diff --git a/libs/langchain_v1/pyproject.toml b/libs/langchain_v1/pyproject.toml
index 736d7f23a12..3c92ddffea8 100644
--- a/libs/langchain_v1/pyproject.toml
+++ b/libs/langchain_v1/pyproject.toml
@@ -111,10 +111,7 @@ enable_error_code = "deprecated"
 warn_unreachable = true
 
 exclude = [
-    # Exclude agents tests except middleware_typing/ which has type-checked tests
     "tests/unit_tests/agents/middleware/",
-    "tests/unit_tests/agents/specifications/",
-    "tests/unit_tests/agents/test_.*\\.py",
 ]
 
 [[tool.mypy.overrides]]
@@ -147,6 +144,7 @@ unfixable = [
 ]
 
 flake8-annotations.allow-star-arg-any = true
+flake8-annotations.mypy-init-return = true
 allowed-confusables = ["–"]
 
 [tool.ruff.lint.flake8-tidy-imports]
@@ -157,13 +155,7 @@ convention = "google"
 ignore-var-parameters = true  # ignore missing documentation for *args and **kwargs parameters
 
 [tool.ruff.lint.extend-per-file-ignores]
-"tests/unit_tests/agents/*" = [
-    "ANN", # Annotations, needs to fix
-]
-"tests/unit_tests/agents/test_responses_spec.py" = ["F821"]
-"tests/unit_tests/agents/test_return_direct_spec.py" = ["F821"]
 "tests/unit_tests/agents/test_react_agent.py" = ["ALL"]
-
 "tests/*" = [
     "D1",      # Documentation rules
     "S101",    # Tests need assertions
diff --git a/libs/langchain_v1/tests/unit_tests/agents/test_agent_name.py b/libs/langchain_v1/tests/unit_tests/agents/test_agent_name.py
index bdb4e5b080a..1ff2914a75b 100644
--- a/libs/langchain_v1/tests/unit_tests/agents/test_agent_name.py
+++ b/libs/langchain_v1/tests/unit_tests/agents/test_agent_name.py
@@ -5,6 +5,8 @@ This module tests that the name parameter correctly sets .name on AIMessage outp
 
 from __future__ import annotations
 
+from typing import Any, cast
+
 from langchain_core.messages import (
     AIMessage,
     HumanMessage,
@@ -119,7 +121,9 @@ def test_lc_agent_name_in_stream_metadata() -> None:
         stream_mode="messages",
     ):
         if "lc_agent_name" in metadata:
-            metadata_with_agent_name.append(metadata["lc_agent_name"])
+            # `stream()` is typed `Iterator[dict[str, Any] | Any]`, so unpacking the
+            # tuple leaves `metadata` as `str | Any`; cast to index it as the dict it is.
+            metadata_with_agent_name.append(cast("dict[str, Any]", metadata)["lc_agent_name"])
 
     assert len(metadata_with_agent_name) > 0
     assert all(name == "streaming_agent" for name in metadata_with_agent_name)
@@ -155,7 +159,9 @@ def test_lc_agent_name_in_stream_metadata_multiple_iterations() -> None:
         stream_mode="messages",
     ):
         if "lc_agent_name" in metadata:
-            metadata_with_agent_name.append(metadata["lc_agent_name"])
+            # `stream()` is typed `Iterator[dict[str, Any] | Any]`, so unpacking the
+            # tuple leaves `metadata` as `str | Any`; cast to index it as the dict it is.
+            metadata_with_agent_name.append(cast("dict[str, Any]", metadata)["lc_agent_name"])
 
