount') THEN
     ALTER TABLE cost_versions
-      ADD CONSTRAINT chk_cost_versions_amount
-      CHECK (amount_fen >= 0);
+      ADD CONSTRAINT chk_cost_versions_amount CHECK (amount_fen >= 0);
   END IF;
   IF NOT EXISTS (SELECT 1 FROM pg_constraint WHERE conname = 'chk_cost_versions_rate') THEN
     ALTER TABLE cost_versions
-      ADD CONSTRAINT chk_cost_versions_rate
-      CHECK (rate_bps >= 0 AND rate_bps <= 10000);
+      ADD CONSTRAINT chk_cost_versions_rate CHECK (rate_bps >= 0 AND rate_bps <= 10000);
   END IF;
   IF NOT EXISTS (SELECT 1 FROM pg_constraint WHERE conname = 'chk_cost_versions_period') THEN
     ALTER TABLE cost_versions
@@ -134,6 +151,10 @@ COMMENT ON COLUMN cost_versions.amount_fen IS '固定金额（分）；percent_o
 COMMENT ON COLUMN cost_versions.rate_bps IS '比例成本，单位万分比；10000=100%';
 COMMENT ON COLUMN cost_versions.scope_type IS '匹配优先级：rule > product > shop > provider > global';
 
+-- ---------------------------------------------------------------------------
+-- 3. 成本快照
+-- ---------------------------------------------------------------------------
+
 CREATE TABLE IF NOT EXISTS cost_snapshots (
   id                   BIGSERIAL PRIMARY KEY,
   subject_type         TEXT NOT NULL,
@@ -141,8 +162,6 @@ CREATE TABLE IF NOT EXISTS cost_snapshots (
   order_id             BIGINT REFERENCES orders(id) ON DELETE SET NULL,
   order_item_id        BIGINT REFERENCES order_items(id) ON DELETE SET NULL,
   work_order_id        BIGINT REFERENCES work_orders(id) ON DELETE SET NULL,
-  manual_case_id       BIGINT,
-  carrier_type         TEXT NOT NULL,
   cost_item_id         BIGINT NOT NULL REFERENCES cost_items(id) ON DELETE RESTRICT,
   cost_version_id      BIGINT NOT NULL REFERENCES cost_versions(id) ON DELETE RESTRICT,
   canonical_product_id BIGINT REFERENCES product_catalog(id) ON DELETE RESTRICT,
@@ -152,18 +171,24 @@ CREATE TABLE IF NOT EXISTS cost_snapshots (
   snapshot_stage       TEXT NOT NULL DEFAULT 'estimated',
   source               TEXT NOT NULL DEFAULT 'resolved',
   metadata_json        JSONB NOT NULL DEFAULT '{}'::jsonb,
+  cost_basis_at        TIMESTAMPTZ NOT NULL DEFAULT NOW(),
   resolved_at          TIMESTAMPTZ NOT NULL,
+  superseded_at        TIMESTAMPTZ,
   created_at           TIMESTAMPTZ NOT NULL
 );
 
-CREATE UNIQUE INDEX IF NOT EXISTS uq_cost_snapshots_subject_item
-  ON cost_snapshots (subject_type, subject_id, cost_item_id);
+CREATE UNIQUE INDEX IF NOT EXISTS uq_cost_snapshots_active_subject_item
+  ON cost_snapshots (subject_type, subject_id, cost_item_id)
+  WHERE superseded_at IS NULL;
 CREATE INDEX IF NOT EXISTS idx_cost_snapshots_item
   ON cost_snapshots (order_item_id);
 CREATE INDEX IF NOT EXISTS idx_cost_snapshots_work_order
   ON cost_snapshots (work_order_id);
 CREATE INDEX IF NOT EXISTS idx_cost_snapshots_product
   ON cost_snapshots (canonical_product_id, resolved_at DESC);
+CREATE INDEX IF NOT EXISTS idx_cost_snapshots_active_subject
+  ON cost_snapshots (subject_type, subject_id)
+  WHERE superseded_at IS NULL;
 
 DO $$
 BEGIN
@@ -172,11 +197,6 @@ BEGIN
       ADD CONSTRAINT chk_cost_snapshots_subject_type
       CHECK (subject_type IN ('order_item', 'work_order', 'manual_case'));
   END IF;
-  IF NOT EXISTS (SELECT 1 FROM pg_constraint WHERE conname = 'chk_cost_snapshots_carrier_type') THEN
-    ALTER TABLE cost_snapshots
-      ADD CONSTRAINT chk_cost_snapshots_carrier_type
-      CHECK (carrier_type IN ('affiliate_dash', 'work_order', 'manual', 'shared'));
-  END IF;
   IF NOT EXISTS (SELECT 1 FROM pg_constraint WHERE conname = 'chk_cost_snapshots_stage') THEN
     ALTER TABLE cost_snapshots
       ADD CONSTRAINT chk_cost_snapshots_stage
@@ -194,8 +214,17 @@ BEGIN
   END IF;
 END $$;
 
-COMMENT ON TABLE cost_snapshots IS '成本快照：订单项/工单/人工单实际使用的成本版本与金额';
+COMMENT ON TABLE cost_snapshots IS
+  '成本快照：订单项/工单/人工单实际使用的成本版本与金额；履约方式看 order_items.fulfillment_mode';
 COMMENT ON COLUMN cost_snapshots.source IS 'resolved=成本中心命中；legacy_rule=旧规则回退；manual_input=人工；missing=未匹配';
+COMMENT ON COLUMN cost_snapshots.cost_basis_at IS
+  '解析成本版本时使用的业务时间：订单支付时间/工单验收时间/人工完成时间';
+COMMENT ON COLUMN cost_snapshots.superseded_at IS
+  '被更新的实际成本替代的时间；非空行只保留历史，不再参与成本汇总';
+
+-- ---------------------------------------------------------------------------
+-- 4. 成本异常队列
+-- ---------------------------------------------------------------------------
 
 CREATE TABLE IF NOT EXISTS cost_exceptions (
   id                       BIGSERIAL PRIMARY KEY,
@@ -203,7 +232,6 @@ CREATE TABLE IF NOT EXISTS cost_exceptions (
   subject_id               BIGINT NOT NULL,
   order_item_id            BIGINT REFERENCES order_items(id) ON DELETE SET NULL,
   work_order_id            BIGINT REFERENCES work_orders(id) ON DELETE SET NULL,
-  carrier_type             TEXT NOT NULL DEFAULT '',
   reason                   TEXT NOT NULL,
   status                   TEXT NOT NULL DEFAULT 'open',
   detail_json              JSONB NOT NULL DEFAULT '{}'::jsonb,
@@ -216,6 +244,11 @@ CREATE UNIQUE INDEX IF NOT EXISTS uq_cost_exceptions_subject_reason
   ON cost_exceptions (subject_type, subject_id, reason);
 CREATE INDEX IF NOT EXISTS idx_cost_exceptions_status_created
   ON cost_exceptions (status, created_at DESC);
+CREATE INDEX IF NOT EXISTS idx_cost_exceptions_item
+  ON cost_exceptions (order_item_id, status);
+-- 按履约方式筛异常时走半连接，这条索引让 order_items 侧只扫目标履约方式的行。
+CREATE INDEX IF NOT EXISTS idx_order_items_fulfillment_id
+  ON order_items (fulfillment_mode, id);
 
