diff --git a/requirements.txt b/requirements.txt
index f149a60..e387fee 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -13,7 +13,7 @@ click==8.1.7 ; python_version >= "3.10" and python_version < "4.0"
 colorama==0.4.6 ; python_version >= "3.10" and python_version < "4.0" and platform_system == "Windows"
 dataclasses-json==0.6.7 ; python_version >= "3.10" and python_version < "4.0"
 distro==1.9.0 ; python_version >= "3.10" and python_version < "4.0"
-duckduckgo-search==6.3.7 ; python_version >= "3.10" and python_version < "4.0"
+duckduckgo-search==7.2.1 ; python_version >= "3.10" and python_version < "4.0"
 exceptiongroup==1.2.2 ; python_version >= "3.10" and python_version < "3.11"
 frozenlist==1.4.1 ; python_version >= "3.10" and python_version < "4.0"
 greenlet==3.1.1 ; python_version < "3.13" and (platform_machine == "aarch64" or platform_machine == "ppc64le" or platform_machine == "x86_64" or platform_machine == "amd64" or platform_machine == "AMD64" or platform_machine == "win32" or platform_machine == "WIN32") and python_version >= "3.10"
@@ -28,9 +28,10 @@ jsonpointer==3.0.0 ; python_version >= "3.10" and python_version < "4.0"
 langchain-community==0.3.13 ; python_version >= "3.10" and python_version < "4.0"
 langchain-core==0.3.28 ; python_version >= "3.10" and python_version < "4.0"
 langchain-openai==0.2.14 ; python_version >= "3.10" and python_version < "4.0"
-langchain-text-splitters==0.3.5 ; python_version >= "3.10" and python_version < "4.0"
+langchain-text-splitters==0.3.4 ; python_version >= "3.10" and python_version < "4.0"
 langchain==0.3.13 ; python_version >= "3.10" and python_version < "4.0"
 langsmith==0.1.147 ; python_version >= "3.10" and python_version < "4.0"
+lxml==5.3.0 ; python_version >= "3.10" and python_version < "4.0"
 marshmallow==3.22.0 ; python_version >= "3.10" and python_version < "4.0"
 multidict==6.1.0 ; python_version >= "3.10" and python_version < "4.0"
 mypy-extensions==1.0.0 ; python_version >= "3.10" and python_version < "4.0"
@@ -38,7 +39,7 @@ numpy==1.26.4 ; python_version >= "3.10" and python_version < "4.0"
 openai==1.58.1 ; python_version >= "3.10" and python_version < "4.0"
 orjson==3.10.7 ; python_version >= "3.10" and python_version < "4.0" and platform_python_implementation != "PyPy"
 packaging==24.1 ; python_version >= "3.10" and python_version < "4.0"
-primp==0.8.1 ; python_version >= "3.10" and python_version < "4.0"
+primp==0.10.0 ; python_version >= "3.10" and python_version < "4.0"
 propcache==0.2.0 ; python_version >= "3.10" and python_version < "4.0"
 pydantic-core==2.23.4 ; python_version >= "3.10" and python_version < "4.0"
 pydantic-settings==2.5.2 ; python_version >= "3.10" and python_version < "4.0"
diff --git a/src/sc_system_ai/agents/classify_agent.py b/src/sc_system_ai/agents/classify_agent.py
index a10d377..f03aac5 100644
--- a/src/sc_system_ai/agents/classify_agent.py
+++ b/src/sc_system_ai/agents/classify_agent.py
@@ -1,14 +1,17 @@
-from collections.abc import Iterator
-from typing import Any, cast
+from collections.abc import AsyncIterator
+from typing import cast
 
 from langchain_openai import AzureChatOpenAI
 
 # from sc_system_ai.agents.tools import magic_function
-from sc_system_ai.agents.search_school_data_agent import SearchSchoolDataAgentResponse
 from sc_system_ai.agents.tools.calling_dummy_agent import calling_dummy_agent
-from sc_system_ai.agents.tools.calling_search_school_data_agent import calling_search_school_data_agent
+from sc_system_ai.agents.tools.calling_search_school_data_agent import (
+    CallingSearchSchoolDataAgent,
+    calling_search_school_data_agent,
+)
+from sc_system_ai.agents.tools.calling_small_talk_agent import calling_small_talk_agent
 from sc_system_ai.agents.tools.classify_role import classify_role
-from sc_system_ai.template.agent import Agent, AgentResponse
+from sc_system_ai.template.agent import Agent, AgentResponse, StreamingAgentResponse
 from sc_system_ai.template.ai_settings import llm
 from sc_system_ai.template.calling_agent import CallingAgent
 from sc_system_ai.template.user_prompts import User
@@ -17,14 +20,14 @@
     # magic_function,
     classify_role,
     calling_dummy_agent,
-    calling_search_school_data_agent
+    calling_search_school_data_agent,
+    calling_small_talk_agent,
 ]
 
 classify_agent_info = """あなたの役割は適切なエージェントを選択し処理を引き継ぐことです。
+あなたがユーザーと会話を行ってはいけません。
 ユーザーの入力、会話の流れから適切なエージェントを選択してください。
 引き継いだエージェントが処理を完了するまで、そのエージェントがユーザーと会話を続けるようにしてください。
-
-適切なエージェントの選択、呼び出しができなかった場合は、そのままユーザーとの会話を続けてください。
 """
 