     # Should have metadata entries for messages from both iterations
     assert len(metadata_with_agent_name) > 0
@@ -176,7 +182,9 @@ async def test_lc_agent_name_in_astream_metadata() -> None:
         stream_mode="messages",
     ):
         if "lc_agent_name" in metadata:
-            metadata_with_agent_name.append(metadata["lc_agent_name"])
+            # `stream()` is typed `Iterator[dict[str, Any] | Any]`, so unpacking the
+            # tuple leaves `metadata` as `str | Any`; cast to index it as the dict it is.
+            metadata_with_agent_name.append(cast("dict[str, Any]", metadata)["lc_agent_name"])
 
     assert len(metadata_with_agent_name) > 0
     assert all(name == "async_streaming_agent" for name in metadata_with_agent_name)
@@ -212,7 +220,9 @@ async def test_lc_agent_name_in_astream_metadata_multiple_iterations() -> None:
         stream_mode="messages",
     ):
         if "lc_agent_name" in metadata:
-            metadata_with_agent_name.append(metadata["lc_agent_name"])
+            # `stream()` is typed `Iterator[dict[str, Any] | Any]`, so unpacking the
+            # tuple leaves `metadata` as `str | Any`; cast to index it as the dict it is.
+            metadata_with_agent_name.append(cast("dict[str, Any]", metadata)["lc_agent_name"])
 
     # Should have metadata entries for messages from both iterations
     assert len(metadata_with_agent_name) > 0
diff --git a/libs/langchain_v1/tests/unit_tests/agents/test_agent_streaming.py b/libs/langchain_v1/tests/unit_tests/agents/test_agent_streaming.py
index 8b5275137ec..4aa58a3f1ae 100644
--- a/libs/langchain_v1/tests/unit_tests/agents/test_agent_streaming.py
+++ b/libs/langchain_v1/tests/unit_tests/agents/test_agent_streaming.py
@@ -113,7 +113,7 @@ class TestAgentStreamV3Sync:
         # by `BaseRunStream.__init__` whenever `MessagesTransformer` is
         # registered. Content population is covered by langgraph tests —
         # here we only assert the agent streamer inherits the built-in.
-        assert "messages" in run._mux.extensions  # type: ignore[attr-defined]
+        assert "messages" in run._mux.extensions
         assert hasattr(run, "messages")
         # Drain so the run closes cleanly.
         for tc in run.tool_calls:
@@ -145,14 +145,14 @@ class TestAgentStreamV3Sync:
             transformers=[_Marker],
         )
         # Both the agent default and the user transformer are registered.
-        assert "tool_calls" in run._mux.extensions  # type: ignore[attr-defined]
-        assert "marker" in run._mux.extensions  # type: ignore[attr-defined]
+        assert "tool_calls" in run._mux.extensions
+        assert "marker" in run._mux.extensions
 
         # `ToolCallTransformer` must come BEFORE the user's transformer in
         # the registration order so it processes `tools` events first. The
         # docstring on `create_agent` promises caller transformers are
         # appended after the built-in.
-        transformers = run._mux._transformers  # type: ignore[attr-defined]
+        transformers = run._mux._transformers
         tool_call_idx = next(
             i for i, t in enumerate(transformers) if isinstance(t, ToolCallTransformer)
         )
diff --git a/libs/langchain_v1/tests/unit_tests/agents/test_create_agent_tool_validation.py b/libs/langchain_v1/tests/unit_tests/agents/test_create_agent_tool_validation.py
index 10047e4dd30..eab76855871 100644
--- a/libs/langchain_v1/tests/unit_tests/agents/test_create_agent_tool_validation.py
+++ b/libs/langchain_v1/tests/unit_tests/agents/test_create_agent_tool_validation.py
@@ -9,6 +9,7 @@ from langgraph.store.base import BaseStore
 from langgraph.store.memory import InMemoryStore
 
 from langchain.agents import AgentState, create_agent
+from langchain.agents.middleware import InputAgentState
 from langchain.tools import InjectedState
 from langchain.tools import tool as dec_tool
 from tests.unit_tests.agents.model import FakeToolCallingModel
@@ -37,6 +38,9 @@ def test_tool_invocation_error_excludes_injected_state() -> None:
     class TestState(AgentState[Any]):
         secret_data: str
 
