PyAgenity Tutorials

Welcome to PyAgenity! This tutorial series will guide you through building intelligent agents and multi-agent workflows, from basic graph construction to advanced patterns like streaming, persistence, and tool integration.

🎯 What You'll Learn

PyAgenity is a lightweight Python framework for building agent graphs on top of LiteLLM. By the end of these tutorials, you'll understand how to:

• Build and execute agent workflows using StateGraph and nodes
• Manage conversation state and message flow with AgentState
• Create tool-calling agents using ToolNode and dependency injection
• Add persistence with checkpointers and memory stores
• Stream real-time responses and monitor execution events
• Use prebuilt agent patterns for common scenarios

🚀 Prerequisites

Before diving in, ensure you have:

• Python 3.12+ installed
• Basic familiarity with async/await patterns
• Experience with LLM APIs (OpenAI, Gemini, etc.)
• Comfort with command-line tools and environment variables

Quick Setup

1. Install PyAgenity with your preferred LLM provider:

   pip install pyagenity[litellm]
   # Optional: add persistence and tools
   pip install pyagenity[pg_checkpoint,mcp]
   

2. Set up environment variables in .env:

   # For LiteLLM examples
   OPENAI_API_KEY=your_openai_key
   # Or use Gemini
   # GEMINI_API_KEY=your_gemini_key
   

3. Clone examples to experiment:

   git clone https://github.com/Iamsdt/PyAgenity.git
   cd PyAgenity/examples/react
   python react_sync.py  # Your first agent!
   

📚 Tutorial Path

Follow these tutorials in order for the best learning experience:

🏗️ Foundation

1. Graph Fundamentals - Build your first agent with StateGraph, nodes, and edges
2. State & Messages - Master conversation state and message schemas
3. Tools & Dependency Injection - Create tool-calling agents with ToolNode
4. React Agent Patterns - Complete guide to ReAct agents: basic patterns, DI, MCP, streaming

🔀 Control & Flow

1. Control Flow & Routing - Conditional edges, interrupts, and error handling
2. Persistence & Memory - Save state with checkpointers and stores
3. Streaming & Events - Real-time responses and observability

🎯 Advanced Patterns

1. Prebuilt Agents & Orchestration - Ready-to-use patterns and multi-agent workflows

💡 Learning Tips

• Run the examples: Every tutorial references working code in examples/. Clone, modify, and experiment!
• Start simple: Build a basic graph first, then add complexity gradually
• Use the console: The ConsolePublisher shows you what's happening under the hood
• Debug with state: Use ResponseGranularity.FULL to inspect complete execution state

📖 Additional Resources

• API Reference - Detailed documentation for all classes and methods
• Examples Directory - Runnable code for every major pattern
• PyProject.toml - Optional dependencies and their features

🔗 Quick Navigation

Tutorial	Focus	Key Files
Graph Fundamentals	StateGraph, nodes, compilation	examples/react/react_sync.py
State & Messages	AgentState, message handling	pyagenity/state/, pyagenity/utils/message.py
Tools & DI	ToolNode, dependency injection	examples/react-injection/, examples/react-mcp/
React Agents	Complete ReAct guide: basic, DI, MCP, streaming	examples/react*/
Control Flow	Conditional routing, interrupts	examples/react/react_weather_agent.py
Persistence	Checkpointers, stores	pyagenity/checkpointer/, pyagenity/store/
Streaming	Real-time responses, events	examples/react_stream/
Advanced	Prebuilt agents, orchestration	pyagenity/prebuilt/agent/

---

Ready to build your first agent? Start with Graph Fundamentals!