Skip to main content
Deprecated: MessageGraph is deprecated in LangGraph v1.0.0 and will be removed in v2.0.0. Please use StateGraph with a messages key instead.
The MessageGraph class is a specialized version of StateGraph where the entire state is a single, append-only list of messages. Defined in: langgraph/graph/message.py:251

​
Overview

MessageGraph is a subclass of StateGraph whose entire state is a single list of messages. Each node in a MessageGraph takes a list of messages as input and returns zero or more messages as output. The add_messages function is used to merge the output messages from each node into the existing list of messages in the graph’s state.

​
Constructor

MessageGraph()
Create a new MessageGraph instance. The state schema is automatically set to Annotated[list[AnyMessage], add_messages].

​
Usage Example

from langgraph.graph.message import MessageGraph
from langchain_core.messages import HumanMessage, AIMessage

builder = MessageGraph()

def chatbot(state: list):
    # state is a list of messages
    return [AIMessage(content="Hello!")]

builder.add_node("chatbot", chatbot)
builder.set_entry_point("chatbot")
builder.set_finish_point("chatbot")

graph = builder.compile()
result = graph.invoke([HumanMessage(content="Hi there.")])
# [
#   HumanMessage(content="Hi there.", id='...'),
#   AIMessage(content="Hello!", id='...')
# ]

​
Migration Guide

Instead of using MessageGraph, use StateGraph with MessagesState or a custom state that includes a messages field:

​
Before (MessageGraph)

from langgraph.graph.message import MessageGraph
from langchain_core.messages import AIMessage

builder = MessageGraph()

def chatbot(state: list):
    return [AIMessage(content="Hello!")]

builder.add_node("chatbot", chatbot)
builder.set_entry_point("chatbot")
builder.set_finish_point("chatbot")
graph = builder.compile()

result = graph.invoke([("user", "Hi there.")])

​
After (StateGraph with MessagesState)

from langgraph.graph import StateGraph, MessagesState, START, END
from langchain_core.messages import AIMessage

builder = StateGraph(MessagesState)

def chatbot(state: MessagesState):
    return {"messages": [AIMessage(content="Hello!")]}

builder.add_node("chatbot", chatbot)
builder.add_edge(START, "chatbot")
builder.add_edge("chatbot", END)
graph = builder.compile()

result = graph.invoke({"messages": [("user", "Hi there.")]})

​
Alternative: Custom State

from typing import Annotated
from typing_extensions import TypedDict
from langgraph.graph import StateGraph, add_messages, START, END
from langchain_core.messages import AnyMessage, AIMessage

class State(TypedDict):
    messages: Annotated[list[AnyMessage], add_messages]
    # Add other fields as needed
    user_id: str

builder = StateGraph(State)

def chatbot(state: State):
    user_id = state["user_id"]
    return {"messages": [AIMessage(content=f"Hello, user {user_id}!")]}

builder.add_node("chatbot", chatbot)
builder.add_edge(START, "chatbot")
builder.add_edge("chatbot", END)
graph = builder.compile()

result = graph.invoke({
    "messages": [("user", "Hi there.")],
    "user_id": "123"
})

​
Methods

MessageGraph inherits all methods from StateGraph. See the StateGraph reference for detailed documentation of available methods:
  • add_node() - Add a node to the graph
  • add_edge() - Add an edge between nodes
  • add_conditional_edges() - Add conditional routing
  • set_entry_point() - Set the starting node
  • set_finish_point() - Set the ending node
  • compile() - Compile into an executable graph

​
Examples

​
Basic Message Workflow

from langgraph.graph.message import MessageGraph
from langchain_core.messages import HumanMessage, AIMessage, ToolMessage

builder = MessageGraph()

builder.add_node(
    "chatbot",
    lambda state: [
        AIMessage(
            content="Hello!",
            tool_calls=[{"name": "search", "id": "123", "args": {"query": "X"}}],
        )
    ],
)

builder.add_node(
    "search",
    lambda state: [ToolMessage(content="Searching...", tool_call_id="123")]
)

builder.set_entry_point("chatbot")
builder.add_edge("chatbot", "search")
builder.set_finish_point("search")

graph = builder.compile()

result = graph.invoke([HumanMessage(content="Hi there. Can you search for X?")])
print(result)
# [
#   HumanMessage(content="Hi there. Can you search for X?", id='...'),
#   AIMessage(content="Hello!", tool_calls=[...], id='...'),
#   ToolMessage(content="Searching...", tool_call_id="123", id='...')
# ]

​
With Conditional Edges

from langgraph.graph.message import MessageGraph
from langchain_core.messages import HumanMessage, AIMessage

builder = MessageGraph()

def should_continue(state: list) -> str:
    last_message = state[-1]
    if hasattr(last_message, 'tool_calls') and last_message.tool_calls:
        return "tools"
    return "end"

builder.add_node("agent", lambda state: [AIMessage(content="Processing...")])
builder.add_node("tools", lambda state: [AIMessage(content="Tool result")])

builder.set_entry_point("agent")
builder.add_conditional_edges(
    "agent",
    should_continue,
    {"tools": "tools", "end": "__end__"}
)
builder.add_edge("tools", "agent")

graph = builder.compile()

​
See Also

Build docs developers (and LLMs) love

Get started for freeTalk to us