+    class TestInputState(InputAgentState):
+        secret_data: str
+
     @dec_tool
     def tool_with_injected_state(
         some_val: int,
@@ -66,10 +70,10 @@ def test_tool_invocation_error_excludes_injected_state() -> None:
     )
 
     result = agent.invoke(
-        {
-            "messages": [HumanMessage("Test message")],
-            "secret_data": "sensitive_secret_123",
-        }
+        TestInputState(
+            messages=[HumanMessage("Test message")],
+            secret_data="sensitive_secret_123",  # noqa: S106
+        )
     )
 
     tool_messages = [m for m in result["messages"] if m.type == "tool"]
@@ -96,6 +100,9 @@ async def test_tool_invocation_error_excludes_injected_state_async() -> None:
     class TestState(AgentState[Any]):
         internal_data: str
 
+    class TestInputState(InputAgentState):
+        internal_data: str
+
     @dec_tool
     async def async_tool_with_injected_state(
         query: str,
@@ -126,10 +133,9 @@ async def test_tool_invocation_error_excludes_injected_state_async() -> None:
     )
 
     result = await agent.ainvoke(
-        {
-            "messages": [HumanMessage("Test async")],
-            "internal_data": "secret_internal_value_xyz",
-        }
+        TestInputState(
+            messages=[HumanMessage("Test async")], internal_data="secret_internal_value_xyz"
+        )
     )
 
     tool_messages = [m for m in result["messages"] if m.type == "tool"]
@@ -342,6 +348,9 @@ def test_create_agent_error_only_model_controllable_params() -> None:
     class StateWithSecrets(AgentState[Any]):
         password: str
 
+    class InputStateWithSecrets(InputAgentState):
+        password: str
+
     @dec_tool
     def secure_tool(
         username: str,
@@ -381,10 +390,10 @@ def test_create_agent_error_only_model_controllable_params() -> None:
     )
 
     result = agent.invoke(
-        {
-            "messages": [HumanMessage("Create account")],
-            "password": "super_secret_password_12345",
-        }
+        InputStateWithSecrets(
+            messages=[HumanMessage("Create account")],
+            password="super_secret_password_12345",  # noqa: S106
+        )
     )
 
     tool_messages = [m for m in result["messages"] if m.type == "tool"]
diff --git a/libs/langchain_v1/tests/unit_tests/agents/test_fetch_last_ai_and_tool_messages.py b/libs/langchain_v1/tests/unit_tests/agents/test_fetch_last_ai_and_tool_messages.py
index bec7adaa93b..a1c9204db32 100644
--- a/libs/langchain_v1/tests/unit_tests/agents/test_fetch_last_ai_and_tool_messages.py
+++ b/libs/langchain_v1/tests/unit_tests/agents/test_fetch_last_ai_and_tool_messages.py
@@ -5,14 +5,14 @@ including the scenario where no AIMessage exists in the message list
 (fixes issue #34792).
 """
 
-from langchain_core.messages import AIMessage, HumanMessage, SystemMessage, ToolMessage
+from langchain_core.messages import AIMessage, AnyMessage, HumanMessage, SystemMessage, ToolMessage
 
 from langchain.agents.factory import _fetch_last_ai_and_tool_messages
 
 
 def test_fetch_last_ai_and_tool_messages_normal() -> None:
     """Test normal case with AIMessage and subsequent ToolMessages."""
-    messages = [
+    messages: list[AnyMessage] = [
         HumanMessage(content="Hello"),
         AIMessage(content="Hi there!", tool_calls=[{"name": "test", "id": "1", "args": {}}]),
         ToolMessage(content="Tool result", tool_call_id="1"),
@@ -29,7 +29,7 @@ def test_fetch_last_ai_and_tool_messages_normal() -> None:
 
 def test_fetch_last_ai_and_tool_messages_multiple_ai() -> None:
     """Test that the last AIMessage is returned when multiple exist."""
-    messages = [
+    messages: list[AnyMessage] = [
         HumanMessage(content="First question"),
         AIMessage(content="First answer", id="ai1"),
         HumanMessage(content="Second question"),
@@ -53,7 +53,7 @@ def test_fetch_last_ai_and_tool_messages_no_ai_message() -> None:
     The function now returns None for the AIMessage, allowing callers to
     handle this edge case explicitly.
     """
-    messages = [
+    messages: list[AnyMessage] = [
         HumanMessage(content="Hello"),
         SystemMessage(content="You are a helpful assistant"),
     ]
@@ -70,7 +70,7 @@ def test_fetch_last_ai_and_tool_messages_empty_list() -> None:
 