 DO $$
 BEGIN
@@ -231,14 +264,212 @@ BEGIN
   END IF;
 END $$;
 
-COMMENT ON TABLE cost_exceptions IS '成本缺失/冲突异常队列；open 数量必须在后台暴露';
+COMMENT ON TABLE cost_exceptions IS
+  '成本缺失/冲突异常队列；open 数量必须在后台暴露；履约方式看 order_items.fulfillment_mode';
+
+-- ---------------------------------------------------------------------------
+-- 5. 接单模板过渡列
+-- ---------------------------------------------------------------------------
 
 ALTER TABLE work_product_rules
-  ADD COLUMN IF NOT EXISTS cost_item_id BIGINT REFERENCES cost_items(id) ON DELETE RESTRICT,
-  ADD COLUMN IF NOT EXISTS legacy_cost_source TEXT NOT NULL DEFAULT 'rule_amount';
+  ADD COLUMN IF NOT EXISTS cost_item_id BIGINT REFERENCES cost_items(id) ON DELETE RESTRICT;
 
 CREATE INDEX IF NOT EXISTS idx_work_product_rules_cost_item
   ON work_product_rules (cost_item_id);
 
-COMMENT ON COLUMN work_product_rules.cost_item_id IS '过渡列：非空则金额优先走成本中心；空则回退 reward_amount/unit_price_fen';
-COMMENT ON COLUMN work_product_rules.legacy_cost_source IS '旧成本回退标记：rule_amount=旧规则金额';
+COMMENT ON COLUMN work_product_rules.cost_item_id IS
+  '过渡列：非空则金额优先走成本中心；空则回退 reward_amount/unit_price_fen';
+
+-- ---------------------------------------------------------------------------
+-- 6. 订单项收入固化
+-- ---------------------------------------------------------------------------
+
+ALTER TABLE order_items
+  ADD COLUMN IF NOT EXISTS revenue_amount_fen BIGINT NOT NULL DEFAULT 0,
+  ADD COLUMN IF NOT EXISTS revenue_basis TEXT NOT NULL DEFAULT 'unknown';
+
+DO $$
+BEGIN
+  IF NOT EXISTS (
+    SELECT 1 FROM pg_constraint WHERE conname = 'chk_order_items_revenue_amount'
+  ) THEN
+    ALTER TABLE order_items
+      ADD CONSTRAINT chk_order_items_revenue_amount
+      CHECK (revenue_amount_fen >= 0);
+  END IF;
+  IF NOT EXISTS (
+    SELECT 1 FROM pg_constraint WHERE conname = 'chk_order_items_revenue_basis'
+  ) THEN
+    ALTER TABLE order_items
+      ADD CONSTRAINT chk_order_items_revenue_basis
+      CHECK (revenue_basis IN ('order_total', 'equal_split', 'platform_item', 'manual', 'unknown'));
+  END IF;
+END $$;
+
+-- 历史订单回填：当前生产快照全部是一单一品，仍按通用规则处理多订单项。
+WITH item_counts AS (
+  SELECT order_id, COUNT(*)::int AS item_count
+  FROM order_items
+  GROUP BY order_id
+),
+ranked AS (
+  SELECT
+    oi.id,
+    GREATEST(COALESCE(o.total_amount, 0), 0)::bigint AS order_revenue_fen,
+    ic.item_count,
+    ROW_NUMBER() OVER (PARTITION BY oi.order_id ORDER BY oi.id ASC) AS item_rank
+  FROM order_items oi
+  JOIN orders o ON o.id = oi.order_id
+  JOIN item_counts ic ON ic.order_id = oi.order_id
+)
+UPDATE order_items oi
+SET
+  revenue_amount_fen = CASE
+    WHEN r.item_count = 1 THEN r.order_revenue_fen
+    ELSE (r.order_revenue_fen / r.item_count)
+      + CASE WHEN r.item_rank = r.item_count THEN (r.order_revenue_fen % r.item_count) ELSE 0 END
+  END,
+  revenue_basis = CASE WHEN r.item_count = 1 THEN 'order_total' ELSE 'equal_split' END,
+  updated_at = NOW()
+FROM ranked r
+WHERE oi.id = r.id
+  AND oi.revenue_basis = 'unknown';
+
+CREATE INDEX IF NOT EXISTS idx_order_items_revenue_basis
+  ON order_items (revenue_basis);
+
+COMMENT ON COLUMN order_items.revenue_amount_fen IS
+  '订单项收入（分）：一单一品=订单实付；多品=当前按等额拆分，basis 标明口径';
+COMMENT ON COLUMN order_items.revenue_basis IS
+  'order_total=整单归属于唯一订单项；equal_split=多订单项等额拆分；platform_item=平台明细；manual/unknown=未定';
+
+-- ---------------------------------------------------------------------------
+-- 7. 成本版本页与成本中心页的聚合覆盖索引
+-- ---------------------------------------------------------------------------
+
+CREATE INDEX IF NOT EXISTS idx_work_orders_rule_cover
+  ON work_orders (product_rule_id, order_id, reward_amount);
+CREATE INDEX IF NOT EXISTS idx_order_items_revenue_cover
+  ON order_items (order_id, revenue_amount_fen);
+
+-- 商品目录按分类名 lateral join work_categories，没有索引时每行都扫全表。
+CREATE INDEX IF NOT EXISTS idx_work_categories_name
+  ON work_categories (name);
+
+-- ---------------------------------------------------------------------------
+-- 8. 系统内置成本项
+-- ---------------------------------------------------------------------------
+
+-- 接单平台打手报酬：实际金额取工单 reward_amount，预估阶段不参与；
+-- 版本金额 0 只是占位，让 cost_snapshots.cost_version_id 有稳定引用。
+INSERT INTO cost_items (
+  code, name, cost_model, pricing_mode, unit_name, currency,
+  status, note, created_at, updated_at
+)
+VALUES (
+  'sys_work_order_reward',
+  '接单平台打手报酬（实际结算）',
+  'work_order_reward',
+  'per_order',
+  '单',
+  'CNY',
+  'active',
+  '系统内置：实际金额取工单 reward_amount；版本金额仅占位，不参与预估成本',
+  NOW(),
+  NOW()
+)
+ON CONFLICT (code) DO UPDATE
+SET
+  name = EXCLUDED.name,
+  cost_model = EXCLUDED.cost_model,
+  pricing_mode = EXCLUDED.pricing_mode,
+  note = EXCLUDED.note,
+  updated_at = NOW();
+
+-- 人工客服履约成本：只能由公司人工支出登记给出实际金额。
+INSERT INTO cost_items (
+  code, name, cost_model, pricing_mode, unit_name, currency,
+  status, note, created_at, updated_at
+)
+VALUES (
+  'sys_manual_service',
+  '人工客服履约成本',
+  'manual_actual',
