Langgraph ainvoke with error: "Must write to at least one of []"

[mrctito]

Checked other resources

• I added a very descriptive title to this question.
• I searched the LangChain documentation with the integrated search.
• I used the GitHub search to find a similar question and didn't find it.

Commit to Help

• I commit to help with one of those options 👆

Example Code

I reproduced almost the same code that I found in examples in the official LangGraph documentation.

Relevant points from my source code:

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

self.tool_node = ToolNode(tools=self.tools)

agent_runnable = create_tool_calling_agent(llm=self.llm_with_tools, tools=self.tools, prompt=prompt_obj)

async def call_model_b(self, state):
messages = state["messages"]
response = await self.agent_runnable.ainvoke(messages)
return {"messages": [response]}

async def should_continue_b(self, state) -> Literal["end", "continue"]:
messages = state["messages"]
last_message = messages[-1]
# If there is no tool call, then we finish
if not last_message.tool_calls:
return "end"
# Otherwise if there is, we continue
else:
return "continue"

graph_builder = StateGraph(StateGraph)
graph_builder.add_node("agent", self.call_model_b)
graph_builder.add_node("action", self.tool_node)
graph_builder.set_entry_point("agent")

graph_builder.add_conditional_edges(
"agent",
self.should_continue_b,
{
# If tools, then we call the tool node.
"continue": "action",
# Otherwise we finish.
"end": END
}
)

graph_builder.add_edge("action", "agent")
graph = graph_builder.compile()

result = await self.graph.ainvoke(inputs, debug=True)

This is where the error happens: "Must write to at least one of []"

Description

Dear friends,

I need help with this error: "Must write to at least one of []"

Context: I cannot get my Langgraph application to work. I reproduced almost the same code that I found in examples in the official LangGraph documentation.

Relevant points from my source code:

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

self.tool_node = ToolNode(tools=self.tools)

agent_runnable = create_tool_calling_agent(llm=self.llm_with_tools, tools=self.tools, prompt=prompt_obj)

async def call_model_b(self, state):
messages = state["messages"]
response = await self.agent_runnable.ainvoke(messages)
return {"messages": [response]}

async def should_continue_b(self, state) -> Literal["end", "continue"]:
messages = state["messages"]
last_message = messages[-1]
# If there is no tool call, then we finish
if not last_message.tool_calls:
return "end"
# Otherwise if there is, we continue
else:
return "continue"

graph_builder = StateGraph(StateGraph)
graph_builder.add_node("agent", self.call_model_b)
graph_builder.add_node("action", self.tool_node)
graph_builder.set_entry_point("agent")

graph_builder.add_conditional_edges(
"agent",
self.should_continue_b,
{
# If tools, then we call the tool node.
"continue": "action",
# Otherwise we finish.
"end": END
}
)

graph_builder.add_edge("action", "agent")
graph = graph_builder.compile()

result = await self.graph.ainvoke(inputs, debug=True)

This is where the error happens: "Must write to at least one of []"

The error message is simply: "Must write to at least one of []"

Can someone please help me?

Thank you!

System Info

langchain==0.2.6
langchain-community==0.2.6
langchain-core==0.2.10
langchain-experimental==0.0.62
langchain-openai==0.1.13
langchain-qdrant==0.1.0
langchain-text-splitters==0.2.1

[dosubot[bot]]

To resolve the "Must write to at least one of []" error in your LangGraph application, you need to ensure that your graph nodes are correctly configured to write to at least one state key. This error typically occurs when a node does not produce any output that updates the state.

In your provided code, each node function should return a dictionary that updates the state. For example, the call_model_b function should return a dictionary with the updated messages, and the should_continue_b function should return a string indicating the next step.

Here is an example of how to ensure that each node function updates the state correctly:

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

self.tool_node = ToolNode(tools=self.tools)

agent_runnable = create_tool_calling_agent(llm=self.llm_with_tools, tools=self.tools, prompt=prompt_obj)

async def call_model_b(self, state):
    messages = state["messages"]
    response = await self.agent_runnable.ainvoke(messages)
    return {"messages": messages + [response]}  # Ensure the state is updated with the new messages

async def should_continue_b(self, state) -> Literal["end", "continue"]:
    messages = state["messages"]
    last_message = messages[-1]
    # If there is no tool call, then we finish
    if not last_message.tool_calls:
        return "end"
    # Otherwise if there is, we continue
    else:
        return "continue"

graph_builder = StateGraph(AgentState)
graph_builder.add_node("agent", self.call_model_b)
graph_builder.add_node("action", self.tool_node)
graph_builder.set_entry_point("agent")

graph_builder.add_conditional_edges(
    "agent",
    self.should_continue_b,
    {
        # If tools, then we call the tool node.
        "continue": "action",
        # Otherwise we finish.
        "end": END
    }
)

graph_builder.add_edge("action", "agent")
graph = graph_builder.compile()

result = await self.graph.ainvoke(inputs, debug=True)

Ensure that each node function in your graph returns a dictionary that updates the state with the necessary keys. This will resolve the "Must write to at least one of []" error ^[1].

To continue talking to Dosu, mention @dosu.

[GaryFail]

@mrctito Not sure if you ever solved this but I've just came across the same error and what solved it for me was changing the state to have a base class of MessagesState from langgraph.graph instead of TypedDict