     This can occur after RemoveMessage(id=REMOVE_ALL_MESSAGES) clears all messages.
     """
-    messages: list = []
+    messages: list[AnyMessage] = []
 
     ai_msg, tool_msgs = _fetch_last_ai_and_tool_messages(messages)
 
@@ -81,7 +81,7 @@ def test_fetch_last_ai_and_tool_messages_empty_list() -> None:
 
 def test_fetch_last_ai_and_tool_messages_only_human_messages() -> None:
     """Test handling when only HumanMessages exist."""
-    messages = [
+    messages: list[AnyMessage] = [
         HumanMessage(content="Hello"),
         HumanMessage(content="Are you there?"),
     ]
@@ -94,7 +94,7 @@ def test_fetch_last_ai_and_tool_messages_only_human_messages() -> None:
 
 def test_fetch_last_ai_and_tool_messages_ai_without_tool_calls() -> None:
     """Test AIMessage without tool_calls returns empty tool messages list."""
-    messages = [
+    messages: list[AnyMessage] = [
         HumanMessage(content="Hello"),
         AIMessage(content="Hi! How can I help you today?"),
     ]
diff --git a/libs/langchain_v1/tests/unit_tests/agents/test_injected_runtime_create_agent.py b/libs/langchain_v1/tests/unit_tests/agents/test_injected_runtime_create_agent.py
index 0f8b70657c3..f9d13afc322 100644
--- a/libs/langchain_v1/tests/unit_tests/agents/test_injected_runtime_create_agent.py
+++ b/libs/langchain_v1/tests/unit_tests/agents/test_injected_runtime_create_agent.py
@@ -25,7 +25,7 @@ from langgraph.store.memory import InMemoryStore
 from typing_extensions import override
 
 from langchain.agents import create_agent
-from langchain.agents.middleware.types import AgentMiddleware, AgentState
+from langchain.agents.middleware.types import AgentMiddleware, AgentState, InputAgentState
 from langchain.tools import InjectedState, ToolRuntime
 from tests.unit_tests.agents.model import FakeToolCallingModel
 
@@ -303,6 +303,9 @@ def test_tool_runtime_with_custom_state() -> None:
     class CustomState(AgentState[Any]):
         custom_field: str
 
+    class CustomInputState(InputAgentState):
+        custom_field: str
+
     runtime_state = {}
 
     @tool
@@ -327,10 +330,10 @@ def test_tool_runtime_with_custom_state() -> None:
     )
 
     result = agent.invoke(
-        {
-            "messages": [HumanMessage("Test custom state")],
-            "custom_field": "custom_value",
-        }
+        CustomInputState(
+            messages=[HumanMessage("Test custom state")],
+            custom_field="custom_value",
+        )
     )
 
     # Verify custom field was accessible
@@ -622,6 +625,10 @@ def test_combined_injected_state_runtime_store() -> None:
         user_id: str
         session_id: str
 
+    class CustomInputState(InputAgentState):
+        user_id: str
+        session_id: str
+
     # Define explicit args schema that only includes LLM-controlled parameters
     weather_schema = {
         "type": "object",
@@ -692,11 +699,11 @@ def test_combined_injected_state_runtime_store() -> None:
 
     # Invoke with custom state fields
     result = agent.invoke(
-        {
-            "messages": [HumanMessage("What's the weather like?")],
-            "user_id": "user_42",
-            "session_id": "session_abc123",
-        }
+        CustomInputState(
+            messages=[HumanMessage("What's the weather like?")],
+            user_id="user_42",
+            session_id="session_abc123",
+        )
     )
 