+  'per_order',
+  '单',
+  'CNY',
+  'active',
+  '系统内置：由公司人工支出登记写入 actual 快照；逐单实际金额，不按工时或固定单价',
+  NOW(),
+  NOW()
+)
+ON CONFLICT (code) DO UPDATE
+SET
+  name = EXCLUDED.name,
+  cost_model = EXCLUDED.cost_model,
+  pricing_mode = EXCLUDED.pricing_mode,
+  note = EXCLUDED.note,
+  updated_at = NOW();
+
+-- 自动发货上游采购成本
+INSERT INTO cost_items (
+  code, name, cost_model, pricing_mode, unit_name, currency,
+  status, note, created_at, updated_at
+)
+VALUES (
+  'sys_affiliate_purchase',
+  '自动发货上游采购成本',
+  'affiliate_purchase',
+  'per_item',
+  '件',
+  'CNY',
+  'active',
+  '系统内置：按上游 SKU 固定成本版本结算',
+  NOW(),
+  NOW()
+)
+ON CONFLICT (code) DO UPDATE
+SET
+  name = EXCLUDED.name,
+  cost_model = EXCLUDED.cost_model,
+  pricing_mode = EXCLUDED.pricing_mode,
+  note = EXCLUDED.note,
+  updated_at = NOW();
+
+-- 接单平台：active 占位版本，供 cost_snapshots.cost_version_id 引用。
+INSERT INTO cost_versions (
+  cost_item_id, scope_type, provider, shop_id, canonical_product_id,
+  work_product_rule_id, amount_fen, rate_bps, effective_from, source,
+  status, change_reason, created_at, updated_at
+)
+SELECT
+  ci.id, 'global', '', '', NULL, NULL, 0, 0,
+  '1970-01-01T00:00:00+08:00', 'manual', 'active',
+  '系统占位：接单平台实际金额取工单 reward_amount',
+  NOW(), NOW()
+FROM cost_items ci
+WHERE ci.code = 'sys_work_order_reward'
+ON CONFLICT DO NOTHING;
+
+-- 人工客服：draft 版本，避免在财务未填写金额时静默按 0 计成本。
+INSERT INTO cost_versions (
+  cost_item_id, scope_type, provider, shop_id, canonical_product_id,
+  work_product_rule_id, amount_fen, rate_bps, effective_from, source,
+  status, change_reason, created_at, updated_at
+)
+SELECT
+  ci.id, 'global', '', '', NULL, NULL, 0, 0,
+  '1970-01-01T00:00:00+08:00', 'manual', 'draft',
+  '草稿占位：待财务填写人工客服实际按单成本后启用',
+  NOW(), NOW()
+FROM cost_items ci
+WHERE ci.code = 'sys_manual_service'
+ON CONFLICT DO NOTHING;
+
+-- ---------------------------------------------------------------------------
+-- 9. 财务角色补齐成本中心维护权限
+-- ---------------------------------------------------------------------------
+
+INSERT INTO admin_role_permissions (role_key, permission_key)
+VALUES ('finance', 'finance.config')
+ON CONFLICT (role_key, permission_key) DO NOTHING;
diff --git a/apps/backend/src/db/migrations/105_cost_center_bootstrap_fix.sql b/apps/backend/src/db/migrations/105_cost_center_bootstrap_fix.sql
deleted file mode 100644
index 8f5aceac..00000000
--- a/apps/backend/src/db/migrations/105_cost_center_bootstrap_fix.sql
+++ /dev/null
@@ -1,19 +0,0 @@
--- 105_cost_center_bootstrap_fix.sql
---
--- 背景：
---   103_product_catalog 已执行，里面把 normalized_quantity 直接用 quantity 填充。
---   quantity 长期几乎恒为 1，会让「5个」类 SKU 的归一数量永远无法再通过解析器修正。
---
--- 处理：
---   1. 清空初始化时写入的归一数量，由成本中心「回填数量」用 resolveSkuNameQuantity 重新解析；
---   2. 给财务角色补 finance.config，允许维护商品映射与成本版本（管理员仍绕过权限检查）。
---
--- 该列是本期新列，清空不会影响既有订单字段；回填接口只处理 normalized_quantity IS NULL。
-
-UPDATE order_items
-SET normalized_quantity = NULL
-WHERE normalized_quantity IS NOT NULL;
-
-INSERT INTO admin_role_permissions (role_key, permission_key)
-VALUES ('finance', 'finance.config')
-ON CONFLICT DO NOTHING;
diff --git a/apps/backend/src/db/migrations/109_sales_daily_facts.sql b/apps/backend/src/db/migrations/105_sales_daily_facts.sql
similarity index 76%
rename from apps/backend/src/db/migrations/109_sales_daily_facts.sql
rename to apps/backend/src/db/migrations/105_sales_daily_facts.sql
index 502aa470..32d8e2e4 100644
--- a/apps/backend/src/db/migrations/109_sales_daily_facts.sql
+++ b/apps/backend/src/db/migrations/105_sales_daily_facts.sql
@@ -1,9 +1,11 @@
--- 109_sales_daily_facts.sql —— 店铺 × 日期 × 标准商品 × 履约载体的销售事实。
+-- 105_sales_daily_facts.sql —— 店铺 × 日期 × 标准商品 × 履约方式的销售事实。
 --
--- 口径（2026-10-22 确认）：
+-- 口径（2026-10-23 重写）：
 --   * 销售日 = orders.paid_at 的 Asia/Shanghai 日期；没有 paid_at 时退回 created_at；
---   * 收入 = order_items.revenue_amount_fen（迁移 108 固化），不再直接拿整单金额；
---   * 未归一商品允许 canonical_product_id 为空，单独成“未归一”行，保证总收入能对平；
+--   * 收入 = order_items.revenue_amount_fen（迁移 104 固化），不再直接拿整单金额；
+--   * 履约方式 = order_items.fulfillment_mode，四种：work_order / affiliate_dash /
+--     legacy_platform / manual；未判定的订单项先归 manual 并由定时任务补齐；
+--   * 未归一商品允许 canonical_product_id 为空，单独成「未归一」行，保证总收入对平；
 --   * 成本来自 cost_snapshots 当前有效行；没有快照的订单项计入 cost_missing_count；
 --   * 退款暂未接入，refund_amount_fen 先按 pay_status=refunded 预留。
 
@@ -13,7 +15,7 @@ CREATE TABLE IF NOT EXISTS sales_daily_facts (
   provider             TEXT NOT NULL,
   shop_id              TEXT NOT NULL DEFAULT '',
   canonical_product_id BIGINT REFERENCES product_catalog(id) ON DELETE RESTRICT,
-  carrier_type         TEXT NOT NULL,
+  fulfillment_mode     TEXT NOT NULL,
   order_count          INTEGER NOT NULL DEFAULT 0,
   item_count           INTEGER NOT NULL DEFAULT 0,
   quantity             NUMERIC(14, 3) NOT NULL DEFAULT 0,
@@ -29,14 +31,14 @@ CREATE TABLE IF NOT EXISTS sales_daily_facts (
   updated_at           TIMESTAMPTZ NOT NULL
 );
 