 # agentクラスの作成
@@ -35,14 +38,10 @@ def __init__(
             self,
             llm: AzureChatOpenAI = llm,
             user_info: User | None = None,
-            is_streaming: bool = True,
-            return_length: int = 5
     ):
         super().__init__(
             llm=llm,
             user_info=user_info if user_info is not None else User(),
-            is_streaming=is_streaming,
-            return_length=return_length
         )
         self.assistant_info = classify_agent_info
         super().set_assistant_info(self.assistant_info)
@@ -53,16 +52,33 @@ def set_tools(self, tools: list) -> None:
         for tool in tools:
             if isinstance(tool, CallingAgent):
                 tool.set_user_info(self.user_info)
-
         super().set_tools(tools)
 
-    def invoke(self, message: str) -> Iterator[str | AgentResponse | SearchSchoolDataAgentResponse]:
-        if self.is_streaming:
-            yield from super().invoke(message)
-        else:
-            # ツールの出力をそのまま返却
-            resp = cast(dict[str, Any], next(super().invoke(message)))
-            yield resp["output"]
+    def invoke(self, message: str) -> AgentResponse:
+        # toolの出力がAgentReaponseで返って来るので整形
+        for tool in self.tool.tools:
+            if isinstance(tool, CallingAgent):
+                tool.cancel_streaming()
+        resp = super().invoke(message)
+        resp.document_id = self._doc_id_checker()
+        return resp
+
+    def _doc_id_checker(self) -> list[str] | None:
+        """
+        ドキュメントIDが存在するか確認する
+        """
+        for tool in self.tool.tools:
+            if isinstance(tool, CallingSearchSchoolDataAgent):
+                if tool.document_id is not None:
+                    return tool.document_id
+        return None
+
+    async def stream(self, message: str, return_length: int = 5) -> AsyncIterator[StreamingAgentResponse]:
+        for tool in self.tool.tools:
+            if isinstance(tool, CallingAgent):
+                tool.setup_streaming(self.queue)
+        async for output in super().stream(message, return_length):
+            yield output
 
 
 if __name__ == "__main__":
@@ -79,7 +95,7 @@ def invoke(self, message: str) -> Iterator[str | AgentResponse | SearchSchoolDat
     user_info.conversations.add_conversations_list(history)
 
     while True:
-        classify_agent = ClassifyAgent(user_info=user_info, is_streaming=False)
+        classify_agent = ClassifyAgent(user_info=user_info)
         # classify_agent.display_agent_info()
         # print(main_agent.get_agent_prompt())
         # classify_agent.display_agent_prompt()
@@ -89,7 +105,7 @@ def invoke(self, message: str) -> Iterator[str | AgentResponse | SearchSchoolDat
             break
 
         # 通常の呼び出し
-        resp = next(classify_agent.invoke(user))
+        resp = classify_agent.invoke(user)
         print(resp)
 
         # ストリーミング呼び出し
@@ -97,9 +113,9 @@ def invoke(self, message: str) -> Iterator[str | AgentResponse | SearchSchoolDat
         #     print(output)
         # resp = classify_agent.get_response()
 
-        if type(resp) is dict:
+        if type(resp) is AgentResponse:
             new_conversation = [
                 ("human", user),
-                ("ai", resp["output"])
+                ("ai", cast(str,resp.output))
             ]
             user_info.conversations.add_conversations_list(new_conversation)
diff --git a/src/sc_system_ai/agents/dummy_agent.py b/src/sc_system_ai/agents/dummy_agent.py
index b864df3..93c7363 100644
--- a/src/sc_system_ai/agents/dummy_agent.py
+++ b/src/sc_system_ai/agents/dummy_agent.py
@@ -1,4 +1,5 @@
 # ダミーのエージェント
+from typing import cast
 
 from langchain_openai import AzureChatOpenAI
 
@@ -49,14 +50,10 @@ def __init__(
             self,
             llm: AzureChatOpenAI = llm,
             user_info: User | None = None,
-            is_streaming: bool = True,
-            return_length: int = 5
     ):
         super().__init__(
             llm=llm,
             user_info=user_info if user_info is not None else User(),
-            is_streaming=is_streaming,
-            return_length=return_length
         )
         self.assistant_info = dummy_agent_info
         super().set_assistant_info(self.assistant_info)
@@ -96,9 +93,9 @@ def __init__(
             print(output)
         resp = dummy_agent.get_response()
 
-        if type(resp) is dict:
+        if resp.error is not None:
             new_conversation = [
                 ("human", user),
-                ("ai", resp["output"])
+                ("ai", cast(str, resp.output))
             ]
             user_info.conversations.add_conversations_list(new_conversation)
diff --git a/src/sc_system_ai/agents/main_agent.py b/src/sc_system_ai/agents/main_agent.py
index 92888e2..ab76f31 100644
--- a/src/sc_system_ai/agents/main_agent.py
+++ b/src/sc_system_ai/agents/main_agent.py
@@ -16,14 +16,10 @@ def __init__(
             self,
             llm: AzureChatOpenAI = llm,
             user_info: User | None = None,
-            is_streaming: bool = True,
-            return_length: int = 5
     ):
         super().__init__(
             llm=llm,
             user_info=user_info if user_info is not None else User(),
-            is_streaming=is_streaming,
-            return_length=return_length
         )
         self.assistant_info = main_agent_info
         super().set_assistant_info(self.assistant_info)
@@ -43,9 +39,9 @@ def __init__(
     user_info = User(name=user_name, major=user_major)
     user_info.conversations.add_conversations_list(history)
 