     # Verify tool executed successfully
@@ -740,6 +747,10 @@ async def test_combined_injected_state_runtime_store_async() -> None:
         api_key: str
         request_id: str
 
+    class CustomInputState(InputAgentState):
+        api_key: str
+        request_id: str
+
     # Define explicit args schema that only includes LLM-controlled parameters
     # Note: state, runtime, and store are NOT in this schema
     search_schema = {
@@ -819,11 +830,11 @@ async def test_combined_injected_state_runtime_store_async() -> None:
 
     # Invoke async
     result = await agent.ainvoke(
-        {
-            "messages": [HumanMessage("Search for something")],
-            "api_key": "sk-test-key-xyz",
-            "request_id": "req_999",
-        }
+        CustomInputState(
+            messages=[HumanMessage("Search for something")],
+            api_key="sk-test-key-xyz",
+            request_id="req_999",
+        )
     )
 
     # Verify tool executed successfully
diff --git a/libs/langchain_v1/tests/unit_tests/agents/test_kwargs_tool_runtime_injection.py b/libs/langchain_v1/tests/unit_tests/agents/test_kwargs_tool_runtime_injection.py
index 1e6637fb6c3..5c817847508 100644
--- a/libs/langchain_v1/tests/unit_tests/agents/test_kwargs_tool_runtime_injection.py
+++ b/libs/langchain_v1/tests/unit_tests/agents/test_kwargs_tool_runtime_injection.py
@@ -16,8 +16,8 @@ class ArgsSchema(BaseModel):
     """Args schema with config and runtime fields."""
 
     query: str = Field(description="The query")
-    config: dict | None = Field(default=None)
-    runtime: dict | None = Field(default=None)
+    config: dict[str, Any] | None = Field(default=None)
+    runtime: dict[str, Any] | None = Field(default=None)
 
 
 def test_config_and_runtime_not_injected_to_kwargs() -> None:
diff --git a/libs/langchain_v1/tests/unit_tests/agents/test_response_format.py b/libs/langchain_v1/tests/unit_tests/agents/test_response_format.py
index 55ff31a0b40..f2c237f0cf6 100644
--- a/libs/langchain_v1/tests/unit_tests/agents/test_response_format.py
+++ b/libs/langchain_v1/tests/unit_tests/agents/test_response_format.py
@@ -905,7 +905,7 @@ class TestSupportsProviderStrategy:
     """Unit tests for `_supports_provider_strategy`."""
 
     @staticmethod
-    def _make_structured_model(model_name: str):
+    def _make_structured_model(model_name: str) -> GenericFakeChatModel:
         class GeminiTestChatModel(GenericFakeChatModel):
             model_name: str
 
diff --git a/libs/langchain_v1/tests/unit_tests/agents/test_responses_spec.py b/libs/langchain_v1/tests/unit_tests/agents/test_responses_spec.py
index 5c86d90f7c5..6c93b753b6b 100644
--- a/libs/langchain_v1/tests/unit_tests/agents/test_responses_spec.py
+++ b/libs/langchain_v1/tests/unit_tests/agents/test_responses_spec.py
@@ -118,7 +118,7 @@ def test_responses_integration_matrix(case: TestCase) -> None:
     # Unwrap nested schema objects
     response_format_spec = [item.get("schema", item) for item in response_format_spec]
     if len(response_format_spec) == 1:
-        tool_output = ToolStrategy(response_format_spec[0])
+        tool_output = ToolStrategy[Any](response_format_spec[0])
     else:
         tool_output = ToolStrategy({"oneOf": response_format_spec})
 
diff --git a/libs/langchain_v1/tests/unit_tests/agents/test_state_schema.py b/libs/langchain_v1/tests/unit_tests/agents/test_state_schema.py
index bd0bffeb35f..65f10da9d2a 100644
--- a/libs/langchain_v1/tests/unit_tests/agents/test_state_schema.py
+++ b/libs/langchain_v1/tests/unit_tests/agents/test_state_schema.py
@@ -16,6 +16,7 @@ from langchain.agents import create_agent, factory
 from langchain.agents.middleware.types import (
     AgentMiddleware,
     AgentState,
+    InputAgentState,
     PrivateStateAttr,
 )
 
