"""
Pattern 23.02: Tool-Use
Agents use external tools and APIs
"""
from typing import TypedDict
from langgraph.graph import StateGraph, END
from langchain_openai import ChatOpenAI
from langchain_core.messages import HumanMessage


class ToolUseState(TypedDict):
    """State for Tool-Use pattern."""
    input: str
    output: str
    metadata: dict


def agent_function(state: ToolUseState) -> ToolUseState:
    """Tool-Use agent implementation."""
    llm = ChatOpenAI()
    prompt = f"Process using Tool-Use: {state['input']}"
    response = llm.invoke([HumanMessage(content=prompt)])
    state["output"] = str(response.content)
    state["metadata"] = {"pattern": "Tool-Use", "status": "completed"}
    return state


def create_tool_use():
    """Create Tool-Use workflow."""
    workflow = StateGraph(ToolUseState)
    workflow.add_node("agent", agent_function)
    workflow.set_entry_point("agent")
    workflow.add_edge("agent", END)
    return workflow.compile()


if __name__ == "__main__":
    workflow = create_tool_use()
    result = workflow.invoke({"input": "Test Tool-Use", "output": "", "metadata": {}})
    print(f"\n=== Tool-Use Pattern ===\nOutput: {result['output'][:200]}...")

    

Agentic Orchestration Pattern: Tool-Use Pattern

02 Tool Use Pattern – Detailed Documentation

Overview

File: 23_domain_specific/02_tool_use.py
Pattern Category: 23 Domain Specific
Implementation: LangGraph/CrewAI

Pattern Description

Pattern 23.02: Tool-Use Agents use external tools and APIs

---

Code Structure

Imports

from typing import TypedDict
from langgraph.graph import StateGraph, END
from langchain_openai import ChatOpenAI
from langchain_core.messages import HumanMessage

Purpose of imports: – TypedDict: For type-safe state definitions – StateGraph: LangGraph’s graph-based workflow orchestration – ChatOpenAI: OpenAI LLM integration – HumanMessage: LangChain message formatting

State Definition

class ToolUseState(TypedDict):
    """State for Tool-Use pattern."""
    input: str
    output: str
    metadata: dict


def agent_function(state: ToolUseState) -> ToolUseState:

ToolUseState: State for Tool-Use pattern.

State Fields: – input: Stores input – output: Stores output – metadata: Stores metadata – def agent_function(state: Stores def agent function(state – prompt = f"Process using Tool-Use: Stores prompt = f”Process using Tool-Use – state["metadata"] = {"pattern": Stores state[“metadata”] = {“pattern”

---

Functions

agent_function()

Line: 18
Arguments: state

Tool-Use agent implementation.

def agent_function(state: ToolUseState) -> ToolUseState:
    """Tool-Use agent implementation."""
    llm = ChatOpenAI()
    prompt = f"Process using Tool-Use: {state['input']}"
    response = llm.invoke([HumanMessage(content=prompt)])
    state["output"] = str(response.content)
    state["metadata"] = {"pattern": "Tool-Use", "status": "completed"}
    return state


def create_tool_use():

create_tool_use()

Line: 28
Arguments: “

Create Tool-Use workflow.

def create_tool_use():
    """Create Tool-Use workflow."""
    workflow = StateGraph(ToolUseState)
    workflow.add_node("agent", agent_function)
    workflow.set_entry_point("agent")
    workflow.add_edge("agent", END)
    return workflow.compile()


if __name__ == "__main__":

---

Line-by-Line Explanation

Line 5: Import statement

from typing import TypedDict

Imports TypedDict for creating type-safe state dictionaries.

Line 6: Import statement

from langgraph.graph import StateGraph, END

Imports StateGraph – the core LangGraph class for building workflows.

Line 7: Import statement

from langchain_openai import ChatOpenAI

Imports ChatOpenAI for LLM interactions.

Line 8: Import statement

from langchain_core.messages import HumanMessage

Imports HumanMessage for formatting prompts to the LLM.

Line 11: Class definition

class ToolUseState(TypedDict):

Defines the state structure using TypedDict for type safety.

Line 12: Module docstring

    """State for Tool-Use pattern."""

Describes the pattern and its purpose.

Line 18: Function agent_function definition

def agent_function(state: ToolUseState) -> ToolUseState:

Defines the agent_function function that processes the workflow state.

Line 19: Module docstring

    """Tool-Use agent implementation."""

Describes the pattern and its purpose.

Line 28: Function create_tool_use definition

def create_tool_use():

Defines the create_tool_use function that processes the workflow state.

Line 29: Module docstring

    """Create Tool-Use workflow."""

Describes the pattern and its purpose.

---

Usage Example

if __name__ == "__main__":
    workflow = create_tool_use()
    result = workflow.invoke({"input": "Test Tool-Use", "output": "", "metadata": {}})
    print(f"\n=== Tool-Use Pattern ===\nOutput: {result['output'][:200]}...")

---

Key Concepts

1. State Management: Uses TypedDict to maintain workflow state
2. Graph-Based Flow: LangGraph orchestrates execution flow
3. Agent Functions: Each node processes and updates state
4. LLM Integration: Uses ChatOpenAI for intelligent processing

Pattern Benefits

• ✅ Type-safe state management
• ✅ Clear workflow visualization
• ✅ Modular and reusable components
• ✅ Easy to test and debug
• ✅ Scalable architecture

Related Patterns

• See other patterns in the same category
• Explore combination possibilities
• Check the main documentation for pattern relationships