-    main_agent = MainAgent(user_info=user_info, is_streaming=False)
+    main_agent = MainAgent(user_info=user_info)
     main_agent.display_agent_info()
     # print(main_agent.get_agent_prompt())
     main_agent.display_agent_prompt()
-    print(next(main_agent.invoke("magic function に３をいれて")))
+    print(main_agent.invoke("magic function に３をいれて"))
 
diff --git a/src/sc_system_ai/agents/search_school_data_agent.py b/src/sc_system_ai/agents/search_school_data_agent.py
index d780c31..e544439 100644
--- a/src/sc_system_ai/agents/search_school_data_agent.py
+++ b/src/sc_system_ai/agents/search_school_data_agent.py
@@ -1,11 +1,10 @@
-from collections.abc import Iterator
-from typing import cast
+from collections.abc import AsyncIterator
 
 from langchain_openai import AzureChatOpenAI
 
 # from sc_system_ai.agents.tools import magic_function
 from sc_system_ai.agents.tools.search_school_data import search_school_database_cosmos
-from sc_system_ai.template.agent import Agent, AgentResponse
+from sc_system_ai.template.agent import Agent, AgentResponse, StreamingAgentResponse
 from sc_system_ai.template.ai_settings import llm
 from sc_system_ai.template.user_prompts import User
 
@@ -19,9 +18,6 @@
 ## 学校の情報
 """
 
-class SearchSchoolDataAgentResponse(AgentResponse):
-    document_id: list[str]
-
 # agentクラスの作成
 
 class SearchSchoolDataAgent(Agent):
@@ -29,32 +25,34 @@ def __init__(
             self,
             llm: AzureChatOpenAI = llm,
             user_info: User | None = None,
-            is_streaming: bool = True,
-            return_length: int = 5
     ):
         super().__init__(
             llm=llm,
             user_info=user_info if user_info is not None else User(),
-            is_streaming=is_streaming,
-            return_length=return_length
         )
         self.assistant_info = search_school_data_agent_info
 
-    def invoke(self, message: str) -> Iterator[SearchSchoolDataAgentResponse]:
-        # Agentクラスのストリーミングを改修後にストリーミング実装
-        self.cancel_streaming()
+    def _add_search_result(self, message: str) -> list[str]:
         search = search_school_database_cosmos(message)
         ids = []
         for doc in search:
             self.assistant_info += f"### {doc.metadata['title']}\n" + doc.page_content + "\n"
             ids.append(doc.metadata["id"])
         super().set_assistant_info(self.assistant_info)
+        return ids
+
+    def invoke(self, message: str) -> AgentResponse:
+        # Agentクラスのストリーミングを改修後にストリーミング実装
+        ids = self._add_search_result(message)
+        resp = super().invoke(message)
+        resp.document_id = ids
+        return resp
 
-        resp = cast(AgentResponse, next(super().invoke(message)))
-        yield {
-            **resp,
-            "document_id": ids
-        }
+    async def stream(self, message: str, return_length: int = 5) -> AsyncIterator[StreamingAgentResponse]:
+        ids = self._add_search_result(message)
+        async for resp in super().stream(message, return_length):
+            yield resp
+        self.result.document_id = ids
 
 if __name__ == "__main__":
     from sc_system_ai.logging_config import setup_logging
@@ -68,5 +66,5 @@ def invoke(self, message: str) -> Iterator[SearchSchoolDataAgentResponse]:
     ]
     user_info = User(name=user_name, major=user_major)
     user_info.conversations.add_conversations_list(history)
-    agent = SearchSchoolDataAgent(user_info=user_info, is_streaming=False)
-    print(next(agent.invoke("京都テックについて教えて")))
+    agent = SearchSchoolDataAgent(user_info=user_info)
+    print(agent.invoke("京都テックについて教えて"))
diff --git a/src/sc_system_ai/agents/small_talk_agent.py b/src/sc_system_ai/agents/small_talk_agent.py
new file mode 100644
index 0000000..38b5897
--- /dev/null
+++ b/src/sc_system_ai/agents/small_talk_agent.py
@@ -0,0 +1,49 @@
+from langchain_openai import AzureChatOpenAI
+
+from sc_system_ai.agents.tools import magic_function
+from sc_system_ai.template.agent import Agent
+from sc_system_ai.template.ai_settings import llm
+from sc_system_ai.template.user_prompts import User
+
+main_agent_tools = [magic_function]
+main_agent_info = """あなたの役割はユーザーと雑談を行うことです。
+ユーザーが楽しめるような会話になるようにしてください。
+"""
+
+# agentクラスの作成
+
+
+class SmallTalkAgent(Agent):
+    def __init__(
+            self,
+            llm: AzureChatOpenAI = llm,
+            user_info: User | None = None,
+    ):
+        super().__init__(
+            llm=llm,
+            user_info=user_info if user_info is not None else User(),
+        )
+        self.assistant_info = main_agent_info
+        super().set_assistant_info(self.assistant_info)
+        super().set_tools(main_agent_tools)
+
+
+if __name__ == "__main__":
+    from sc_system_ai.logging_config import setup_logging
+    setup_logging()
+    # ユーザー情報
+    user_name = "hogehoge"
+    user_major = "fugafuga専攻"
+    history = [
+        ("human", "こんにちは!"),
+        ("ai", "本日はどのようなご用件でしょうか？")
+    ]
+    user_info = User(name=user_name, major=user_major)
+    user_info.conversations.add_conversations_list(history)
+
+    agent = SmallTalkAgent(user_info=user_info)
+    agent.display_agent_info()
+    # print(main_agent.get_agent_prompt())
+    agent.display_agent_prompt()
+    print(agent.invoke("magic function に３をいれて"))
+
diff --git a/src/sc_system_ai/agents/tools/calling_search_school_data_agent.py b/src/sc_system_ai/agents/tools/calling_search_school_data_agent.py
index 2ce47f1..c4aaaf1 100644
--- a/src/sc_system_ai/agents/tools/calling_search_school_data_agent.py
+++ b/src/sc_system_ai/agents/tools/calling_search_school_data_agent.py
@@ -10,6 +10,9 @@
 
