diff --git a/digdag-docs/src/operators/py.md b/digdag-docs/src/operators/py.md
index be466f6bf3..62eb3d9db7 100644
--- a/digdag-docs/src/operators/py.md
+++ b/digdag-docs/src/operators/py.md
@@ -76,6 +76,114 @@ See [Python API documents](../python_api.html) for details including variable ma
     print("simple execution")
   ```
 
+  You can pass arguments to class for initialization　by defining arguments under the `py>:` operation as the following:
+  ```yaml
+  # sample.dig
+  +some_task:
+    py>: tasks.MyWorkflow.my_task
+    required1_1: awesome execution
+    required1_2: "awesome execution"
+    required2: {a: "a"}
+    required3: 1
+    required4: 1.0
+    required5: [a, 1, 1.0, "a"]
+  ```
+
+  Also, you can do the same thing using `_export` as the following:
+  ```yaml
+  # sample.dig
+  +some_task:
+    _export:
+      required1_1: awesome execution
+      required1_2: "awesome execution"
+      required2: {a: "a"}
+      required3: 1
+      required4: 1.0
+      required5: [a, 1, 1.0, "a"]
+    py>: tasks.MyWorkflow.my_task
+  ```
+
+  This example assume following Python script:
+
+  ```python
+  # tasks.py
+  from typing import Union
+
+
+  class MyWorkflow(object):
+      def __init__(
+        self,
+        required1_1: str,
+        required1_2: str,
+        required2: dict[str, str],
+        required3: int,
+        required4: float,
+        required5: list[Union[str, int, float]]
+      ):
+          print(f"{required1_1} same as {required1_2}")
+          self.arg2 = required2
+          print(f"{float(required3)} same as {required4}")
+          self.arg5 = required5
+      
+      def my_task(self):
+          pass
+  ```
+
+  Or, you can pass arguments to function as the following:
+
+  ```yaml
+  # sample.dig
+  +some_task:
+    py>: simple_tasks.my_func
+    required1: simple execution
+    required2: {a: "a"}
+  ```
+
+  ```yaml
+  # simple_sample.dig
+  +some_task:
+    _export:
+      required1: simple execution
+      required2: {a: "a"}
+    py>: simple_tasks.my_func
+  ```
+
+  ```python
+  # simple_tasks.py
+  def my_func(required1: str, required2: dict[str, str]):
+    print(f"{required1}: {required2}")
+  ```
+
+  Finally, you can pass combination (must have different names) of class and mehtod arguments to Python script as the following:
+  
+  ```yaml
+  # sample.dig
+  +some_task:
+    py>: tasks.MyWorkflow.my_task
+    required_class_arg: awesome execution
+    required_method_arg: ["a", "b"]
+  ```
+
+  ```yaml
+  # sample.dig
+  +some_task:
+    _export:
+      required_class_arg: awesome execution
+      required_method_arg: ["a", "b"]
+    py>: tasks.MyWorkflow.my_task
+  ```
+
+  ```python
+  # tasks.py
+  class MyWorkflow:
+    def __init__(self, required_class_arg: str):
+      self.arg = required_class_arg
+    
+    def my_task(self, required_method_arg: list[str]):
+      print(f"{self.arg}: {required_method_arg}")
+  ```
+
+
 * **python**: PATH STRING or COMMAND ARGUMENTS LIST
 
   The python defaults to `python`. If an alternate python and options are desired, use the `python` option.
diff --git a/examples/python_args.dig b/examples/python_args.dig
index 86cf108771..0cec0ac8f5 100644
--- a/examples/python_args.dig
+++ b/examples/python_args.dig
@@ -19,3 +19,9 @@ timezone: UTC
   key1: "a"
   key2: "val2"
 
++e:
+  _export:
+    key1: "a"
+    key2: {b: "c"}
+  py>: tasks.python_args.exported_arguments
+