AgentPatterns Examples: Python AI Agent Patterns (ReAct, RAG, Routing, Tool Calling)

Production-ready and beginner-friendly Python AI agent examples that map directly to guides on agentpatterns.tech.

Build and understand LLM agents with real execution loops: tool calling, memory, routing, orchestration, guardrails, and recovery.

Topics Covered

• Python AI agents
• LLM agent architecture
• ReAct, RAG, and routing agent patterns
• Tool calling and function calling
• Multi-agent orchestration and collaboration
• Guardrails, fallback, and recovery

Main Agent Patterns

Core section of this repo. If you are here for practical reusable architectures, start with these patterns.

Pattern	What you learn	Local path	Article
ReAct Agent	Thought-action-observation loops	agent-patterns/react-agent/python	Read article
Routing Agent	Intent-based task dispatch	agent-patterns/routing-agent/python	Read article
Task Decomposition Agent	Break work into executable sub-tasks	agent-patterns/task-decomposition-agent/python	Read article
RAG Agent	Retrieval-augmented reasoning	agent-patterns/rag-agent/python	Read article
Supervisor Agent	Controller-worker delegation	agent-patterns/supervisor-agent/python	Read article
Orchestrator Agent	Multi-worker execution orchestration	agent-patterns/orchestrator-agent/python	Read article
Research Agent	Structured web-style investigation flow	agent-patterns/research-agent/python	Read article
Data Analysis Agent	Analysis workflows with tool constraints	agent-patterns/data-analysis-agent/python	Read article
Reflection Agent	Self-review and iterative improvement	agent-patterns/reflection-agent/python	Read article
Self-Critique Agent	Audit and revise model output	agent-patterns/self-critique-agent/python	Read article
Memory-Augmented Agent	Short-term and durable memory usage	agent-patterns/memory-augmented-agent/python	Read article
Multi-Agent Collaboration	Role-based agent teamwork	agent-patterns/multi-agent-collaboration/python	Read article
Guarded Policy Agent	Policy checks and safe action gating	agent-patterns/guarded-policy-agent/python	Read article
Fallback Recovery Agent	Checkpoints, retries, and fallback logic	agent-patterns/fallback-recovery-agent/python	Read article
Code Execution Agent	Controlled code execution loops	agent-patterns/code-execution-agent/python	Read article

Quick Start

1. Choose an example.
2. Install dependencies.
3. Run the agent.

CATEGORY=agent-patterns
EXAMPLE=react-agent
cd $CATEGORY/$EXAMPLE/python

python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
export OPENAI_API_KEY="sk-..."
python main.py

Learning Path

• start-here for your first runnable agent.
• foundations for core building blocks.
• agent-patterns for production-style architectures.

Start Here

Example	Local path	Article
Write Your First Agent	start-here/write-your-first-agent/python	Read article

Foundations

Example	Local path	Article
Tool Calling Basics	foundations/tool-calling-basics/python	Read article
Tool Calling	foundations/tool-calling/python	Read article
Agent Memory	foundations/agent-memory/python	Read article
Allowed Actions	foundations/allowed-actions/python	Read article
Planning vs Reactive	foundations/planning-vs-reactive/python	Read article
Stop Conditions	foundations/stop-conditions/python	Read article
LLM Limits in Agents	foundations/llm-limits-agents/python	Read article

Extra Example

Example	Local path	Article
Support Agent	examples/support-agent/python	Code example

Why This Repository

• Real runnable AI agent code, not pseudo-code.
• Clear path from beginner examples to advanced patterns.
• Minimal production-style implementations you can reuse.
• Direct mapping to the architecture guides on agentpatterns.tech.

Prerequisites

• Python 3.10+ (3.11+ recommended)
• OPENAI_API_KEY environment variable
• macOS/Linux shell commands above (adapt for Windows if needed)

License

MIT