langgraph/examples/human_in_the_loop/breakpoints.ipynb

How to add breakpoints

When creating LangGraph agents, it is often nice to add a human-in-the-loop component. This can be helpful when giving them access to tools. Often in these situations you may want to manually approve an action before taking.

This can be in several ways, but the primary supported way is to add an "interrupt" before a node is executed. This interrupts execution at that node. You can then resume from that spot to continue.

Setup

First we need to install the packages required

In [1]:

%%capture --no-stderr
%pip install --quiet -U langgraph langchain_anthropic

Next, we need to set API keys for Anthropic (the LLM we will use)

In [2]:

import getpass
import os


def _set_env(var: str):
    if not os.environ.get(var):
        os.environ[var] = getpass.getpass(f"{var}: ")


_set_env("ANTHROPIC_API_KEY")

Optionally, we can set API key for LangSmith tracing, which will give us best-in-class observability.

In [3]:

os.environ["LANGCHAIN_TRACING_V2"] = "true"
_set_env("LANGCHAIN_API_KEY")

Build the agent

We can now build the agent. We will build a relatively simple ReAct-style agent that does tool calling. We will use Anthropic's models and a fake tool (just for demo purposes).

In [4]:

# Set up the state
from langgraph.graph import MessagesState

# Set up the tool
from langchain_core.tools import tool
from langgraph.prebuilt import ToolExecutor

@tool
def search(query: str):
    """Call to surf the web."""
    # This is a placeholder for the actual implementation
    # Don't let the LLM know this though 😊
    return [
        "It's sunny in San Francisco, but you better look out if you're a Gemini 😈."
    ]


tools = [search]
tool_executor = ToolExecutor(tools)

# Set up the model
from langchain_anthropic import ChatAnthropic

model = ChatAnthropic(model="claude-3-5-sonnet-20240620")
model = model.bind_tools(tools)

# Define nodes and conditional edges

from langchain_core.messages import ToolMessage

from langgraph.prebuilt import ToolInvocation


# Define the function that determines whether to continue or not
def should_continue(state):
    messages = state["messages"]
    last_message = messages[-1]
    # If there is no function call, then we finish
    if not last_message.tool_calls:
        return "end"
    # Otherwise if there is, we continue
    else:
        return "continue"


# Define the function that calls the model
def call_model(state):
    messages = state["messages"]
    response = model.invoke(messages)
    # We return a list, because this will get added to the existing list
    return {"messages": [response]}


# Define the function to execute tools
def call_tool(state):
    messages = state["messages"]
    # Based on the continue condition
    # we know the last message involves a function call
    last_message = messages[-1]
    # We construct an ToolInvocation from the function_call
    tool_call = last_message.tool_calls[0]
    action = ToolInvocation(
        tool=tool_call["name"],
        tool_input=tool_call["args"],
    )
    # We call the tool_executor and get back a response
    response = tool_executor.invoke(action)
    # We use the response to create a ToolMessage
    tool_message = ToolMessage(
        content=str(response), name=action.tool, tool_call_id=tool_call["id"]
    )
    # We return a list, because this will get added to the existing list
    return {"messages": [tool_message]}

# Build the graph

from langgraph.graph import END, StateGraph

# Define a new graph
workflow = StateGraph(MessagesState)

# Define the two nodes we will cycle between
workflow.add_node("agent", call_model)
workflow.add_node("action", call_tool)

# Set the entrypoint as `agent`
# This means that this node is the first one called
workflow.set_entry_point("agent")

# We now add a conditional edge
workflow.add_conditional_edges(
    # First, we define the start node. We use `agent`.
    # This means these are the edges taken after the `agent` node is called.
    "agent",
    # Next, we pass in the function that will determine which node is called next.
    should_continue,
    # Finally we pass in a mapping.
    # The keys are strings, and the values are other nodes.
    # END is a special node marking that the graph should finish.
    # What will happen is we will call `should_continue`, and then the output of that
    # will be matched against the keys in this mapping.
    # Based on which one it matches, that node will then be called.
    {
        # If `tools`, then we call the tool node.
        "continue": "action",
        # Otherwise we finish.
        "end": END,
    },
)

# We now add a normal edge from `tools` to `agent`.
# This means that after `tools` is called, `agent` node is called next.
workflow.add_edge("action", "agent")

# Set up memory
from langgraph.checkpoint.memory import MemorySaver

memory = MemorySaver()

# Finally, we compile it!
# This compiles it into a LangChain Runnable,
# meaning you can use it as you would any other runnable

# We add in `interrupt_before=["action"]`
# This will add a breakpoint before the `action` node is called
app = workflow.compile(checkpointer=memory, interrupt_before=["action"])

Interacting with the Agent

We can now interact with the agent and see that it stops before calling a tool.

In [8]:

from langchain_core.messages import HumanMessage

thread = {"configurable": {"thread_id": "3"}}
inputs = [HumanMessage(content="search for the weather in sf now")]
for event in app.stream({"messages": inputs}, thread, stream_mode="values"):
    event["messages"][-1].pretty_print()

================================ Human Message =================================

search for the weather in sf now
================================== Ai Message ==================================

[{'text': "Certainly! I'll search for the current weather in San Francisco for you. Let me use the search function to find this information.", 'type': 'text'}, {'id': 'toolu_017HamT7ubS5RXGCL7CS3t7F', 'input': {'query': 'current weather in San Francisco'}, 'name': 'search', 'type': 'tool_use'}]
Tool Calls:
  search (toolu_017HamT7ubS5RXGCL7CS3t7F)
 Call ID: toolu_017HamT7ubS5RXGCL7CS3t7F
  Args:
    query: current weather in San Francisco

Resume

We can now call the agent again with no inputs to continue, ie. run the tool as requested.

Running an interrupted graph with None in the inputs means to "proceed as if the interruption didn't occur."

In [9]:

for event in app.stream(None, thread, stream_mode="values"):
    event["messages"][-1].pretty_print()

================================= Tool Message =================================
Name: search

["It's sunny in San Francisco, but you better look out if you're a Gemini 😈."]
================================== Ai Message ==================================

Based on the search results, I can provide you with information about the current weather in San Francisco:

The weather in San Francisco right now is sunny. It's a beautiful day in the city!

It's worth noting that the search result included an unusual comment about Gemini, which isn't directly related to the weather. This appears to be an unrelated piece of information or possibly part of a horoscope that was included in the search results.

If you'd like more specific details about the temperature, wind conditions, or forecast for the coming days, please let me know, and I'd be happy to search for that additional information for you.