boom

libs/langchain_v1/tests/unit_tests/agents/middleware/core/test_framework.py

@@ -9,7 +9,6 @@ from langchain_core.tools import InjectedToolCallId, tool
 from langgraph.checkpoint.base import BaseCheckpointSaver
 from langgraph.checkpoint.memory import InMemorySaver
 from langgraph.runtime import Runtime
-from langgraph.types import GraphOutput
 from pydantic import BaseModel, Field
 from syrupy.assertion import SnapshotAssertion
 from typing_extensions import override
@@ -96,10 +95,7 @@ def test_create_agent_invoke(
 
     thread1 = {"configurable": {"thread_id": "1"}}
     result = agent_one.invoke({"messages": ["hello"]}, thread1)
-    # v2 stream format returns GraphOutput; unwrap to get the dict
-    assert isinstance(result, GraphOutput)
-    result_dict = result.value
-    assert result_dict == {
+    assert result.value == {
         "messages": [
             _AnyIdHumanMessage(content="hello"),
             AIMessage(
@@ -201,9 +197,7 @@ def test_create_agent_jump(
 
     thread1 = {"configurable": {"thread_id": "1"}}
     result = agent_one.invoke({"messages": []}, thread1)
-    assert isinstance(result, GraphOutput)
-    result_dict = result.value
-    assert result_dict == {"messages": []}
+    assert result.value == {"messages": []}
     assert calls == ["NoopSeven.before_model", "NoopEight.before_model"]
 
 
@@ -701,9 +695,7 @@ async def test_create_agent_async_jump() -> None:
 
     result = await agent.ainvoke({"messages": []})
 
-    assert isinstance(result, GraphOutput)
-    result_dict = result.value
-    assert result_dict == {"messages": []}
+    assert result.value == {"messages": []}
     assert calls == ["AsyncMiddlewareOne.abefore_model", "AsyncMiddlewareTwo.abefore_model"]
 
 

libs/langchain_v1/tests/unit_tests/agents/middleware/core/test_wrap_model_call_state_update.py

@@ -11,7 +11,7 @@ import pytest
 from langchain_core.language_models.fake_chat_models import GenericFakeChatModel
 from langchain_core.messages import AIMessage, HumanMessage
 from langgraph.errors import InvalidUpdateError
-from langgraph.types import Command, GraphOutput
+from langgraph.types import Command
 
 from langchain.agents import AgentState, create_agent
 from langchain.agents.middleware.types import (
@@ -630,10 +630,8 @@ class TestComposition:
 
         result = agent.invoke({"messages": [HumanMessage("Hi")]})
 
-        assert isinstance(result, GraphOutput)
-        result_dict = result.value
-        assert result_dict.get("structured_response") == {"key": "value"}
-        messages = result_dict["messages"]
+        assert result.value.get("structured_response") == {"key": "value"}
+        messages = result.value["messages"]
         assert len(messages) == 2
         assert messages[1].content == "Hello"
 

libs/langchain_v1/tests/unit_tests/agents/middleware/implementations/test_tool_call_limit_parallel_bug.py

@@ -1,16 +1,13 @@
-"""Debug: trace exactly what happens when middleware processes parallel calls.
+"""Test that middleware correctly handles parallel tool calls with limits.
 
-Instruments after_model and model_to_tools to see if the ToolMessage injected
-by middleware is visible to the routing edge.
+Verifies that when middleware blocks some parallel tool calls, only the
+permitted calls execute and interrupts propagate correctly.
 """
 
-from unittest.mock import patch
-
-from langchain_core.messages import AIMessage, HumanMessage, ToolCall, ToolMessage
+from langchain_core.messages import HumanMessage, ToolCall
 from langchain_core.tools import tool
-from langgraph.checkpoint.memory import InMemorySaver, MemorySaver
-from langgraph.graph import MessagesState, StateGraph
-from langgraph.types import GraphOutput, interrupt
+from langgraph.checkpoint.memory import MemorySaver
+from langgraph.types import interrupt
 
 from langchain.agents.factory import create_agent
 from langchain.agents.middleware.tool_call_limit import ToolCallLimitMiddleware
@@ -47,61 +44,22 @@ def test_instrument_middleware_and_routing() -> None:
     fruit_mw = ToolCallLimitMiddleware(tool_name="ask_fruit_expert", run_limit=1)
     veggie_mw = ToolCallLimitMiddleware(tool_name="ask_veggie_expert", run_limit=1)
 
-    # Wrap after_model to trace its return value
-    original_after_model = ToolCallLimitMiddleware.after_model
-
-    def traced_after_model(self, state, runtime):
-        result = original_after_model(self, state, runtime)
-        msgs = state.get("messages", [])
-        ai_msgs = [m for m in msgs if isinstance(m, AIMessage)]
-        tool_msgs = [m for m in msgs if isinstance(m, ToolMessage)]
-        print(f"\n  [{self.name}].after_model:")
-        print(f"    state has {len(msgs)} messages ({len(ai_msgs)} AI, {len(tool_msgs)} Tool)")
-        if ai_msgs:
-            last_ai = ai_msgs[-1]
-            print(f"    last AI tool_calls: {[tc['id'] for tc in last_ai.tool_calls]}")
-        print(f"    returning: {result}")
-        if result and "messages" in result:
-            for m in result["messages"]:
-                if isinstance(m, ToolMessage):
-                    print(f"      -> injecting ToolMessage(call_id={m.tool_call_id}, status={m.status})")
-        return result
-
     checkpointer = MemorySaver()
 