@@ -40,6 +41,9 @@ def test_state_schema_single_custom_field() -> None:
     class CustomState(AgentState[Any]):
         custom_field: str
 
+    class CustomInputState(InputAgentState):
+        custom_field: str
+
     agent = create_agent(
         model=FakeToolCallingModel(
             tool_calls=[[{"args": {"x": 1}, "id": "call_1", "name": "simple_tool"}], []]
@@ -48,7 +52,9 @@ def test_state_schema_single_custom_field() -> None:
         state_schema=CustomState,
     )
 
-    result = agent.invoke({"messages": [HumanMessage("Test")], "custom_field": "test_value"})
+    result = agent.invoke(
+        CustomInputState(messages=[HumanMessage("Test")], custom_field="test_value")
+    )
 
     assert result["custom_field"] == "test_value"
     assert len(result["messages"]) == 4
@@ -62,6 +68,11 @@ def test_state_schema_multiple_custom_fields() -> None:
         session_id: str
         context: str
 
+    class CustomInputState(InputAgentState):
+        user_id: str
+        session_id: str
+        context: str
+
     agent = create_agent(
         model=FakeToolCallingModel(
             tool_calls=[[{"args": {"x": 1}, "id": "call_1", "name": "simple_tool"}], []]
@@ -71,12 +82,12 @@ def test_state_schema_multiple_custom_fields() -> None:
     )
 
     result = agent.invoke(
-        {
-            "messages": [HumanMessage("Test")],
-            "user_id": "user_123",
-            "session_id": "session_456",
-            "context": "test_ctx",
-        }
+        CustomInputState(
+            messages=[HumanMessage("Test")],
+            user_id="user_123",
+            session_id="session_456",
+            context="test_ctx",
+        )
     )
 
     assert result["user_id"] == "user_123"
@@ -91,6 +102,9 @@ def test_state_schema_with_tool_runtime() -> None:
     class ExtendedState(AgentState[Any]):
         counter: int
 
+    class ExtendedInputState(InputAgentState):
+        counter: int
+
     runtime_data = {}
 
     @tool
@@ -107,7 +121,7 @@ def test_state_schema_with_tool_runtime() -> None:
         state_schema=ExtendedState,
     )
 
-    result = agent.invoke({"messages": [HumanMessage("Test")], "counter": 5})
+    result = agent.invoke(ExtendedInputState(messages=[HumanMessage("Test")], counter=5))
 
     assert runtime_data["counter"] == 5
     assert "Counter is 5" in result["messages"][2].content
@@ -122,6 +136,10 @@ def test_state_schema_with_middleware() -> None:
     class MiddlewareState(AgentState[Any]):
         middleware_data: str
 
+    class UserAndMiddlewareInputState(InputAgentState):
+        user_name: str
+        middleware_data: str
+
     middleware_calls = []
 
     class TestMiddleware(AgentMiddleware[MiddlewareState, None]):
@@ -142,11 +160,11 @@ def test_state_schema_with_middleware() -> None:
     )
 
     result = agent.invoke(
-        {
-            "messages": [HumanMessage("Test")],
-            "user_name": "Alice",
-            "middleware_data": "test_data",
-        }
+        UserAndMiddlewareInputState(
+            messages=[HumanMessage("Test")],
+            user_name="Alice",
+            middleware_data="test_data",
+        )
     )
 
     assert result["user_name"] == "Alice"
@@ -176,6 +194,9 @@ async def test_state_schema_async() -> None:
     class AsyncState(AgentState[Any]):
         async_field: str
 