--- canonical_product_id 为空时唯一索引仍要生效，因此用 COALESCE(..., 0) 作为“未归一”桶。
+-- canonical_product_id 为空时唯一索引仍要生效，因此用 COALESCE(..., 0) 作为「未归一」桶。
 CREATE UNIQUE INDEX IF NOT EXISTS uq_sales_daily_facts_key
   ON sales_daily_facts (
     stat_date,
     provider,
     shop_id,
     COALESCE(canonical_product_id, 0),
-    carrier_type
+    fulfillment_mode
   );
 CREATE INDEX IF NOT EXISTS idx_sales_daily_facts_date
   ON sales_daily_facts (stat_date DESC);
@@ -44,15 +46,19 @@ CREATE INDEX IF NOT EXISTS idx_sales_daily_facts_shop
   ON sales_daily_facts (provider, shop_id, stat_date DESC);
 CREATE INDEX IF NOT EXISTS idx_sales_daily_facts_product
   ON sales_daily_facts (canonical_product_id, stat_date DESC);
+CREATE INDEX IF NOT EXISTS idx_sales_daily_facts_mode
+  ON sales_daily_facts (fulfillment_mode, stat_date DESC);
 
 DO $$
 BEGIN
   IF NOT EXISTS (
-    SELECT 1 FROM pg_constraint WHERE conname = 'chk_sales_daily_facts_carrier'
+    SELECT 1 FROM pg_constraint WHERE conname = 'chk_sales_daily_facts_fulfillment_mode'
   ) THEN
     ALTER TABLE sales_daily_facts
-      ADD CONSTRAINT chk_sales_daily_facts_carrier
-      CHECK (carrier_type IN ('affiliate_dash', 'work_order', 'manual', 'shared', 'unknown'));
+      ADD CONSTRAINT chk_sales_daily_facts_fulfillment_mode
+      CHECK (fulfillment_mode IN (
+        'work_order', 'affiliate_dash', 'legacy_platform', 'manual'
+      ));
   END IF;
   IF NOT EXISTS (
     SELECT 1 FROM pg_constraint WHERE conname = 'chk_sales_daily_facts_cost_stage'
@@ -87,9 +93,11 @@ BEGIN
 END $$;
 
 COMMENT ON TABLE sales_daily_facts IS
-  '店铺×日期(Asia/Shanghai)×标准商品×履约载体的销售与成本事实；报表直接读这张表';
+  '店铺×日期(Asia/Shanghai)×标准商品×履约方式的销售与成本事实；报表直接读这张表';
 COMMENT ON COLUMN sales_daily_facts.canonical_product_id IS
-  '标准商品；为空表示渠道映射未命中，报表单独展示“未归一”占比';
+  '标准商品；为空表示渠道映射未命中，报表单独展示「未归一」占比';
+COMMENT ON COLUMN sales_daily_facts.fulfillment_mode IS
+  '履约方式：work_order 接单平台 / affiliate_dash 自动发货 / legacy_platform 历史平台 / manual 人工';
 COMMENT ON COLUMN sales_daily_facts.cost_stage IS
   'missing=全部无成本快照；partial=部分无快照；estimated=全部预估；actual=全部实际';
 COMMENT ON COLUMN sales_daily_facts.cost_missing_count IS
diff --git a/apps/backend/src/db/migrations/111_manual_expense_records.sql b/apps/backend/src/db/migrations/106_manual_expense_records.sql
similarity index 76%
rename from apps/backend/src/db/migrations/111_manual_expense_records.sql
rename to apps/backend/src/db/migrations/106_manual_expense_records.sql
index 067cb985..76b14845 100644
--- a/apps/backend/src/db/migrations/111_manual_expense_records.sql
+++ b/apps/backend/src/db/migrations/106_manual_expense_records.sql
@@ -1,11 +1,15 @@
--- 111_manual_expense_records.sql —— 公司人工支出/垫付登记。
+-- 106_manual_expense_records.sql —— 公司人工支出登记 + 备用金账户编号可读化。
 --
--- 与迁移 100 的 `manual_expenses` 严格区分：
+-- 与迁移 100 的 manual_expenses 严格区分：
 --   manual_expenses        = 客服备用金支出明细（减少客服余额）
---   manual_expense_records = 公司口径人工支出/垫付登记（谁为哪个订单付了多少钱）
+--   manual_expense_records = 公司口径人工支出登记（谁为哪个订单付了多少钱）
 -- 两者通过 fund_expense_id 关联，不合并、不改名、不共用索引名。
 --
--- 本轮只做登记、审批、报销与冲正预留；影子总账分录在 finance_* 阶段接入。
+-- 口径（2026-10-23 确认）：
+--   * 人工成本逐单、按实际金额录入，不按工时、不按固定单量/固定单价；
+--   * 人工支出从客服备用金账户扣款，客服角色只能用本人账户；
+--   * 历史人工支出不补录，只取少量样本做校准；
+--   * 客服必须在上线后 24 小时内为人工单录入成本。
 
 CREATE TABLE IF NOT EXISTS manual_expense_records (
   id                    BIGSERIAL PRIMARY KEY,
@@ -14,7 +18,7 @@ CREATE TABLE IF NOT EXISTS manual_expense_records (
   order_item_id         BIGINT REFERENCES order_items(id) ON DELETE SET NULL,
   work_order_id         BIGINT REFERENCES work_orders(id) ON DELETE SET NULL,
   fund_expense_id       BIGINT REFERENCES manual_expenses(id) ON DELETE SET NULL,
-  cost_item_id          BIGINT REFERENCES cost_items(id) ON DELETE RESTRICT,
+  cost_item_id          BIGINT NOT NULL REFERENCES cost_items(id) ON DELETE RESTRICT,
   expense_category      TEXT NOT NULL DEFAULT 'other',
   amount_fen            BIGINT NOT NULL,
   currency              TEXT NOT NULL DEFAULT 'CNY',
@@ -129,9 +133,9 @@ BEGIN
 END $$;
 
 COMMENT ON TABLE manual_expense_records IS
-  '公司人工支出登记：付款项目/金额/订单号/付款人/收款人/凭证；与客服备用金 manual_expenses 通过 fund_expense_id 关联';
+  '公司人工支出登记：逐单实际金额；付款项目/收款方/凭证/审批；与客服备用金 manual_expenses 通过 fund_expense_id 关联';
 COMMENT ON COLUMN manual_expense_records.fund_expense_id IS
-  '来源客服备用金支出（manual_expenses.id）；公司直付为空';
+  '来源客服备用金支出（manual_expenses.id）；人工支出必须从备用金扣款，因此正常必有值';
 COMMENT ON COLUMN manual_expense_records.cost_item_id IS
   '付款项目 -> cost_items；为空表示未归类，报表必须暴露未归类占比';
 COMMENT ON COLUMN manual_expense_records.evidence_json IS
@@ -159,3 +163,39 @@ CREATE INDEX IF NOT EXISTS idx_manual_expense_events_type
 
