🌟 Awesome Agentic System Design

A curated collection of resources for designing, evaluating, and deploying agentic AI systems—covering architecture patterns, frameworks, safety considerations, and real-world applications.

---

📝 Motivation

AI agents are quickly moving from research prototypes to real-world applications, and with that shift comes a new challenge: agentic system design.

Just as machine learning system design became a must-have skill for engineers and researchers, agentic system design is poised to be the next big area of demand. Companies are already looking for people who can reason about how agents should plan, act, use tools, and collaborate in complex environments.

This repository helps you stay ahead of the curve—bringing together the most useful resources, frameworks, and design patterns to sharpen your skills for future agentic system design interviews.

---

📚 Resources

🔹 Industry Guides & Whitepapers

• IBM x Anthropic — Architecting Secure Enterprise AI Agents with MCP
  Enterprise-focused blueprint for building secure AI agents using the Model Context Protocol (MCP).

• Google — AI Whitepaper: Agents
  Google’s vision on agent capabilities and their role in next-generation AI systems.

• Antonio Gulli (Google) - Agentic Design Patterns: A Hands-On Guide to Building Intelligent Systems

• Databricks

  • Agent System Design Patterns: Patterns for designing, orchestrating, and scaling agents in enterprise data/AI workflows.
  • The Big Book of GenAI: 5 ways to leverage your data to build production-quality GenAl applications.

• OpenAI — A Practical Guide to Building Agents
  Hands-on playbook for constructing agent workflows, planning strategies, and evaluation.

• Anthropic — Building effective agents
  Practical advice for developers on building effective agents

• Vectorize — Designing Agentic AI Systems (Series)
  Multi-part blog series introducing agent architectures, design considerations, and real-world use cases.

• MongoDB — 7 Practical Design Patterns for Agentic Systems
  Seven practical design patterns for agentic systems that have implemented in real-world use cases.

📚 Research Papers

Surveys & Architecture

• Agentic Large Language Models, a Survey
• Large Language Model Agent: A Survey on Methodology, Applications and Challenges
• A Survey of Self-Evolving Agents
• Agentic Retrieval-Augmented Generation: A Survey on Agentic RAG
• Multi-Agent Collaboration Mechanisms: A Survey of LLMs
• A Survey of Agent Interoperability Protocols: MCP, ACP, A2A, and ANP
• Agentic AI Frameworks: Architectures, Protocols, and Design Challenges
• Agentic Design Patterns: A System-Theoretic Framework
• Architectures for Building Agentic AI

Agent Design & Optimization

• CORAL: Towards Autonomous Multi-Agent Evolution for Open-Ended Discovery
• Weak-for-Strong: Training Weak Meta-Agent to Harness Strong Executors
• MetaAgent: Automatically Constructing Multi-Agent Systems Based on Finite State Machines
• AFlow: Automating Agentic Workflow Generation
• FlowReasoner: Reinforcing Query-Level Meta-Agents
• Multi-Agent Design: Optimizing Agents with Better Prompts and Topologies
• Towards a Science of Scaling Agent Systems
• On the Role of Feedback in Test-Time Scaling of Agentic AI Workflows
• RewardHarness: Self-Evolving Agentic Post-Training
• Budget-Aware Tool-Use Enables Effective Agent Scaling
• A Practical Guide for Designing, Developing, and Deploying Production-Grade Agentic AI Workflows

Agent Memory & Evaluation

• A-MEM: Agentic Memory for LLM Agents
• Survey on Evaluation of LLM-based Agents
• MultiAgentBench: Evaluating the Collaboration and Competition of LLM Agents
• PerspectiveGap: A Benchmark for Multi-Agent Orchestration Prompting
• ClawBench: Can AI Agents Complete Everyday Online Tasks?

Agent Safety & Interpretability

• Because we have LLMs, we Can and Should Pursue Agentic Interpretability
• AgenTracer: Who Is Inducing Failure in the LLM Agentic Systems?
• Toward a Safe Internet of Agents
• Towards a Science of AI Agent Reliability

🤖 Agentic Frameworks

• Accenture AI Refinery SDK
• AG2: Open-Source AgentOS for AI Agents
• Agent S: Use Computer Like a Human
• AgentScope: Agent-Oriented Programming for Building LLM Applications
• Agno: Lightweight Multi-Modal Agent Runtime
• AutoGen: A framework for building AI agents and applications
• Claude Agent SDK
• CrewAI: Fast and Flexible Multi-Agent Automation Framework
• Dify: Open-Source Platform for Agentic Workflow Development
• Google Agent Development Kit (ADK)
• LangGraph: Orchestration Framework for Stateful Agents
• Mastra: TypeScript-First Agent Framework
• MetaGPT: The Multi-Agent Framework
• Microsoft Agent Framework
• Microsoft 365 Agents SDK
• OpenAI Agents SDK
• OpenAI Agent Builder
• Pydantic AI: Type-Safe Agent Framework
• SmolAgents: Code-Centric Agent Framework by Hugging Face
• Darkmoon - Open source (GPL-3.0) autonomous AI penetration testing platform covering web, API, Active Directory and Kubernetes, with proof of exploitation.
• Strands Agents: AWS Open-Source Agent SDK

🤝 Contributing

Have a resource, paper, or framework that belongs here?
Feel free to open a PR and help grow the ecosystem of agentic system design.

---