 
 class CallingSearchSchoolDataAgent(CallingAgent):
+    # tool側でidを保持する
+    document_id: list[str] | None = None
+
     def __init__(self) -> None:
         super().__init__()
         self.set_tool_info(
@@ -18,6 +21,11 @@ def __init__(self) -> None:
             agent=SearchSchoolDataAgent
         )
 
+    def _run(self, user_input: str) -> str:
+        resp = super()._run(user_input)
+        self.document_id = self.response.document_id if self.response is not None else None
+        return resp
+
 calling_search_school_data_agent = CallingSearchSchoolDataAgent()
 
 if __name__ == "__main__":
diff --git a/src/sc_system_ai/agents/tools/calling_small_talk_agent.py b/src/sc_system_ai/agents/tools/calling_small_talk_agent.py
new file mode 100644
index 0000000..888eef4
--- /dev/null
+++ b/src/sc_system_ai/agents/tools/calling_small_talk_agent.py
@@ -0,0 +1,26 @@
+import logging
+
+from sc_system_ai.agents.small_talk_agent import SmallTalkAgent
+from sc_system_ai.template.calling_agent import CallingAgent
+from sc_system_ai.template.user_prompts import User
+
+logger = logging.getLogger(__name__)
+
+
+class CallingSmallTalkAgent(CallingAgent):
+    def __init__(self) -> None:
+        super().__init__()
+        self.set_tool_info(
+            name="calling_small_talk_agent",
+            description="雑談エージェントを呼び出すツール",
+            agent=SmallTalkAgent,
+        )
+
+calling_small_talk_agent = CallingSmallTalkAgent()
+
+if __name__ == "__main__":
+    from sc_system_ai.logging_config import setup_logging
+    setup_logging()
+
+    calling_small_talk_agent.set_user_info(User(name="hogehoge", major="fugafuga専攻"))
+    print(calling_small_talk_agent.invoke({"user_input": "こんにちは"}))
diff --git a/src/sc_system_ai/main.py b/src/sc_system_ai/main.py
index 6af0a7c..aa0deb8 100644
--- a/src/sc_system_ai/main.py
+++ b/src/sc_system_ai/main.py
@@ -58,9 +58,9 @@
 """
 
 import logging
-from collections.abc import Iterator
+from collections.abc import AsyncIterator
 from importlib import import_module
-from typing import Literal, TypedDict, cast
+from typing import Literal, TypedDict
 
 from sc_system_ai.template.agent import Agent
 from sc_system_ai.template.ai_settings import llm
@@ -68,13 +68,18 @@
 
 logger = logging.getLogger(__name__)
 
-AGENT = Literal["classify", "dummy", "search_school_data"]
+AGENT = Literal["classify", "dummy", "search_school_data", "small_talk"]
 
 class Response(TypedDict):
     output: str | None
     error: str | None
     document_id: list[str] | None
 
+class StreamResponse(TypedDict):
+    output: str | None
+    error: str | None
+    status: str | None
+
 class Chat:
     """Chatクラス
     ユーザー情報と会話履歴を保持し、エージェントとのチャットを行うクラス
@@ -114,16 +119,12 @@ def __init__(
         user_name: str,
         user_major: str,
         conversation: list[tuple[str, str]] | None = None,
-        is_streaming: bool = True,
-        return_length: int = 5
     ) -> None:
         self.user = User(name=user_name, major=user_major)
         if conversation is None:
             conversation = []
 
         self.user.conversations.add_conversations_list(conversation)
-        self.is_streaming = is_streaming
-        self.return_length = return_length
         self._agent: Agent | None = None
 
     @property
@@ -144,7 +145,7 @@ def invoke(
         self,
         message: str,
         command: AGENT = "classify"
-    ) -> Iterator[Response]:
+    ) -> Response:
         """エージェントを呼び出し、チャットを行う関数
 
         Args:
@@ -164,13 +165,46 @@ def invoke(
         - dummy: ダミーエージェント
         """
         self._call_agent(command)
-        if self.is_streaming:
-            for resp in self.agent.invoke(message):
-                if type(resp) is str:
-                    yield self._create_response({"output": resp})
-        else:
-            resp = next(self.agent.invoke(message))
-            yield self._create_response(cast(dict, resp))
+        resp = self.agent.invoke(message)
+        return {
+            "output": resp.output,
+            "error": resp.error,
+            "document_id": resp.document_id
+        }
+
+    async def stream(
+        self,
+        message: str,
+        return_length: int = 5,
+        command: AGENT = "classify"
+    ) -> AsyncIterator[StreamResponse]:
+        """エージェントを呼び出し、ストリーミングチャットを行う関数
+
+        Args:
+            message (str): メッセージ
+            return_length (int, optional): ストリーミングモード時の返答数. デフォルトは5
+            command (AGENT, optional): 呼び出すエージェント。デフォルトでは分類エージェントを呼び出します。
+
+        Returns:
+            Iterator[str]: エージェントからの返答
+
+        コマンドでエージェントを指定して、エージェントを呼び出す場合。
+        ```python
+        for resp in chat.stream(message="私の名前と専攻は何ですか？", command="dummy"):
+            print(resp)
+        ```
+
+        呼び出し可能なエージェント:
+        - classify: 分類エージェント
+        - dummy: ダミーエージェント
+        """
+        self._call_agent(command)
+        async for resp in self.agent.stream(message, return_length):
+            yield {
+                "output": resp.output,
+                "error": resp.error,
+                "status": resp.status
+            }
 
