graph-construction-in-langgraph

Graph Construction in LangGraph

Purpose

This skill provides guidance on constructing LangGraph workflows using StateGraph. Graph construction transforms isolated node functions into orchestrated workflows with explicit control flow, enabling complex multi-step agentic systems.

Compatibility

This skill is compatible with LangGraph 1.x (tested with v1.0.5, December 2025).

Feature-specific version requirements:

• interrupt() function: requires LangGraph >= 0.2.31
• add_sequence() method: requires LangGraph >= 0.2.46

When to Use This Skill

Use this skill when:

• Creating a new LangGraph workflow from scratch
• Adding nodes or edges to existing graphs
• Defining entry points and exit conditions
• Setting up conditional routing between nodes
• Compiling and visualizing graph structures
• Troubleshooting graph connectivity issues

Core Concepts

StateGraph Basics

StateGraph is the foundation of LangGraph workflows:

from langgraph.graph import StateGraph, START, END
from typing import TypedDict  # Use typing_extensions.TypedDict for Python 3.9

class State(TypedDict):
    messages: list
    step: str

# Create graph with state schema
workflow = StateGraph(State)

Using MessagesState for Chat Applications

For chat-based applications, LangGraph provides a built-in MessagesState with automatic message management:

from langgraph.graph import StateGraph, MessagesState, START, END

# MessagesState provides a 'messages' field with add_messages reducer
workflow = StateGraph(MessagesState)

def chat_node(state: MessagesState) -> dict:
    # Access messages from state
    last_message = state["messages"][-1]
    response = generate_response(last_message)  # Your LLM call here
    return {"messages": [response]}

workflow.add_node("chat", chat_node)
workflow.add_edge(START, "chat")
workflow.add_edge("chat", END)

MessagesState automatically handles message deduplication and updates based on message IDs.

Adding Nodes