---

🧰 LangGraph Agent Toolkit

CI/Testing
Package
Meta

📋 Introduction

A comprehensive toolkit for building, deploying, and managing AI agents using LangGraph, FastAPI, and Streamlit. It provides a production-ready framework for creating conversational AI agents with features like multi-provider LLM support, streaming responses, observability, memory and prompt management.

What is langGraph-agent-toolkit?

The langgraph-agent-toolkit is a full-featured framework for developing and deploying AI agent services. Built on the foundation of:

• LangGraph for agent creation with advanced flows and human-in-the-loop capabilities
• FastAPI for robust, high-performance API services with streaming support
• Streamlit for intuitive user interfaces

Key components include:

• Data structures and settings built with Pydantic
• LiteLLM proxy for universal multi-provider LLM support
• Comprehensive memory management and persistence using PostgreSQL/SQLite
• Advanced observability tooling via Langfuse and Langsmith
• Modular architecture allowing customization while maintaining a consistent application structure

Whether you're building a simple chatbot or complex multi-agent system, this toolkit provides the infrastructure to develop, test, and deploy your LangGraph-based agents with confidence.

You can use DeepWiki to learn more about this repository.

📑 Contents

• Introduction
• Quickstart
• Installation Options
• Architecture
• Key Features
• Environment Setup
• Project Structure
• Setup and Usage
• Documentation
• Useful Resources
• Development and Contributing
• License

🚀 Quickstart

1. Create a .env file based on .env.example

2. Option 1: Run with Python from source

   # Install dependencies
   pip install uv
   uv sync --frozen
   source .venv/bin/activate
   
   # Start the service
   python langgraph_agent_toolkit/run_api.py
   
   # In another terminal
   source .venv/bin/activate
   streamlit run langgraph_agent_toolkit/run_app.py

3. Option 2: Run with Python from PyPi repository

   pip install langgraph-agent-toolkit

   ℹ️ For more details on installation options, see the Installation Documentation.

4. Option 3: Run with Docker

   docker compose watch

📦 Installation Options

The toolkit supports multiple installation options using "extras" to include just the dependencies you need.

For detailed installation instructions and available extras, see the Installation Documentation.

🏗️ Architecture

✨ Key Features

1. LangGraph Integration

   • Latest LangGraph v0.3 features
   • Human-in-the-loop with interrupt()
   • Flow control with Command and langgraph-supervisor

2. API Service

   • FastAPI with streaming and non-streaming endpoints
   • Support for both token-based and message-based streaming
   • Multiple agent support with URL path routing
   • Available agents and models listed at /info endpoint
   • Supports different runners (unicorn, gunicorn, mangum, azure functions)

3. Developer Experience

   • Asynchronous design with async/await
   • Docker configuration with live reloading
   • Comprehensive testing suite

4. Enterprise Components

   • Configurable PostgreSQL/SQLite connection pools
   • Observability via Langfuse and Langsmith
   • User feedback system
   • Prompt management system
   • LiteLLM proxy integration

For more details on features, see the Usage Documentation.

⚙️ Environment Setup

For detailed environment setup instructions, including creating your .env file and configuring LiteLLM, see the Environment Setup Documentation.

📂 Project Structure

The repository contains:

• langgraph_agent_toolkit/agents/blueprints/: Agent definitions
• langgraph_agent_toolkit/agents/agent_executor.py: Agent execution control
• langgraph_agent_toolkit/schema/: Protocol schema definitions
• langgraph_agent_toolkit/core/: Core modules (LLM, memory, settings)
• langgraph_agent_toolkit/service/service.py: FastAPI service
• langgraph_agent_toolkit/client/client.py: Service client
• langgraph_agent_toolkit/run_app.py: Chat interface
• docker/: Docker configurations
• tests/: Test suite

🛠️ Setup and Usage

For detailed setup and usage instructions, including building your own agent, Docker setup, using the AgentClient, and local development, see the Usage Documentation.

📚 Documentation

Full documentation is available at GitHub repository and includes:

• Installation Guide
• Environment Setup
• Usage Guide

📚 Useful Resources

• LangGraph documentation
• LangGraph Memory Concept
• LangGraph Memory Persistence
• LangGraph Memory Template
• LangGraph Human in the Loop
• LangGraph 101 - blueprints
• LangGraph - Examples
• Complex data extraction with function calling
• How to edit graph state
• Memory in the background
• Building an agent with LangGraph
• How to create tools in Langchain
• Simple Serverless FastAPI with AWS Lambda
• LangGraph Middleware

👥 Development and Contributing

Thank you for considering contributing to Langgraph Agent Toolkit! We encourage the community to post Issues and Pull Requests.

Before you get started, please see our Contribution Guide.

📄 License

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