tool-calling-in-langgraph

Tool Calling in LangGraph

Tool calling enables LangGraph workflows to use external tools (APIs, functions, databases) through standardized tool nodes that execute based on LLM decisions.

Core Pattern

from typing import Annotated
from typing_extensions import TypedDict
from langgraph.graph import StateGraph, START, END
from langgraph.graph.message import add_messages
from langgraph.prebuilt import ToolNode, tools_condition
from langchain_core.tools import tool
from langchain_anthropic import ChatAnthropic

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

# Define tools using @tool decorator
@tool
def search_function(query: str) -> str:
    """Search the web for information. Input should be a search query."""
    return f"Results for: {query}"

@tool
def calculator(a: int, b: int) -> int:
    """Perform mathematical calculations. Add two numbers together."""
    return a + b

tools = [search_function, calculator]

# Create tool node
tool_node = ToolNode(tools)

# Bind tools to LLM
llm = ChatAnthropic(model="claude-sonnet-4-5-20250929")
llm_with_tools = llm.bind_tools(tools)

# Build graph
workflow = StateGraph(AgentState)
workflow.add_node("agent", lambda state: {"messages": [llm_with_tools.invoke(state["messages"])]})
workflow.add_node("tools", tool_node)

# Add edges
workflow.add_edge(START, "agent")
workflow.add_conditional_edges("agent", tools_condition)
workflow.add_edge("tools", "agent")

# Compile
graph = workflow.compile()

ReAct Tool Loop

Using Prebuilt tools_condition (Recommended)

from langgraph.graph import StateGraph, START, END
from langgraph.prebuilt import ToolNode, tools_condition

workflow = StateGraph(AgentState)
workflow.add_node("agent", agent_node)
workflow.add_node("tools", tool_node)

# Use prebuilt condition - returns "tools" or "__end__"
workflow.add_conditional_edges("agent", tools_condition)
workflow.add_edge("tools", "agent")  # Loop back after tool use
workflow.add_edge(START, "agent")

graph = workflow.compile()