-    with patch.object(ToolCallLimitMiddleware, "after_model", traced_after_model):
-        agent = create_agent(
-            model=model,
-            tools=[ask_fruit_expert, ask_veggie_expert],
-            middleware=[fruit_mw, veggie_mw],
-            checkpointer=checkpointer,
-        )
+    agent = create_agent(
+        model=model,
+        tools=[ask_fruit_expert, ask_veggie_expert],
+        middleware=[fruit_mw, veggie_mw],
+        checkpointer=checkpointer,
+    )
 
-        # Also trace the model_to_tools edge
-        # Get the compiled graph and print its structure
-        print("\n=== Graph nodes ===")
-        for name in agent.nodes:
-            print(f"  {name}")
-
-        config = {"configurable": {"thread_id": "debug1"}}
-        result = agent.invoke(
-            {"messages": [HumanMessage("Tell me about apples and bananas")]},
-            config,
-        )
-
-    # v2 stream format: unwrap GraphOutput
-    assert isinstance(result, GraphOutput)
-    result_dict = result.value
-    interrupts = list(result.interrupts)
-    print(f"\n=== Results ===")
-    print(f"call_log:   {call_log}")
-    print(f"interrupts: {len(interrupts)}")
-
-    tool_msgs = [m for m in result_dict["messages"] if isinstance(m, ToolMessage)]
-    error_msgs = [m for m in tool_msgs if m.status == "error"]
-    print(f"error tool messages: {len(error_msgs)}")
-    for m in error_msgs:
-        print(f"  call_id={m.tool_call_id} content={m.content!r}")
+    config = {"configurable": {"thread_id": "debug1"}}
+    result = agent.invoke(
+        {"messages": [HumanMessage("Tell me about apples and bananas")]},
+        config,
+    )
 
     assert len(call_log) == 1, f"Expected 1 execution, got {len(call_log)}: {call_log}"
-    assert len(interrupts) == 1, f"Expected 1 interrupt, got {len(interrupts)}"
+    assert len(result.interrupts) == 1, (
+        f"Expected 1 interrupt, got {len(result.interrupts)}"
+    )

libs/langchain_v1/uv.lock

@@ -1983,7 +1983,7 @@ requires-dist = [
     { name = "langchain-perplexity", marker = "extra == 'perplexity'" },
     { name = "langchain-together", marker = "extra == 'together'" },
     { name = "langchain-xai", marker = "extra == 'xai'" },
-    { name = "langgraph", git = "https://github.com/langchain-ai/langgraph.git?subdirectory=libs%2Flanggraph&branch=1.1" },
+    { name = "langgraph", git = "https://github.com/langchain-ai/langgraph.git?subdirectory=libs%2Flanggraph&branch=sr%2F1.1-v2-default" },
     { name = "pydantic", specifier = ">=2.7.4,<3.0.0" },
 ]
 provides-extras = ["community", "anthropic", "openai", "azure-ai", "google-vertexai", "google-genai", "fireworks", "ollama", "together", "mistralai", "huggingface", "groq", "aws", "deepseek", "xai", "perplexity"]
@@ -2536,7 +2536,7 @@ wheels = [
 [[package]]
 name = "langgraph"
 version = "1.0.10rc1"
-source = { git = "https://github.com/langchain-ai/langgraph.git?subdirectory=libs%2Flanggraph&branch=1.1#dec15d13dc47e032886816a8cbd16e4e99d6d45a" }
+source = { git = "https://github.com/langchain-ai/langgraph.git?subdirectory=libs%2Flanggraph&branch=sr%2F1.1-v2-default#33289349a938ee0d6a8d3d5f01db989d43fc28d7" }
 dependencies = [
     { name = "langchain-core" },
     { name = "langgraph-checkpoint" },
@@ -2549,7 +2549,7 @@ dependencies = [
 [[package]]
 name = "langgraph-checkpoint"
 version = "4.0.1rc4"
-source = { git = "https://github.com/langchain-ai/langgraph.git?subdirectory=libs%2Fcheckpoint&branch=1.1#dec15d13dc47e032886816a8cbd16e4e99d6d45a" }
+source = { git = "https://github.com/langchain-ai/langgraph.git?subdirectory=libs%2Fcheckpoint&branch=sr%2F1.1-v2-default#33289349a938ee0d6a8d3d5f01db989d43fc28d7" }
 dependencies = [
     { name = "langchain-core" },
     { name = "ormsgpack" },
@@ -2558,7 +2558,7 @@ dependencies = [
 [[package]]
 name = "langgraph-prebuilt"
 version = "1.0.8"
-source = { git = "https://github.com/langchain-ai/langgraph.git?subdirectory=libs%2Fprebuilt&branch=1.1#dec15d13dc47e032886816a8cbd16e4e99d6d45a" }
+source = { git = "https://github.com/langchain-ai/langgraph.git?subdirectory=libs%2Fprebuilt&branch=sr%2F1.1-v2-default#33289349a938ee0d6a8d3d5f01db989d43fc28d7" }
 dependencies = [
     { name = "langchain-core" },
     { name = "langgraph-checkpoint" },
@@ -2567,7 +2567,7 @@ dependencies = [
 [[package]]
 name = "langgraph-sdk"
 version = "0.3.9"
-source = { git = "https://github.com/langchain-ai/langgraph.git?subdirectory=libs%2Fsdk-py&branch=1.1#dec15d13dc47e032886816a8cbd16e4e99d6d45a" }
+source = { git = "https://github.com/langchain-ai/langgraph.git?subdirectory=libs%2Fsdk-py&branch=sr%2F1.1-v2-default#33289349a938ee0d6a8d3d5f01db989d43fc28d7" }
 dependencies = [
     { name = "httpx" },
     { name = "orjson" },