+    class AsyncInputState(InputAgentState):
+        async_field: str
+
     @tool
     async def async_tool(x: int) -> str:
         """Async tool."""
@@ -190,10 +211,10 @@ async def test_state_schema_async() -> None:
     )
 
     result = await agent.ainvoke(
-        {
-            "messages": [HumanMessage("Test async")],
-            "async_field": "async_value",
-        }
+        AsyncInputState(
+            messages=[HumanMessage("Test async")],
+            async_field="async_value",
+        )
     )
 
     assert result["async_field"] == "async_value"
@@ -213,6 +234,10 @@ def test_state_schema_with_private_state_field() -> None:
         public_field: str
         private_field: Annotated[str, PrivateStateAttr]
 
+    class InputStateWithPrivateField(InputAgentState):
+        public_field: str
+        private_field: Annotated[str, PrivateStateAttr]
+
     captured_state = {}
 
     @tool
@@ -234,11 +259,11 @@ def test_state_schema_with_private_state_field() -> None:
 
     # Invoke the agent with BOTH public and private fields
     result = agent.invoke(
-        {
-            "messages": [HumanMessage("Test private state")],
-            "public_field": "public_value",
-            "private_field": "private_value",  # This should be filtered out
-        }
+        InputStateWithPrivateField(
+            messages=[HumanMessage("Test private state")],
+            public_field="public_value",
+            private_field="private_value",  # This should be filtered out
+        )
     )
 
     # Assert that public_field is preserved in the result
diff --git a/libs/langchain_v1/tests/unit_tests/agents/test_subagent_streaming.py b/libs/langchain_v1/tests/unit_tests/agents/test_subagent_streaming.py
index 8505dec3d11..220011108e6 100644
--- a/libs/langchain_v1/tests/unit_tests/agents/test_subagent_streaming.py
+++ b/libs/langchain_v1/tests/unit_tests/agents/test_subagent_streaming.py
@@ -8,14 +8,21 @@ tools were dropped during `stream(..., subgraphs=True)`.
 
 from __future__ import annotations
 
-from langchain_core.messages import HumanMessage, ToolCall
-from langchain_core.tools import tool
+from typing import TYPE_CHECKING, Any, cast
 
-from langchain.agents import create_agent
+from langchain_core.messages import HumanMessage, ToolCall
+from langchain_core.tools import BaseTool, tool
+
+from langchain.agents import AgentState, create_agent
 from tests.unit_tests.agents.model import FakeToolCallingModel
 
+if TYPE_CHECKING:
+    from langgraph.graph.state import CompiledStateGraph
 
-def _make_subagent_caller_tool():
+    from langchain.agents.middleware import InputAgentState, OutputAgentState
+
+
+def _make_subagent_caller_tool() -> BaseTool:
     """Build a subagent and a tool that invokes it."""
     subagent = create_agent(
         model=FakeToolCallingModel(tool_calls=[[]]),
@@ -26,12 +33,18 @@ def _make_subagent_caller_tool():
     def call_subagent(query: str) -> str:
         """Delegate the query to a sub-agent."""
         result = subagent.invoke({"messages": [HumanMessage(query)]})
-        return result["messages"][-1].text
+        # `invoke()` returns an untyped state, so `.text` is `Any`; it is really a
+        # `str` (`TextAccessor`), so narrow it to satisfy the `-> str` return type.
+        return cast("str", result["messages"][-1].text)
 
     return call_subagent
 
 
-def _make_parent_agent(call_subagent_tool) -> object:
+# Return type mirrors `create_agent`'s overload; the context slot is unparameterized
+# here, so it resolves to `None` (the `ContextT` default).
+def _make_parent_agent(
+    call_subagent_tool: BaseTool,
+) -> CompiledStateGraph[AgentState[Any], None, InputAgentState, OutputAgentState[Any]]:
     parent_tool_calls: list[list[ToolCall]] = [
         [{"args": {"query": "hi"}, "id": "call_1", "name": "call_subagent"}],
         [],
diff --git a/libs/langchain_v1/tests/unit_tests/agents/test_subagent_transformer.py b/libs/langchain_v1/tests/unit_tests/agents/test_subagent_transformer.py
index d5c3202e350..dc6f1bd99cd 100644
--- a/libs/langchain_v1/tests/unit_tests/agents/test_subagent_transformer.py
+++ b/libs/langchain_v1/tests/unit_tests/agents/test_subagent_transformer.py
@@ -9,6 +9,7 @@ dispatches a nested `create_agent` from a tool, giving true end-to-end coverage.
 from __future__ import annotations
 