 COMMENT ON TABLE manual_expense_events IS
   '人工支出状态变更审计：只追加，不更新、不删除';
+
+-- ---------------------------------------------------------------------------
+-- 备用金账户编号改为可读的 FA-<用户名>
+--
+-- 历史编号是 FA + 10 位随机十六进制（如 FA7683D6B58B），在「备用金」列表里
+-- 无法一眼对应到具体客服。新规则：FA-<大写用户名>，重名追加 -2、-3。
+-- ---------------------------------------------------------------------------
+
+WITH base AS (
+  SELECT
+    fa.id,
+    'FA-' || COALESCE(
+      NULLIF(
+        trim(
+          both '_' FROM upper(
+            regexp_replace(COALESCE(NULLIF(au.username, ''), 'account-' || fa.id), '[^A-Za-z0-9_]+', '_', 'g')
+          )
+        ),
+        ''
+      ),
+      'ACCOUNT-' || fa.id
+    ) AS base_no
+  FROM fund_accounts fa
+  LEFT JOIN admin_users au ON au.id = fa.owner_admin_user_id
+),
+numbered AS (
+  SELECT
+    id,
+    base_no,
+    row_number() OVER (PARTITION BY base_no ORDER BY id) AS rn
+  FROM base
+)
+UPDATE fund_accounts fa
+SET account_no = CASE WHEN n.rn = 1 THEN n.base_no ELSE n.base_no || '-' || n.rn END
+FROM numbered n
+WHERE fa.id = n.id;
diff --git a/apps/backend/src/db/migrations/106_product_mapping_match_keys.sql b/apps/backend/src/db/migrations/106_product_mapping_match_keys.sql
deleted file mode 100644
index 24e13de6..00000000
--- a/apps/backend/src/db/migrations/106_product_mapping_match_keys.sql
+++ /dev/null
@@ -1,72 +0,0 @@
--- 106_product_mapping_match_keys.sql —— 渠道商品映射支持 SKU 名称/系列消歧。
---
--- 背景：91 的 sku_code 会在同一店铺内被多个商品复用，单靠
--- provider + shop_id + platform_sku_code 无法唯一确定标准商品。
---
--- 规则：
---   1. sku_code：编码唯一，直接命中；
---   2. sku_code_name：编码冲突，但编码 + 精确 SKU 名称可唯一命中；
---   3. sku_code_series：编码冲突，但编码 + SKU 名称系列可唯一命中；
---   4. sku_name_series：编码无法提供时，用来源 + 店铺 + SKU 名称系列兜底；
---   5. 任何一层出现多个候选都不自动选择，交由冲突清单人工处理。
-
-ALTER TABLE channel_product_mappings
-  ADD COLUMN IF NOT EXISTS match_type TEXT NOT NULL DEFAULT 'sku_code',
-  ADD COLUMN IF NOT EXISTS platform_sku_name_key TEXT NOT NULL DEFAULT '';
-
-ALTER TABLE order_items
-  ADD COLUMN IF NOT EXISTS normalized_sku_name TEXT,
-  ADD COLUMN IF NOT EXISTS normalized_sku_series TEXT;
-
-DO $$
-BEGIN
-  IF NOT EXISTS (
-    SELECT 1 FROM pg_constraint WHERE conname = 'chk_channel_product_mappings_match_type'
-  ) THEN
-    ALTER TABLE channel_product_mappings
-      ADD CONSTRAINT chk_channel_product_mappings_match_type
-      CHECK (match_type IN ('sku_code', 'sku_code_name', 'sku_code_series', 'sku_name_series'));
-  END IF;
-
-  IF NOT EXISTS (
-    SELECT 1 FROM pg_constraint WHERE conname = 'chk_channel_product_mappings_match_key'
-  ) THEN
-    ALTER TABLE channel_product_mappings
-      ADD CONSTRAINT chk_channel_product_mappings_match_key
-      CHECK (platform_sku_code <> '' OR platform_sku_name_key <> '');
-  END IF;
-END $$;
-
-DROP INDEX IF EXISTS uq_channel_product_mappings_key;
-
-CREATE UNIQUE INDEX IF NOT EXISTS uq_channel_product_mappings_match_key
-  ON channel_product_mappings (
-    provider,
-    shop_id,
-    platform_sku_code,
-    match_type,
-    platform_sku_name_key
-  );
-
-DROP INDEX IF EXISTS idx_channel_product_mappings_lookup;
-CREATE INDEX IF NOT EXISTS idx_channel_product_mappings_match_lookup
-  ON channel_product_mappings (
-    provider,
-    shop_id,
-    platform_sku_code,
-    match_type,
-    platform_sku_name_key,
-    enabled
-  );
-
-CREATE INDEX IF NOT EXISTS idx_order_items_normalized_sku_lookup
-  ON order_items (normalized_sku_name, normalized_sku_series);
-
-COMMENT ON COLUMN channel_product_mappings.match_type IS
-  '映射匹配方式：sku_code / sku_code_name / sku_code_series / sku_name_series';
-COMMENT ON COLUMN channel_product_mappings.platform_sku_name_key IS
-  '归一后的 SKU 名称或数量系列键；sku_code 模式为空串';
-COMMENT ON COLUMN order_items.normalized_sku_name IS
-  'NFKC、小写、去空白后的 SKU 名称，用于编码 + 名称消歧';
-COMMENT ON COLUMN order_items.normalized_sku_series IS
-  '剥离数量并归一同义字后的 SKU 名称系列，用于数量系列匹配';
diff --git a/apps/backend/src/db/migrations/107_order_item_fulfillment_mode.sql b/apps/backend/src/db/migrations/107_order_item_fulfillment_mode.sql
new file mode 100644
index 00000000..2b630a18
--- /dev/null
+++ b/apps/backend/src/db/migrations/107_order_item_fulfillment_mode.sql
@@ -0,0 +1,48 @@
+-- 107_order_item_fulfillment_mode.sql —— 统一履约事实。
+--
+-- 目标：订单页、成本快照、销售事实只读 order_items.fulfillment_mode，
+-- 不再各自用 provider / voucher / task 临时推断。
+--
+-- 设计要点（2026-10-23 重写）：
+--   1. 履约方式只有一个真源，就是 order_items.fulfillment_mode；
+--   2. 商品表、成本项表、快照表都不再存载体字段，避免四处口径分叉；
+--   3. 核销码只是状态，不参与履约判定；
+--   4. 字段允许短暂为 NULL（订单项创建时通常还没有工单/任务），
+--      由统一决策服务 + 定时兜底在 24 小时内补齐。
+--
+-- 本迁移只做 DDL；31 万行回填由 scripts/backfill-fulfillment-mode.ts 分批执行。
+
+ALTER TABLE order_items
+  ADD COLUMN IF NOT EXISTS fulfillment_mode TEXT,
+  ADD COLUMN IF NOT EXISTS fulfillment_decided_at TIMESTAMPTZ,
+  ADD COLUMN IF NOT EXISTS fulfillment_reason TEXT;
+
+DO $$
+BEGIN
+  IF NOT EXISTS (
+    SELECT 1 FROM pg_constraint WHERE conname = 'chk_order_items_fulfillment_mode'
+  ) THEN
+    ALTER TABLE order_items
+      ADD CONSTRAINT chk_order_items_fulfillment_mode
