How to retrieve and use the stored memory in Langgraph?

[Jayapradha05]

Hi everyone,

I'm having an issue with using TypedDict and Annotated for storing chat history in a class. When I define my AgentState class like this, I can retrieve the chat history without any problems:

from typing import TypedDict, Annotated

class AgentState(TypedDict):
    messages: Annotated[list, add_messages]

However, when I extend the AgentState class to include more fields, I'm unable to retrieve the chat memory:

from typing import TypedDict, Annotated

class AgentState(TypedDict):
    input: str
    output: str
    decision: str
    messages: Annotated[list, add_messages]

In this case, I can no longer retrieve the chat memory. Does anyone know why this might be happening and how I can fix it? Any help would be greatly appreciated!

Additionally, I'd like to understand how the AgentState handles input. When I pass the input to the agent, how is it processed and stored within the AgentState structure? Any insights or explanations on the internal workings would be very helpful.

Thanks!

[vbarda]

hi @Jayapradha05 . could you please provide some additional code with how you're trying to retrieve chat memory (full graph definition + how you're invoking it)? i am assuming that by "chat history" you mean accessing messages key in the state.

as long as you invoke your graph with messages key to begin with, it should be available to all of the nodes/edges in the graph, as they all share access to the state, i.e.

graph.invoke({"messages": [("human", "hi")]})

if you want to retrieve the values of the state after the graph finishes executtion, you would need to also provide a checkpointer to your graph, see more examples here https://langchain-ai.github.io/langgraph/how-tos/persistence/

to answer your second question -- AgentState doesn't handle input, it is the input, it's a data structure that describes the schema of the state, specifically which keys will be available in the state, as well as annotations that specify how to handle state updates (i.e. reducers). state is not stored anywhere if you're not using a checkpointer, but it is passed around between nodes and edges as the graph is executed. see more information here https://langchain-ai.github.io/langgraph/concepts/low_level/#state

[Jayapradha05]

Hi @vbarda
Thank you for the detailed response. Here's the additional information you requested:

Full Graph Structure and Invocation Method
Here is the structure of my StateGraph:

from langchain_ai.langgraph import StateGraph
class AgentState(TypedDict):
    input: str
    output: str
    decision: str
    messages: Annotated[list, add_messages]

memory = SqliteSaver.from_conn_string(":memory:")

workflow = StateGraph(AgentState)

workflow.add_node("analyze", analyze_question)
workflow.add_node("code_agent", answer_code_question)
workflow.add_node("generic_agent", answer_generic_question)
workflow.add_node("rag_agent", answer_rag_question_)

workflow.add_conditional_edges(
    "analyze",
    lambda x: x["decision"],
    {
        "code": "code_agent",
        "rag": "rag_agent",
        "general": "generic_agent"
    }
)

workflow.set_entry_point("analyze")
workflow.add_edge("code_agent", END)
workflow.add_edge("rag_agent", END)
workflow.add_edge("generic_agent", END)

app = workflow.compile(checkpointer=memory)

config = {"configurable": {"thread_id": 5}}
user_input = "can you tell me which food is famous in Japan."

# The config is the second positional argument to stream() or invoke()!
events = app.stream(
    {"messages": [("user", user_input)]}, config, stream_mode="values"
)
for event in events:
    event["messages"][-1].pretty_print()

The Issue
After some questions, when I ask about which country I'm talking about, the model does not remember the country. Here's an example of such a scenario:

User Input: "Can you tell me which food is famous in Japan?"
Response: "Sushi is famous in Japan."
User Input: "Which country am I talking about?"
Response: The model does not remember the country.

Could you please provide insights on how to ensure that the state, especially messages, is preserved and utilized correctly across multiple interactions? Do I need to adjust the way I'm invoking the graph or modify the AgentState definition to maintain the context better?

Thank you for your assistance!