diff --git a/docs/tutorials/customize.md b/docs/tutorials/customize.md
index 293c6f2..f7d76f4 100644
--- a/docs/tutorials/customize.md
+++ b/docs/tutorials/customize.md
@@ -7,7 +7,7 @@ You can customize schema classes on your own code, if you need for some reasons.
 For example, you can make `Attribute` allow `attribute.json` not to require `description` field as follows:
 
 ```python title="custom_attribute.py"
-from dataclasses import dataclass
+from attrs import define, field
 from typing import Any
 
 from typing_extensions import Self
@@ -16,27 +16,16 @@ from t4_devkit.schema import SCHEMAS, SchemaName, SchemaBase
 from t4_devkit.common.io import load_json
 
 
-@dataclass
+@define
 @SCHEMAS.register(SchemaName.ATTRIBUTE, force=True)
 class CustomAttribute(SchemaBase):
-    """Custom Attribute class ignoring if there is no description field.
-    Note that `description` field is mandatory by the original definition.
+    """Custom Attribute class ignoring if there is no `description` field.
+    Note that `description` field is mandatory in the original `Attribute` class.
+
+    `@SCHEMAS.register(SchemaName.ATTRIBUTE, force=True)` performs that
+    it forces to update the attribute table in the schema registry.
     """
 
-    token: str
     name: str
-    description: str | None
-
-    @classmehod
-    def from_json(cls, filepath: str) -> list[Self]:
-        objs: list[Self] = []
-
-        record_list: list[dict[str, Any]] = load_json(filepath)
-        for record in record_list:
-            token: str = record["token"]
-            name: str = record["name"]
-            # Return None if record does not have description field
-            description: str | None = record.get("description")
-            objs.append(cls(token=token, name=name, description=description))
-        return objs
+    description: str | None = field(default=None)
 ```