+      CHECK (
+        fulfillment_mode IS NULL
+        OR fulfillment_mode IN (
+          'work_order', 'affiliate_dash', 'legacy_platform', 'manual'
+        )
+      );
+  END IF;
+END $$;
+
+CREATE INDEX IF NOT EXISTS idx_order_items_fulfillment_mode
+  ON order_items (fulfillment_mode, order_id);
+
+CREATE INDEX IF NOT EXISTS idx_order_items_fulfillment_pending
+  ON order_items (order_id)
+  WHERE fulfillment_mode IS NULL;
+
+COMMENT ON COLUMN order_items.fulfillment_mode IS
+  '履约方式：work_order 接单平台 / affiliate_dash 自动发货 / legacy_platform 历史平台（已下线）/ manual 人工；唯一真源，报表与后台只读此字段';
+COMMENT ON COLUMN order_items.fulfillment_decided_at IS
+  '履约方式最后一次判定时间';
+COMMENT ON COLUMN order_items.fulfillment_reason IS
+  '履约方式判定来源：订单同步 / 定时归集 / 历史回填等';
diff --git a/apps/backend/src/db/migrations/107_product_catalog_carrier.sql b/apps/backend/src/db/migrations/107_product_catalog_carrier.sql
deleted file mode 100644
index 34320fae..00000000
--- a/apps/backend/src/db/migrations/107_product_catalog_carrier.sql
+++ /dev/null
@@ -1,95 +0,0 @@
--- 107_product_catalog_carrier.sql —— 标准商品增加「默认履约载体」。
---
--- 背景：商品主数据此前只有游戏/品类分类，成本项目与成本版本却按
--- 自动发货 / 接单平台 / 人工 三类执行载体配置，后台配置时找不到对应商品。
---
--- 口径（2026-10-22 修正）：
---   1. 只有真正走 affiliate-dash（自动发货执行器）的商品才算 affiliate_dash；
---      以 fulfillment_tasks.executor_key = 'affiliate_dash' 的履约证据为准，
---      按 context_json.skuCode / skuName 建标准商品，SKU 固定为 `aff:<skuCode>`。
---   2. 其余标准商品（接单模板初始化出来的 tpl:*）全部是 work_order；
---      渠道映射只代表「平台 SKU 能归一到标准商品」，不代表自动发货。
---   3. manual / shared 不回填，由后台按需手工调整。
---
--- 2026-10-22 生产快照口径：自动发货任务 41,676 条，涉及 28 个 skuCode。
-
-ALTER TABLE product_catalog
-  ADD COLUMN IF NOT EXISTS carrier_type TEXT NOT NULL DEFAULT 'work_order';
-
-DO $$
-BEGIN
-  IF NOT EXISTS (
-    SELECT 1 FROM pg_constraint WHERE conname = 'chk_product_catalog_carrier_type'
-  ) THEN
-    ALTER TABLE product_catalog
-      ADD CONSTRAINT chk_product_catalog_carrier_type
-      CHECK (carrier_type IN ('affiliate_dash', 'work_order', 'manual', 'shared'));
-  END IF;
-END $$;
-
-CREATE INDEX IF NOT EXISTS idx_product_catalog_status_carrier
-  ON product_catalog (status, carrier_type, canonical_name);
-
--- 1) 先把非自动发货商品统一归到接单平台（模板商品、历史误标都回到 work_order）。
-UPDATE product_catalog
-SET carrier_type = 'work_order',
-    updated_at = NOW()
-WHERE carrier_type IS DISTINCT FROM 'work_order'
-  AND canonical_sku NOT LIKE 'aff:%';
-
--- 2) 按自动发货履约证据建商品：每个 skuCode 取一个最像人名的名称。
-WITH auto_task AS (
-  SELECT
-    btrim(context_json->>'skuCode') AS sku_code,
-    btrim(context_json->>'skuName') AS sku_name,
-    created_at
-  FROM fulfillment_tasks
-  WHERE executor_key = 'affiliate_dash'
-    AND btrim(COALESCE(context_json->>'skuCode', '')) <> ''
-    AND btrim(COALESCE(context_json->>'skuName', '')) <> ''
-),
-agg AS (
-  SELECT sku_code, count(*)::int AS task_count, max(created_at) AS last_fulfilled_at
-  FROM auto_task
-  GROUP BY sku_code
-),
-ranked AS (
-  SELECT
-    sku_code,
-    sku_name,
-    row_number() OVER (
-      PARTITION BY sku_code
-      ORDER BY (sku_name <> sku_code) DESC, count(*) DESC, sku_name
-    ) AS rn
-  FROM auto_task
-  GROUP BY sku_code, sku_name
-)
-INSERT INTO product_catalog (
-  canonical_sku, canonical_name, category, unit_name, carrier_type, status,
-  attributes_json, created_at, updated_at
-)
-SELECT
-  'aff:' || ranked.sku_code,
-  ranked.sku_name,
-  '',
-  CASE WHEN ranked.sku_name LIKE '%个' THEN '个' ELSE '件' END,
-  'affiliate_dash',
-  'active',
-  jsonb_build_object(
-    'source', 'affiliate_dash_task',
-    'skuCode', ranked.sku_code,
-    'taskCount', agg.task_count,
-    'lastFulfilledAt', agg.last_fulfilled_at
-  ),
-  NOW(),
-  NOW()
-FROM ranked
-JOIN agg ON agg.sku_code = ranked.sku_code
-WHERE ranked.rn = 1
-ON CONFLICT (canonical_sku) DO UPDATE
-SET carrier_type = 'affiliate_dash',
-    attributes_json = EXCLUDED.attributes_json,
-    updated_at = NOW();
-
-COMMENT ON COLUMN product_catalog.carrier_type IS
-  '默认履约载体：affiliate_dash=自动发货，work_order=接单平台，manual=人工，shared=多载体共用';
diff --git a/apps/backend/src/db/migrations/108_cost_snapshot_runtime.sql b/apps/backend/src/db/migrations/108_cost_snapshot_runtime.sql
deleted file mode 100644
index 4302579f..00000000
--- a/apps/backend/src/db/migrations/108_cost_snapshot_runtime.sql
+++ /dev/null
@@ -1,98 +0,0 @@
--- 108_cost_snapshot_runtime.sql —— 成本快照进入运行层。
---
--- 目标：
---   1. 成本版本按业务发生时间（订单支付/工单验收）解析，而不是迁移或回填时的 now()；
---   2. 预估值可以在结算时被实际值替代，替代后旧快照保留但不再参与汇总；
---   3. 订单项固化收入金额，销售事实可以按订单项归集，不再直接拿整单金额。
---
--- 口径（2026-10-22 确认）：
---   * 收入按 paid_at 归属销售日；成本版本按 paid_at / accepted_at 解析；
---   * 一单一品时订单收入全部归该订单项；多订单项按等额拆分，余数给最后一项，
---     并把 revenue_basis 标成 equal_split，后续有平台明细时再替换；
---   * 历史快照不删除，只把被替代行标记 superseded_at。
-
-ALTER TABLE cost_snapshots
-  ADD COLUMN IF NOT EXISTS cost_basis_at TIMESTAMPTZ,
-  ADD COLUMN IF NOT EXISTS superseded_at TIMESTAMPTZ;
-
--- 已存在的快照发生在旧逻辑的 now()，没有更好的业务时间，先用 resolved_at 兜底。
-UPDATE cost_snapshots
-SET cost_basis_at = resolved_at
