diff --git a/server/aps_domain/rush.py b/server/aps_domain/rush.py
index eb2f952..f25c153 100644
--- a/server/aps_domain/rush.py
+++ b/server/aps_domain/rush.py
@@ -308,11 +308,37 @@ def _strategy_cn(s: str) -> str:
     }.get(s, s)
 
 
+class _FlexSolveResultView:
+    """PoolEngine 的 dict 结果适配成主轨结果对象的属性读取口径。
+
+    workflow 的 rush.apply 回复文案按属性读 versionNo/orderCount/conflictCount/
+    totalTardiness；柔性引擎返回 dict，这里做一次只读视图转换。
+    """
+
+    def __init__(self, payload: dict[str, Any]) -> None:
+        self.versionNo = payload.get("versionNo")
+        self.orderCount = int(payload.get("orderCount") or 0)
+        self.conflictCount = int(payload.get("conflictCount") or 0)
+        self.totalTardiness = float(payload.get("totalTardiness") or 0.0)
+        self.solveStatus = payload.get("solveStatus")
+
+
 def apply_rush(world: World, next_id, payload: dict[str, Any]) -> dict[str, Any]:
     """采用插单：写主干订单池 + 生成 DRAFT 排产版本。调用方负责 checkpoint/审计。"""
     rush = resolve_rush_payload(world, payload.get("payload") or payload)
     strategy = str(payload.get("strategy") or "DELIVERY_FIRST").upper()
     order = inject_rush_trunk(world, next_id, rush)
+    planning_context = world.get("planningContext") or {}
+    if planning_context.get("trialOnly"):
+        # 试排世界必须走柔性能力池引擎：主轨 RULE 引擎读不到 flex* 主数据，
+        # 全量重排只会产出 0 工单的空壳版本，排产推演/查看排产结果都看不到急单。
+        # 改为先把主干急单桥接进柔性订单池，再重跑 PoolEngine 试排（与 run_flex_schedule 同轨）。
+        from server.aps_domain.orders import sync_sales_orders_to_flex
+        from server.engines import PoolEngine
+
+        sync_sales_orders_to_flex(world)
+        flex_result = PoolEngine().solve(world, next_id, trial=True)
+        return {"order": order, "result": _FlexSolveResultView(flex_result), "strategy": strategy}
     # 正式试排写草稿版本
     from server.aps_domain.constraints import engine_constraint_flags
     from server.aps_domain.params import get_schedule_params