LangGraph Examples

December 4, 2025 ยท View on GitHub

Updated 2025: This directory contains examples and templates for building AI agents using LangGraph, a framework for orchestrating multi-agent workflows.

LangGraph v1.0 (October 22, 2025): LangGraph reached v1.0, marking the first stable major release in the durable agent framework space. It offers node-based architecture for building multi-agent systems with structured state management, durable state persistence, built-in persistence for saving/resuming workflows, and human-in-the-loop patterns. The LangGraph Platform enables teams to run agent workflows at scale with tools to track, debug, and manage them in production.

Overview

This folder provides starter code and examples to help you build AI agents with LangGraph. It includes basic and advanced examples covering various aspects of agent development, such as multi-agent orchestration, workflow management, and more.

2025 Features:

  • Production-ready platform for running agent workflows at scale
  • Advanced tracing and debugging tools
  • Structured state management for complex multi-agent systems
  • Visual workflow debugging and monitoring

Prerequisites

  • Python 3.8 or higher
  • OpenAI API key (set as an environment variable or in a .env file)

Usage

  1. Install dependencies: 
    pip install langgraph openai
  2. Set your OpenAI API key: 
    export OPENAI_API_KEY="your_api_key_here"
  3. Explore the docs and code examples to build your own agents.

Contents

Docs

Code

  • Starter codes for building agents with LangGraph can be found in the code including:
    • 0_hello_world_graph.py: Simple "Hello World" example for LangGraph.
    • 0_langgraph_hello_world.py: Another basic example for LangGraph.
    • 1_build_basic_chatbot.py: Basic chatbot implementation.
    • 2_add_tools_chatbot.py: Chatbot with tool integration.
    • 3_add_memory_chatbot.py: Chatbot with memory and context management.
    • advanced_workflow.py: Advanced workflow example.
    • basic_workflow.py: Basic workflow example.
    • complete_chatbot.py: Full-featured chatbot implementation.
    • customer_support_bot.py: Customer support bot example.

Key Concepts Demonstrated

State Management

  • Using TypedDict for type-safe state
  • Managing conversation history
  • Handling multiple state fields

Graph Structure

  • Node creation and connection
  • Conditional routing
  • Entry point definition
  • Edge management

Tool Integration

  • Custom tool definition with @tool decorator
  • Function calling from agents
  • Tool result handling

Memory and Context

  • Conversation history management
  • Context preservation between turns
  • State persistence

Human Interaction

  • Human-in-the-loop capabilities
  • Input handling
  • Interactive conversation flow

Output Files

Each example generates an output file:

  • basic_workflow_output.txt
  • advanced_workflow_output.txt
  • complete_chatbot_output.txt

These files contain the conversation history and results of each run.

Best Practices

  1. State Management

    • Use TypedDict for type-safe state
    • Keep state minimal and focused
    • Handle state updates carefully

  2. Tool Design

    • Make tools focused and specific
    • Include clear documentation
    • Handle errors gracefully

  3. Graph Structure

    • Keep the graph simple and clear
    • Use meaningful node names
    • Document the flow between nodes

  4. Error Handling

    • Implement proper error handling
    • Provide meaningful error messages
    • Handle edge cases appropriately

Resources

Contributing

Feel free to:

  1. Add new examples
  2. Improve existing examples
  3. Add more documentation
  4. Report issues or suggest improvements

License

This project is licensed under the MIT License - see the LICENSE file for details.