-WHERE cost_basis_at IS NULL;
-
-ALTER TABLE cost_snapshots
-  ALTER COLUMN cost_basis_at SET DEFAULT NOW();
-
--- 同一订单项/工单/人工单的同一成本项，只允许一条“当前有效”快照。
-DROP INDEX IF EXISTS uq_cost_snapshots_subject_item;
-CREATE UNIQUE INDEX IF NOT EXISTS uq_cost_snapshots_active_subject_item
-  ON cost_snapshots (subject_type, subject_id, cost_item_id)
-  WHERE superseded_at IS NULL;
-
-CREATE INDEX IF NOT EXISTS idx_cost_snapshots_active_subject
-  ON cost_snapshots (subject_type, subject_id)
-  WHERE superseded_at IS NULL;
-
-COMMENT ON COLUMN cost_snapshots.cost_basis_at IS
-  '解析成本版本时使用的业务时间：订单支付时间/工单验收时间/人工完成时间';
-COMMENT ON COLUMN cost_snapshots.superseded_at IS
-  '被更新的实际成本替代的时间；非空行只保留历史，不再参与成本汇总';
-
-ALTER TABLE order_items
-  ADD COLUMN IF NOT EXISTS revenue_amount_fen BIGINT NOT NULL DEFAULT 0,
-  ADD COLUMN IF NOT EXISTS revenue_basis TEXT NOT NULL DEFAULT 'unknown';
-
-DO $$
-BEGIN
-  IF NOT EXISTS (
-    SELECT 1 FROM pg_constraint WHERE conname = 'chk_order_items_revenue_amount'
-  ) THEN
-    ALTER TABLE order_items
-      ADD CONSTRAINT chk_order_items_revenue_amount
-      CHECK (revenue_amount_fen >= 0);
-  END IF;
-  IF NOT EXISTS (
-    SELECT 1 FROM pg_constraint WHERE conname = 'chk_order_items_revenue_basis'
-  ) THEN
-    ALTER TABLE order_items
-      ADD CONSTRAINT chk_order_items_revenue_basis
-      CHECK (revenue_basis IN ('order_total', 'equal_split', 'platform_item', 'manual', 'unknown'));
-  END IF;
-END $$;
-
--- 历史订单回填：当前生产快照全部是一单一品，仍按通用规则处理多订单项。
-WITH item_counts AS (
-  SELECT order_id, COUNT(*)::int AS item_count
-  FROM order_items
-  GROUP BY order_id
-),
-ranked AS (
-  SELECT
-    oi.id,
-    GREATEST(COALESCE(o.total_amount, 0), 0)::bigint AS order_revenue_fen,
-    ic.item_count,
-    ROW_NUMBER() OVER (PARTITION BY oi.order_id ORDER BY oi.id ASC) AS item_rank
-  FROM order_items oi
-  JOIN orders o ON o.id = oi.order_id
-  JOIN item_counts ic ON ic.order_id = oi.order_id
-)
-UPDATE order_items oi
-SET
-  revenue_amount_fen = CASE
-    WHEN r.item_count = 1 THEN r.order_revenue_fen
-    ELSE (r.order_revenue_fen / r.item_count)
-      + CASE WHEN r.item_rank = r.item_count THEN (r.order_revenue_fen % r.item_count) ELSE 0 END
-  END,
-  revenue_basis = CASE WHEN r.item_count = 1 THEN 'order_total' ELSE 'equal_split' END,
-  updated_at = NOW()
-FROM ranked r
-WHERE oi.id = r.id
-  AND oi.revenue_basis = 'unknown';
-
-CREATE INDEX IF NOT EXISTS idx_order_items_revenue_basis
-  ON order_items (revenue_basis);
-
-COMMENT ON COLUMN order_items.revenue_amount_fen IS
-  '订单项收入（分）：一单一品=订单实付；多品=当前按等额拆分，basis 标明口径';
-COMMENT ON COLUMN order_items.revenue_basis IS
-  'order_total=整单归属于唯一订单项；equal_split=多订单项等额拆分；platform_item=平台明细；manual/unknown=未定';
diff --git a/apps/backend/src/db/migrations/110_system_cost_items.sql b/apps/backend/src/db/migrations/110_system_cost_items.sql
deleted file mode 100644
index 260a79ca..00000000
--- a/apps/backend/src/db/migrations/110_system_cost_items.sql
+++ /dev/null
@@ -1,95 +0,0 @@
--- 110_system_cost_items.sql —— 内置接单平台实际报酬与人工客服成本项。
---
--- 背景：接单平台的成本金额在验收时已经由 work_orders.reward_amount 确定，
--- 不需要财务再按模板猜一个预估价；人工客服成本则必须由人工支出登记给出。
---
--- 口径：
---   1. sys_work_order_reward 由系统维护，预估阶段不参与；工单验收后按实际
---      reward_amount 直接写 actual 快照。版本金额 0 只是占位，不参与汇总。
---   2. sys_manual_service 由人工支出登记使用，先建草稿版本占位；
---      财务填写真实按单金额并启用后，才参与成本解析。
-
-INSERT INTO cost_items (
-  code, name, cost_type, carrier_type, pricing_mode, unit_name, currency,
-  status, note, created_at, updated_at
-)
-VALUES (
-  'sys_work_order_reward',
-  '接单平台打手报酬（实际结算）',
-  'work_order_reward',
-  'work_order',
-  'per_order',
-  '单',
-  'CNY',
-  'active',
-  '系统内置：实际金额取工单 reward_amount；版本金额仅占位，不参与预估成本',
-  NOW(),
-  NOW()
-)
-ON CONFLICT (code) DO UPDATE
-SET
-  name = EXCLUDED.name,
-  cost_type = EXCLUDED.cost_type,
-  carrier_type = EXCLUDED.carrier_type,
-  pricing_mode = EXCLUDED.pricing_mode,
-  note = EXCLUDED.note,
-  updated_at = NOW();
-
-INSERT INTO cost_items (
-  code, name, cost_type, carrier_type, pricing_mode, unit_name, currency,
-  status, note, created_at, updated_at
-)
-VALUES (
-  'sys_manual_service',
-  '人工客服履约成本',
-  'manual_service',
-  'manual',
-  'per_order',
-  '单',
-  'CNY',
-  'active',
-  '系统内置：由公司人工支出登记写入 actual 快照；草稿版本待财务填写实际单价',
-  NOW(),
-  NOW()
-)
-ON CONFLICT (code) DO UPDATE
-SET
-  name = EXCLUDED.name,
-  cost_type = EXCLUDED.cost_type,
-  carrier_type = EXCLUDED.carrier_type,
-  pricing_mode = EXCLUDED.pricing_mode,
-  note = EXCLUDED.note,
-  updated_at = NOW();
-
--- 接单平台：active 占位版本，供 cost_snapshots.cost_version_id 引用。
-INSERT INTO cost_versions (
-  cost_item_id, scope_type, provider, shop_id, canonical_product_id,
-  work_product_rule_id, amount_fen, rate_bps, effective_from, source,
-  status, change_reason, created_at, updated_at
-)
-SELECT
-  ci.id, 'global', '', '', NULL, NULL, 0, 0,
-  '1970-01-01T00:00:00+08:00', 'manual', 'active',
-  '系统占位：接单平台实际金额取工单 reward_amount',
-  NOW(), NOW()
-FROM cost_items ci
