AI Agents Super Reference

Learn about AI agents fast

secret agent

This article is a curated list of high quality resources on AI agents. Please comment with any great resources you’ve found.

An Agent is a system that leverages an AI model to interact with its environment in order to achieve a user-defined objective. It combines reasoning, planning, and the execution of actions (often via external tools) to fulfill tasks

Articles

Building effective agents, Anthropic

• https://www.anthropic.com/engineering/building-effective-agents

• Workflows: Orchestration through predefined code paths

• Agents: LLMs dynamically direct their own processes and tools, maintaining control on how they accomplish tasks

• The augmented LLM: A basic building block for agent is an LLM that has access to retrieval, tools and memory

Workflow

• Prompt chaining: Decompose a task into a sequence of steps

• Routing: Classify an input and direct it to a specialized follow up task

• Parallelization: Perform work at the same time and programmatically aggregate results. Sectioning involves dividing work, voting involves performing the task multiple ways to get diverse outputs

• Orchestrator worker: The LLM breaks down tasks and delegates them to worker LLMs

• Evaluator-optimizer: One LLM generates content and another improves the content

Practical agent usage: “agents add the most value for tasks that require both conversation and action, have clear success criteria, enable feedback loops, and integrate meaningful human oversight.”

Introduction to Agents, Hugging Face

• https://huggingface.co/docs/smolagents/en/conceptual_guides/intro_agents

• Different levels of agency can be given to LLMs to control a workflow

  • Simple processor: LLM has no impact on program flow

  • Router: LLM determines if/else switch

  • Tool caller: LLM output determines tool execution

  • Multi-step agent: LLM controls iteration and program continuation

  • Multi-agent: One agent workflow can start another workflow

Model Context Protocol, Core Architecture

• https://modelcontextprotocol.io/docs/concepts/architecture

• https://github.com/modelcontextprotocol

• Hosts: LLM applications that initiate connections

• Clients: Maintain 1:1 connections with servers, inside the host application

• Servers: Provide context, tools and prompts to clients

From the MCP introduction: “MCP is an open protocol that standardizes how applications provide context to LLMs. Think of MCP like a USB-C port for AI applications. Just as USB-C provides a standardized way to connect your devices to various peripherals and accessories, MCP provides a standardized way to connect AI models to different data sources and tools.”

Agent Architectures, LangGraph

• https://langchain-ai.github.io/langgraph/tutorials/#agent-architectures

• Multi-agent systems

  • Network: Enable two or more agents to collaborate on a task

  • Supervisor: Use an LLM to orchestrate and delegate to individual agents

  • Hierarchical teams: Orchestrate nested teams of agents to solve problems

• Planning agents

  • Plan-and-execute: Implement a basic planning and execution agent

  • Planning with observation: Reduce re-planning by saving observations as variables

• Reflection and critique

  • Basic reflection: Prompt the agent to reflect on and revise its outputs

• Experimental

  • Web research (STORM): Generate Wikipedia-like articles via research and multi-perspective QA

Videos

Papers

• ReAct: Synergizing Reasoning and Acting in Language Models

  • Abstract: https://arxiv.org/abs/2210.03629

  • Code: https://github.com/ysymyth/ReAct

• Executable Code Actions Elicit Better LLM Agents

  • Abstract: https://arxiv.org/abs/2402.01030

  • Code: https://github.com/xingyaoww/code-act

• AI Agents That Matter

  • Abstract: https://arxiv.org/abs/2407.01502

• AI Agents: Evolution, Architecture, and Real-World Applications

  • Abstract: https://arxiv.org/abs/2503.12687

• The Rise and Potential of Large Language Model Based Agents: A Survey

  • Abstract: https://arxiv.org/abs/2309.07864

• Additional papers on AI Agents

  • https://huggingface.co/collections/m-ric/agents-65ba776fbd9e29f771c07d4e

Courses

• Hugging Face, AI Agents Course (free) https://huggingface.co/learn/agents-course/en/unit0/introduction

• Microsoft, AI Agents for Beginners (free) https://github.com/microsoft/ai-agents-for-beginners

• DeepLearning.AI, AI Agents in LangGraph (free) https://www.deeplearning.ai/short-courses/ai-agents-in-langgraph/

Frameworks and Tools

• Hugging Face SmolAgents https://huggingface.co/docs/smolagents/index

• LangGraph https://langchain-ai.github.io/langgraph/

• Llama Index https://docs.llamaindex.ai/en/stable/

• Microsoft Semantic Kernel https://learn.microsoft.com/en-us/semantic-kernel/overview/

• N8N https://github.com/n8n-io/n8n

• MCP https://github.com/modelcontextprotocol

Code Resources

• ML Reference Designs

  • Github repo maintained by the author showing various AI topics including AI agents and framework usage

  • https://github.com/ccozad/ml-reference-designs

• Microsoft Semantic Kernel Examples

  • C#, dotnet

  • https://github.com/microsoft/semantic-kernel/tree/main/dotnet/samples/Concepts

• LangGraph

  • https://github.com/langchain-ai/langgraph

• SmolAgents

  • https://github.com/huggingface/smolagents