Agentic-Design-Patterns

https://assets.jimmysong.io/books/agentic-design-patterns-zh-20251011.pdf

https://agent-patterns.readthedocs.io/en/latest/index.html

Agent-Patterns?

A Python library providing reusable, extensible, and well-documented base classes for common AI agent workflows using LangGraph and LangChain.

Overview?

Agent-Patterns implements proven design patterns for AI agents, reducing boilerplate and encouraging consistent best practices. The library is especially useful for developers needing to quickly build or customize advanced AI workflows without reinventing the wheel.

Guides?

• Getting Started Guide: Installation and your first agent

• Pattern Selection Guide: Choosing the right pattern for your use case

• Advanced Customization Guide: Extending the library

• Troubleshooting Guide: Solving common issues

• Deployment Guide: Deploying in production environments

• Migration Guide: Migrating from other frameworks

See all guides in the Guides Index.

Tutorials?

• Research Assistant: Building an AI research assistant

• Customer Support Bot: Creating a support agent

• Code Generation Agent: Developing a code generator

• Multi-Agent System: Building a collaborative system

See all tutorials in the Tutorials Index.

Supported Patterns?

• ReAct (Reason + Act): Iterative reasoning and action for tool-based problem solving (arXiv Paper)

• Plan & Solve: Decoupled planning and execution phases (arXiv Paper)

• Reflection: Periodic reflection for strategic adjustments

• Reflexion: ReAct with reflection capabilities for self-improvement (arXiv Paper)

• LLM Compiler: Dynamic execution graph construction and optimization (arXiv Paper)

• ReWOO: Reason, World model, Observe, Outcome for simulation (arXiv Paper)

• LATS: LangChain Agents Tracing System for comprehensive observability (arXiv Paper)

• STORM: Self-evaluation, Think of options, Options for reasoning, Reason step by step, Mistake detection (NAACL Paper)

• Self-Discovery: Agents that discover their own capabilities

• Reflection and Refinement: Structured reflection with explicit refinement steps

See all patterns in the Patterns Index.

https://github.com/fanqingsong/Agentic-Design-Patterns

Agentic Design Patterns - A Hands-On Guide to Building Intelligent Systems

This repository contains the complete materials for "Agentic Design Patterns: A Hands-On Guide to Building Intelligent Systems" by Antonio Gulli. It includes all chapters in PDF format and accompanying code notebooks for hands-on learning.

Note: All author royalties are donated to Save the Children ??

This comprehensive guide covers 21 chapters and 7 appendices on building intelligent AI agent systems, including:

• Foundational Patterns: Prompt chaining, routing, parallelization
• Advanced Techniques: Reflection, tool use, planning, multi-agent systems
• Memory & Learning: Memory management, adaptation, goal setting
• Production Patterns: Exception handling, human-in-the-loop, RAG
• Optimization: Resource-aware patterns, reasoning techniques, guardrails
• Real-world Applications: From GUI to real-world environments

1. Chapter 1: Prompt Chaining - Sequential task decomposition
2. Chapter 2: Routing - Dynamic path selection
3. Chapter 3: Parallelization - Concurrent processing
4. Chapter 4: Reflection - Self-improvement mechanisms
5. Chapter 5: Tool Use - External capability integration
6. Chapter 6: Planning - Strategic task management
7. Chapter 7: Multi-Agent - Collaborative systems

8. Chapter 8: Memory Management - State persistence
9. Chapter 9: Learning and Adaptation - Dynamic improvement
10. Chapter 10: Model Context Protocol (MCP) - Standardized interfaces
11. Chapter 11: Goal Setting and Monitoring - Objective tracking

12. Chapter 12: Exception Handling and Recovery - Robust error management
13. Chapter 13: Human-in-the-Loop - Human-AI collaboration
14. Chapter 14: Knowledge Retrieval (RAG) - Information access patterns

15. Chapter 15: Inter-Agent Communication (A2A) - Agent networking
16. Chapter 16: Resource-Aware Optimization - Efficient resource usage
17. Chapter 17: Reasoning Techniques - Advanced decision-making
18. Chapter 18: Guardrails/Safety Patterns - Risk mitigation
19. Chapter 19: Evaluation and Monitoring - Performance tracking
20. Chapter 20: Prioritization - Task management
21. Chapter 21: Exploration and Discovery - Autonomous learning

• Appendix A: Advanced Prompting Techniques
• Appendix B: AI Agentic: From GUI to Real world environment
• Appendix C: Quick overview of Agentic Frameworks
• Appendix D: Building an Agent with AgentSpace
• Appendix E: AI Agents on the CLI
• Appendix F: Under the Hood: Reasoning Engines
• Appendix G: Coding agents