     def _call_agent(self, command: AGENT) -> None:
         try:
@@ -182,21 +216,11 @@ def _call_agent(self, command: AGENT) -> None:
             self.agent = agent_class(
                 llm=llm,
                 user_info=self.user,
-                is_streaming=self.is_streaming,
-                return_length=self.return_length
             )
         except (ModuleNotFoundError, AttributeError, ValueError):
             logger.error(f"エージェントが見つかりません: {command}")
             raise ValueError(f"エージェントが見つかりません: {command}") from None
 
-    def _create_response(self, resp: dict) -> Response:
-        return {
-            "output": resp.get("output"),
-            "error": resp.get("error"),
-            "document_id": resp.get("document_id")
-        }
-
-
 
 def static_chat() -> None:
     # ユーザー情報
@@ -212,16 +236,16 @@ def static_chat() -> None:
     user.conversations.add_conversations_list(conversation)
 
     # エージェントの設定
-    agent = Agent(llm=llm, user_info=user, is_streaming=False)
+    agent = Agent(llm=llm, user_info=user)
 
     # メッセージを送信
     message = "私の名前と専攻は何ですか？"
-    resp = next(agent.invoke(message))
+    resp = agent.invoke(message)
     print(resp)
 
 
 
-def streaming_chat() -> None:
+async def streaming_chat() -> None:
     # ユーザー情報
     user_name = "hogehoge"
     user_major = "fugafuga専攻"
@@ -235,15 +259,17 @@ def streaming_chat() -> None:
     user.conversations.add_conversations_list(conversation)
 
     # エージェントの設定
-    agent = Agent(llm=llm, user_info=user, is_streaming=True)
+    agent = Agent(llm=llm, user_info=user)
 
     # メッセージを送信
     message = "私の名前と専攻は何ですか？"
-    for resp in agent.invoke(message):
+    async for resp in agent.stream(message):
         print(resp)
 
 
 if __name__ == "__main__":
+    import asyncio
+
     from sc_system_ai.logging_config import setup_logging
     setup_logging()
 
@@ -257,7 +283,6 @@ def streaming_chat() -> None:
             ("human", "こんにちは!"),
             ("ai", "本日はどのようなご用件でしょうか？")
         ],
-        is_streaming=False,
     )
     message = "私の名前と専攻は何ですか？"
 
@@ -267,11 +292,12 @@ def streaming_chat() -> None:
     #     pass
 
     # # 通常呼び出し
-    resp = next(chat.invoke(message=message, command="dummy"))
-    print(resp)
+    # resp = chat.invoke(message=message, command="dummy")
+    # print(resp)
 
     # ストリーミング呼び出し
-    # chat.is_streaming = True
-    # for r in chat.invoke(message=message, command="dummy"):
-    #     print(r)
-    # chat.agent.get_response()
+    async def stream() -> None:
+        async for r in chat.stream(message="京都テックについて教えて"):
+            print(r)
+    asyncio.run(stream())
+
diff --git a/src/sc_system_ai/template/agent.py b/src/sc_system_ai/template/agent.py
index 30841f3..7a79ff3 100644
--- a/src/sc_system_ai/template/agent.py
+++ b/src/sc_system_ai/template/agent.py
@@ -7,16 +7,17 @@
 
 """
 import logging
-from collections.abc import Iterator
+from collections.abc import AsyncIterator
 from queue import Queue
 from threading import Thread
-from typing import Any, TypedDict, TypeGuard
+from typing import Any, Literal
 
 from langchain.agents import AgentExecutor, create_tool_calling_agent
 from langchain.tools import BaseTool
 from langchain_community.tools import DuckDuckGoSearchRun
 from langchain_core.messages import AIMessage, HumanMessage
 from langchain_openai import AzureChatOpenAI
+from pydantic import BaseModel
 
 from sc_system_ai.agents.tools import magic_function, search_duckduckgo
 from sc_system_ai.template.ai_settings import llm
@@ -39,39 +40,20 @@ class ToolManager:
     """
     def __init__(
             self,
+            queue: Queue,
             tools: list | None = None,
-            is_streaming: bool = True,
-            queue: Queue | None = None,
     ):
         self.tools: list[BaseTool] = []
-        self._is_streaming = is_streaming
-        self.queue = queue if queue is not None else Queue()
-
+        self.queue = queue
+        self.handler = StreamingToolHandler(self.queue)
         if tools is not None:
             self.set_tools(tools)
 
