feat：意图识别优化

app/schemas/requirement.py

@@ -2,9 +2,10 @@
 
 from typing import Literal, Self
 
-from pydantic import Field, model_validator
+from pydantic import Field, field_validator, model_validator
 
 from .base import ExtensibleModel, JsonObject, NonBlankString
+from .operations import OPERATION_GUIDE, normalize_operation
 
 Intent = Literal["create_screen", "modify_screen", "qa", "unclear"]
 TargetScope = Literal["nodes", "canvas", "screen", "unresolved"]
@@ -22,14 +23,18 @@ class RequirementChange(ExtensibleModel):
     target_description: NonBlankString | None = Field(
         default=None, description="目标的自然语言描述，例如右侧区域、用户所说的那个图"
     )
-    operation: NonBlankString = Field(
-        description="需求操作，例如 move、resize、add、delete、restyle；允许后续扩展"
-    )
+    operation: NonBlankString = Field(description=f"需求操作。{OPERATION_GUIDE}")
     parameters: JsonObject = Field(
         default_factory=dict,
         description="需求参数，例如 delta_x=50；保留用户要求的单位与相对变化语义",
     )
 
+    @field_validator("operation")
+    @classmethod
+    def normalize_operation_name(cls, value: str) -> str:
+        # 在协议层统一同义操作名，模型输出、测试构造和检查点恢复都经过这里
+        return normalize_operation(value)
+
     @model_validator(mode="after")
     def validate_target(self) -> Self:
         if len(self.target_ids) != len(set(self.target_ids)):