 import sys
+from typing import cast
 
 import pytest
 from langchain_core.messages import HumanMessage
@@ -38,7 +39,9 @@ def test_subagents_surfaces_named_subagent() -> None:
     def call_weather(city: str) -> str:
         """Call the weather agent."""
         result = weather_agent.invoke({"messages": [HumanMessage(f"weather in {city}")]})
-        return result["messages"][-1].text
+        # `invoke()` returns an untyped state, so `.text` is `Any`; it is really a
+        # `str` (`TextAccessor`), so narrow it to satisfy the `-> str` return type.
+        return cast("str", result["messages"][-1].text)
 
     supervisor = create_agent(
         model=_supervisor_model(),
@@ -85,7 +88,9 @@ async def test_subagents_surfaces_named_subagent_async() -> None:
     async def call_weather(city: str) -> str:
         """Call the weather agent."""
         result = await weather_agent.ainvoke({"messages": [HumanMessage(f"weather in {city}")]})
-        return result["messages"][-1].text
+        # `invoke()` returns an untyped state, so `.text` is `Any`; it is really a
+        # `str` (`TextAccessor`), so narrow it to satisfy the `-> str` return type.
+        return cast("str", result["messages"][-1].text)
 
     supervisor = create_agent(
         model=_supervisor_model(),
@@ -148,7 +153,9 @@ def test_unnamed_inner_agent_surfaces_with_inherited_name() -> None:
     def call_weather(city: str) -> str:
         """Call an unnamed inner agent."""
         result = inner_agent.invoke({"messages": [HumanMessage(f"weather in {city}")]})
-        return result["messages"][-1].text
+        # `invoke()` returns an untyped state, so `.text` is `Any`; it is really a
+        # `str` (`TextAccessor`), so narrow it to satisfy the `-> str` return type.
+        return cast("str", result["messages"][-1].text)
 
     supervisor = create_agent(
         model=_supervisor_model(),
@@ -183,7 +190,9 @@ def test_same_name_nested_agent_surfaced() -> None:
     def call_weather(city: str) -> str:
         """Call a same-named inner agent."""
         result = inner_agent.invoke({"messages": [HumanMessage(f"weather in {city}")]})
-        return result["messages"][-1].text
+        # `invoke()` returns an untyped state, so `.text` is `Any`; it is really a
+        # `str` (`TextAccessor`), so narrow it to satisfy the `-> str` return type.
+        return cast("str", result["messages"][-1].text)
 
     # The parent agent shares the inner agent's name.
     supervisor = create_agent(
diff --git a/libs/langchain_v1/tests/unit_tests/agents/test_system_message.py b/libs/langchain_v1/tests/unit_tests/agents/test_system_message.py
index 9c141feaf31..d7d4e24cabf 100644
--- a/libs/langchain_v1/tests/unit_tests/agents/test_system_message.py
+++ b/libs/langchain_v1/tests/unit_tests/agents/test_system_message.py
@@ -765,7 +765,7 @@ class TestDynamicSystemPromptMiddleware:
     def test_middleware_can_return_system_message(self) -> None:
         """Test that middleware can return a SystemMessage with dynamic content."""
 
-        def dynamic_system_prompt_middleware(request: ModelRequest) -> SystemMessage:
+        def dynamic_system_prompt_middleware(request: ModelRequest[Any]) -> SystemMessage:
             """Return a SystemMessage with dynamic content."""
             region = getattr(request.runtime.context, "region", "n/a")
             return SystemMessage(content=f"You are a helpful assistant. Region: {region}")
@@ -786,7 +786,6 @@ class TestDynamicSystemPromptMiddleware:
             runtime=runtime,
             model_settings={},
         )
-
         new_system_message = dynamic_system_prompt_middleware(request)
 
         assert isinstance(new_system_message, SystemMessage)