-    @property
-    def is_streaming(self) -> bool:
-        return self._is_streaming
-
-    @is_streaming.setter
-    def is_streaming(self, is_streaming: bool) -> None:
-        self._is_streaming = is_streaming
-
-        if self._is_streaming:
-            self.tools = self.setup_streaming(self.tools)
-        else:
-            self.cancel_streaming()
-
-    def setup_streaming(self, tools: list[BaseTool]) -> list[BaseTool]:
+    def setup_streaming(self) -> None:
         """ストリーミングのセットアップを行う関数"""
-        self.handler = StreamingToolHandler(self.queue)
-
-        for tool in tools:
+        for tool in self.tools:
             tool.callbacks = [self.handler]
 
-        return tools
-
     def cancel_streaming(self) -> None:
         """ストリーミングのセットアップを解除する関数"""
         for tool in self.tools:
@@ -110,14 +92,22 @@ def _tool_checker(self, tool: Any) -> bool:
     tools: {tools}
 -------------------
 """
-
 # Agentのレスポンスの型
-class AgentResponse(TypedDict, total=False):
+class BaseAgentResponse(BaseModel):
+    """Agentのレスポンスの型"""
+    output: str | None = None
+    error: str | None = None
+
+class AgentResponse(BaseAgentResponse):
     """Agentのレスポンスの型"""
-    chat_history: list[HumanMessage | AIMessage]
-    messages: str
-    output: str
-    error: str
+    chat_history: list[HumanMessage | AIMessage] | None = None
+    messages: str | None = None
+    document_id: list[str] | None = None
+
+
+class StreamingAgentResponse(BaseAgentResponse):
+    """Agentのストリーミングレスポンスの型"""
+    status: Literal["processing", "completed", "error"] | None = None
 
 # Agentクラスの作成
 class Agent:
@@ -134,65 +124,34 @@ def __init__(
             self,
             llm: AzureChatOpenAI = llm,
             user_info: User | None = None,
-            is_streaming: bool = True,
-            return_length: int = 5
     ):
         self.llm = llm
         self.user_info = user_info if user_info is not None else User()
 
         self.result: AgentResponse
-        self._is_streaming = is_streaming
-        self._return_length = return_length
         self.queue: Queue = Queue()
+        self.handler = StreamingAgentHandler(self.queue)
 
         # assistant_infoとtoolsは各エージェントで設定する
         self.assistant_info = ""
-        self.tool = ToolManager(tools=template_tools, is_streaming=self._is_streaming, queue=self.queue)
+        self.tool = ToolManager(tools=template_tools, queue=self.queue)
 
         self.prompt_template = PromptTemplate(assistant_info=self.assistant_info, user_info=self.user_info)
 
-        if self._is_streaming:
-            self.setup_streaming()
-
         self.get_agent_info()
 
-    @property
-    def is_streaming(self) -> bool:
-        return self._is_streaming
-
-    @is_streaming.setter
-    def is_streaming(self, is_streaming: bool) -> None:
-        self._is_streaming = is_streaming
-        self.tool.is_streaming = is_streaming
-
-        if self._is_streaming:
-            self.setup_streaming()
-        else:
-            self.cancel_streaming()
-
-    @property
-    def return_length(self) -> int:
-        return self._return_length
-
-    @return_length.setter
-    def return_length(self, return_length: int) -> None:
-        if return_length <= 0:
-            raise ValueError("return_lengthは1以上の整数である必要があります。")
-        self._return_length = return_length
-
     def setup_streaming(self) -> None:
         """ストリーミング時のセットアップを行う関数"""
         self.clear_queue()
-        self.handler = StreamingAgentHandler(self.queue)
-
-        # llmの設定
         self.llm.streaming = True
         self.llm.callbacks = [self.handler]
+        self.tool.setup_streaming()
 
     def cancel_streaming(self) -> None:
         """ストリーミング時のセットアップを解除する関数"""
         self.llm.streaming = False
         self.llm.callbacks = None
+        self.tool.cancel_streaming()
 
     def clear_queue(self) -> None:
         """キューをクリアする関数"""
@@ -208,7 +167,7 @@ def set_tools(self, tools: list) -> None:
         """ツールを設定する関数"""
         self.tool.set_tools(tools)
 
-    def invoke(self, message: str) -> Iterator[str | AgentResponse]:
+    def invoke(self, message: str) -> AgentResponse:
         """
         エージェントを実行する関数
 