-WHERE ci.code = 'sys_work_order_reward'
-ON CONFLICT DO NOTHING;
-
--- 人工客服：draft 版本，避免在财务未填写金额时静默按 0 计成本。
-INSERT INTO cost_versions (
-  cost_item_id, scope_type, provider, shop_id, canonical_product_id,
-  work_product_rule_id, amount_fen, rate_bps, effective_from, source,
-  status, change_reason, created_at, updated_at
-)
-SELECT
-  ci.id, 'global', '', '', NULL, NULL, 0, 0,
-  '1970-01-01T00:00:00+08:00', 'manual', 'draft',
-  '草稿占位：待财务填写人工客服实际按单成本后启用',
-  NOW(), NOW()
-FROM cost_items ci
-WHERE ci.code = 'sys_manual_service'
-ON CONFLICT DO NOTHING;
-
-COMMENT ON COLUMN cost_items.code IS
-  '成本项编码；sys_ 前缀为系统内置项，业务侧不要删除或改编码';
diff --git a/apps/backend/src/db/migrations/112_cost_reporting_indexes.sql b/apps/backend/src/db/migrations/112_cost_reporting_indexes.sql
deleted file mode 100644
index f1f160c2..00000000
--- a/apps/backend/src/db/migrations/112_cost_reporting_indexes.sql
+++ /dev/null
@@ -1,14 +0,0 @@
--- 112_cost_reporting_indexes.sql —— 成本版本页按接单规则汇总用户实付。
---
--- 背景：成本版本列表需要同时展示「用户实付均价」和「模板发单金额」。
--- 原查询按 work_orders.product_rule_id 聚合约 11 万工单 + 31 万订单项，
--- 没有覆盖索引时需要 1 秒以上；以下两个覆盖索引把聚合降到约 150 毫秒。
-
-CREATE INDEX IF NOT EXISTS idx_work_orders_rule_cover
-  ON work_orders (product_rule_id)
-  INCLUDE (order_item_id)
-  WHERE order_item_id IS NOT NULL;
-
-CREATE INDEX IF NOT EXISTS idx_order_items_revenue_cover
-  ON order_items (id)
-  INCLUDE (order_id, revenue_amount_fen);
diff --git a/apps/backend/src/db/migrations/113_fund_account_no_readable.sql b/apps/backend/src/db/migrations/113_fund_account_no_readable.sql
deleted file mode 100644
index a9d25c92..00000000
--- a/apps/backend/src/db/migrations/113_fund_account_no_readable.sql
+++ /dev/null
@@ -1,35 +0,0 @@
--- 113_fund_account_no_readable.sql —— 备用金账户编号改为可读的 FA-<用户名>。
---
--- 背景：历史编号是 FA + 10 位随机十六进制（如 FA7683D6B58B），
--- 在「备用金」列表里无法一眼对应到具体客服。
--- 新规则：FA-<大写用户名>，例如 FA-KF_2、FA-JIAJIALI；
--- 同一用户名重名时追加 -2、-3。
-
-WITH base AS (
-  SELECT
-    fa.id,
-    'FA-' || COALESCE(
-      NULLIF(
-        trim(
-          both '_' FROM upper(
-            regexp_replace(COALESCE(NULLIF(au.username, ''), 'account-' || fa.id), '[^A-Za-z0-9_]+', '_', 'g')
-          )
-        ),
-        ''
-      ),
-      'ACCOUNT-' || fa.id
-    ) AS base_no
-  FROM fund_accounts fa
-  LEFT JOIN admin_users au ON au.id = fa.owner_admin_user_id
-),
-numbered AS (
-  SELECT
-    id,
-    base_no,
-    row_number() OVER (PARTITION BY base_no ORDER BY id) AS rn
-  FROM base
-)
-UPDATE fund_accounts fa
-SET account_no = CASE WHEN n.rn = 1 THEN n.base_no ELSE n.base_no || '-' || n.rn END
-FROM numbered n
-WHERE fa.id = n.id;
diff --git a/apps/backend/src/db/migrations/114_order_item_fulfillment_mode.sql b/apps/backend/src/db/migrations/114_order_item_fulfillment_mode.sql
deleted file mode 100644
index cfdfcb27..00000000
--- a/apps/backend/src/db/migrations/114_order_item_fulfillment_mode.sql
+++ /dev/null
@@ -1,45 +0,0 @@
--- 114_order_item_fulfillment_mode.sql
---
--- 统一履约事实：订单页、成本快照、销售事实只读 order_items.fulfillment_mode，
--- 不再各自用 provider / voucher / task 临时推断。
---
--- 说明：
--- 1. 本迁移只做 DDL；31 万行回填由 scripts/backfill-fulfillment-mode.ts 分批执行。
--- 2. product_catalog / cost_items / cost_snapshots / sales_daily_facts 的
---    carrier_type CHECK 目前都不接受 legacy_platform，必须同时放宽，
---    否则历史平台数据写入会在运行时报错。
-
-ALTER TABLE order_items
-  ADD COLUMN IF NOT EXISTS fulfillment_mode text,
-  ADD COLUMN IF NOT EXISTS fulfillment_decided_at timestamptz,
-  ADD COLUMN IF NOT EXISTS fulfillment_reason text;
-
-ALTER TABLE order_items
-  ADD CONSTRAINT chk_order_items_fulfillment_mode
-  CHECK (
-    fulfillment_mode IS NULL
-    OR fulfillment_mode IN ('work_order', 'affiliate_dash', 'legacy_platform', 'manual')
-  );
-
-CREATE INDEX IF NOT EXISTS idx_order_items_fulfillment_mode
-  ON order_items (fulfillment_mode, order_id);
-
-CREATE INDEX IF NOT EXISTS idx_order_items_fulfillment_pending
-  ON order_items (order_id)
-  WHERE fulfillment_mode IS NULL;
-
-ALTER TABLE product_catalog DROP CONSTRAINT IF EXISTS chk_product_catalog_carrier_type;
-ALTER TABLE product_catalog ADD CONSTRAINT chk_product_catalog_carrier_type
-  CHECK (carrier_type IN ('affiliate_dash', 'work_order', 'legacy_platform', 'manual', 'shared'));
-
-ALTER TABLE cost_items DROP CONSTRAINT IF EXISTS chk_cost_items_carrier_type;
-ALTER TABLE cost_items ADD CONSTRAINT chk_cost_items_carrier_type
-  CHECK (carrier_type IN ('affiliate_dash', 'work_order', 'legacy_platform', 'manual', 'shared'));
-
-ALTER TABLE cost_snapshots DROP CONSTRAINT IF EXISTS chk_cost_snapshots_carrier_type;
-ALTER TABLE cost_snapshots ADD CONSTRAINT chk_cost_snapshots_carrier_type
-  CHECK (carrier_type IN ('affiliate_dash', 'work_order', 'legacy_platform', 'manual', 'shared'));
-
-ALTER TABLE sales_daily_facts DROP CONSTRAINT IF EXISTS chk_sales_daily_facts_carrier;
-ALTER TABLE sales_daily_facts ADD CONSTRAINT chk_sales_daily_facts_carrier
-  CHECK (carrier_type IN ('affiliate_dash', 'work_order', 'legacy_platform', 'manual', 'shared', 'unknown'));