@@ -226,82 +185,119 @@ def invoke(self, message: str) -> Iterator[str | AgentResponse]:
         resp = next(agent.invoke("user message"))
         ```
         """
+        self.cancel_streaming()
+        self._invoke(message, False)
+        return self.get_response()
+
+    async def stream(
+        self,
+        message: str,
+        return_length: int = 5
+    ) -> AsyncIterator[StreamingAgentResponse]:
+        """
+        エージェントをストリーミングで実行する関数
+
+        Args:
+            message (str): ユーザーからのメッセージ
+
+        ```python
+        for output in agent.stream("user message"):
+            print(output)
+        ```
+        """
+        self.setup_streaming()
+        phrase = ""
+        thread = Thread(target=self._invoke, args=(message, True,))
+        thread.start()
+        try:
+            while True:
+                if self.queue.empty():
+                    continue
+
+                token = self.handler.queue.get()
+                if token is None:
+                    logger.debug("エージェントの実行が終了しました。")
+                    break
+                phrase += token
+                if len(phrase) >= return_length:
+                    yield StreamingAgentResponse(
+                        output=phrase, error=None, status="processing"
+                    )
+                    phrase = ""
+        except Exception as e:
+            logger.error(f"エラーが発生しました:{e}")
+            yield StreamingAgentResponse(
+                output=None, error=f"エラーが発生しました:{e}", status="error"
+            )
+
+        if thread and thread.is_alive():
+            thread.join()
+        yield StreamingAgentResponse(output=phrase, error=None, status="completed")
+
+    def _invoke(self, message: str, streaming: bool) -> None:
         agent = create_tool_calling_agent(
             llm=self.llm,
             tools=self.tool.tools,
             prompt=self.prompt_template.full_prompt
         )
-        self.agent_executor = AgentExecutor(
+        agent_executor = AgentExecutor(
             agent=agent,
             tools=self.tool.tools,
-            callbacks= [self.handler] if self._is_streaming else None
+            callbacks= [self.handler] if streaming else None
         )
-
-        if self._is_streaming:
-            yield from self._streaming_invoke(message)
-        else:
-            self._invoke(message)
-            if "error" in self.result:
-                yield self.result["error"]
-            else:
-                yield self.result
-
-    def _invoke(self, message: str) -> None:
         try: # エージェントの実行
             logger.info("エージェントの実行を開始します。\n-------------------\n")
             logger.debug(f"最終的なプロンプト: {self.prompt_template.full_prompt.messages}")
-            resp = self.agent_executor.invoke({
+            resp = agent_executor.invoke({
                 "chat_history": self.user_info.conversations.format_conversation(),
                 "messages": message,
             })
 
-            if self._response_checker(resp):
-                self.result = resp
+            if "output" in resp:
+                self.result = AgentResponse(
+                    chat_history=resp.get("chat_history"),
+                    messages=resp.get("messages"),
+                    output=resp.get("output"),
+                )
             else:
                 logger.error("エージェントの実行結果取得に失敗しました。")
                 logger.debug(f"エージェントの実行結果: {resp}")
                 raise RuntimeError("エージェントの実行結果取得に失敗しました。")
         except Exception as e:
             logger.error(f"エージェントの実行に失敗しました。エラー内容: {e}")
-            self.result = {"error": f"エージェントの実行に失敗しました。エラー内容: {e}"}
-
-    def _response_checker(self, response: Any) -> TypeGuard[AgentResponse]:
-        """レスポンスの型チェック"""
-        if type(response) is dict:
-            if all(key in response for key in ["chat_history", "messages", "output"]):
-                return True
-        return False
+            self.result = AgentResponse(error=f"エージェントの実行に失敗しました。エラー内容: {e}")
 
-    def _streaming_invoke(self, message: str) -> Iterator[str]:
-        phrase = ""
-        thread = Thread(target=self._invoke, args=(message,))
-
-        thread.start()
-        try:
-            while True:
-                token = self.handler.queue.get()
-                if token is None:
-                    break
-
-                phrase += token
-                if len(phrase) >= self._return_length:
-                    yield phrase
-                    phrase = ""
-        except Exception as e:
-            logger.error(f"エラーが発生しました:{e}")
-        finally:
-            yield phrase
+    async def stream_on_tool(self, message: str) -> None:
+        """ツール上でストリーミングでエージェントを実行する関数"""
+        self.handler.queue = self.queue
+        self.setup_streaming()
+        agent = create_tool_calling_agent(
+            llm=self.llm,
+            tools=self.tool.tools,
+            prompt=self.prompt_template.full_prompt
+        )
+        agent_executor = AgentExecutor(
+            agent=agent,
+            tools=self.tool.tools,
+            callbacks= [self.handler],
+        )
+        resp = await agent_executor.ainvoke({
+            "chat_history": self.user_info.conversations.format_conversation(),
+            "messages": message,
+        })
+        self.result = AgentResponse(
+            chat_history=resp.get("chat_history"),
+            messages=resp.get("messages"),
+            output=resp.get("output"),
+        )
 
-            #クリーンアップ
-            if thread and thread.is_alive():
-                thread.join()
 
     def get_response(self) -> AgentResponse:
         """エージェントのレスポンスを取得する関数"""
         try:
             resp = self.result
         except AttributeError:
-            return {"error": "エージェントの実行結果がありません。"}
+            return AgentResponse(error="エージェントの実行結果がありません。")
         else:
             return resp
 
@@ -330,6 +326,8 @@ def display_agent_prompt(self) -> None:
 
 
 if __name__ == "__main__":
+    import asyncio
+
     from sc_system_ai.logging_config import setup_logging
     setup_logging()
     # ユーザー情報
@@ -346,15 +344,15 @@ def display_agent_prompt(self) -> None:
     agent = Agent(
         user_info=user_info,
         llm=llm,
-        is_streaming=False
     )
     agent.assistant_info = "あなたは優秀な校正者です。"
     agent.tool.set_tools(tools)
 
-    result = next(agent.invoke("magic function に３"))
-    print(result)
+    # result = agent.invoke("magic function に３")
+    # print(result)
+
+    async def main() -> None:
+        async for output in agent.stream("magic function に３", 5):
+            print(output.output)
 
-    agent.is_streaming = True
-    for output in agent.invoke("magic function に３"):
-        print(output)
-    print(agent.get_response())
+    asyncio.run(main())
diff --git a/src/sc_system_ai/template/calling_agent.py b/src/sc_system_ai/template/calling_agent.py
index fadce2d..cc5603d 100644
--- a/src/sc_system_ai/template/calling_agent.py
+++ b/src/sc_system_ai/template/calling_agent.py
@@ -1,10 +1,12 @@
+import asyncio
 import logging
-from typing import Any, cast
+from queue import Queue
+from typing import cast
 
 from langchain_core.tools import BaseTool
 from pydantic import BaseModel, ConfigDict, Field
 
-from sc_system_ai.template.agent import Agent
+from sc_system_ai.template.agent import Agent, AgentResponse
 from sc_system_ai.template.user_prompts import User
 
 logger = logging.getLogger(__name__)
@@ -50,6 +52,12 @@ def __init__(self):
 
     user_info: User = Field(description="ユーザー情報", default=User())
     agent: type[Agent] = Agent
+    # AgentResponseを保持する変数
+    response: AgentResponse | None = None
+
+    # ストリーミングのセットアップ
+    queue: Queue = Queue()
+    is_streaming: bool = False
 
     def __init__(self) -> None:
         super().__init__()
@@ -57,21 +65,29 @@ def __init__(self) -> None:
     def _run(
             self,
             user_input: str,
-        ) -> dict[str, Any]:
+        ) -> str:
         logger.info(f"Calling Agent Toolが次の値で呼び出されました: {user_input}")
 
         # エージェントの呼び出し
         try:
-            agent = self.agent(user_info=self.user_info, is_streaming=False)
+            agent = self.agent(user_info=self.user_info)
+            agent.queue = self.queue
         except Exception as e:
             logger.error(f"エージェントの呼び出しに失敗しました: {e}")
             raise e
         else:
             logger.debug(f"エージェントの呼び出しに成功しました: {self.agent}")
 
-        resp = next(agent.invoke(user_input))
+        if self.is_streaming:
+            asyncio.run(agent.stream_on_tool(user_input))
+            resp = agent.get_response()
+        else:
+            resp = agent.invoke(user_input)
+            self.response = resp
+            if resp.error is not None:
+                return resp.error
+        return cast(str, resp.output)
 
-        return cast(dict[str, Any], resp)
 
     def set_user_info(self, user_info: User) -> None:
         """ユーザー情報の設定
@@ -102,6 +118,16 @@ def set_tool_info(
         self.description = description
         self.agent = agent
 
+    def setup_streaming(self, queue: Queue) -> None:
+        """ストリーミングのセットアップ"""
+        self.is_streaming = True
+        self.queue = queue
+
+    def cancel_streaming(self) -> None:
+        """ストリーミングのキャンセル"""
+        self.is_streaming = False
+        self.queue = Queue()
+
 
 calling_agent = CallingAgent()
 
diff --git a/studies/streaming_resp.py b/studies/streaming_resp.py
index 14556db..1fe94e3 100644
--- a/studies/streaming_resp.py
+++ b/studies/streaming_resp.py
@@ -1,12 +1,14 @@
-from dotenv import load_dotenv
+import asyncio
 from queue import Queue
 from threading import Thread
+from typing import Any, AsyncIterator
 
+from dotenv import load_dotenv
+from langchain.agents import AgentExecutor, create_tool_calling_agent
 from langchain.callbacks.base import BaseCallbackHandler
-from langchain.agents import create_tool_calling_agent, AgentExecutor
 
-from sc_system_ai.template.ai_settings import llm
 from sc_system_ai.agents.tools.magic_function import magic_function
+from sc_system_ai.template.ai_settings import llm
 from sc_system_ai.template.system_prompt import PromptTemplate
 from sc_system_ai.template.user_prompts import User
 
@@ -16,21 +18,22 @@ class StreamingHandler(BaseCallbackHandler):
     def __init__(self, queue: Queue):
         super().__init__()
         self.queue = queue
-    
-    def on_llm_new_token(self, token, **kwargs):
+
+    def on_llm_new_token(self, token: str, **kwargs: Any) -> None:
+        print(token)
         if token:
             self.queue.put(token)
 
-    def on_agent_finish(self, response, **kwargs):
+    def on_agent_finish(self, finish, **kwargs: Any) -> None:
         print("end")
         self.queue.put(None)
 
-    def on_llm_error(self, error, **kwargs):
+    def on_llm_error(self, **kwargs: Any) -> None:
         self.queue.put(None)
 
 
-def main():
-    queue = Queue()
+async def main() -> AsyncIterator[str]:
+    queue: Queue = Queue()
     handler = StreamingHandler(queue)
 
     llm.streaming=True
@@ -46,9 +49,9 @@ def main():
         prompt=prompt.full_prompt
     )
     agent_executor = AgentExecutor(agent=agent, tools=tool, callbacks=[handler])
-    def task():
+    def task() -> None:
         result = agent_executor.invoke({
-            "messages": "magic functionに3を入れて"
+            "messages": "悲しい歌を歌ってください"
         })
         print(result)
 
@@ -56,21 +59,23 @@ def task():
     thread.start()
 
     resp = ""
+    while True:
+        if queue.empty():
+            continue
+        token = queue.get()
+        if token is None:
+            break
+        resp += token
+        if len(resp) > 1:
+            yield resp
+            resp = ""
+    thread.join()
+    yield resp
+
+async def job() -> None:
+    async for i in main():
+        print(i)
+
+if __name__ == "__main__":
+    asyncio.run(job())
 
-    try:
-        while True:
-            token = queue.get()
-            if token is None:
-                break
-
-            resp += token
-            if len(resp) >= 5:
-                yield resp
-                resp = ""
-    finally:
-        yield resp
-        if thread and thread.is_alive():
-            thread.join()
-
-for output in main():
-    print(output)