AI Agents Guide — AI skill for Claude Code

AI Agents Guide

A practical guide to understanding, building, and working with AI agents.

Overview

AI agents are systems that use models to interpret goals, decide what to do, use available tools, and produce results. This repository collects the core concepts and practical patterns needed to understand how agent systems work.

The focus is on clear explanations and small, understandable examples rather than unnecessary framework complexity.

What this guide covers

Agent fundamentals

Understand the basic components of an agent system:

• Model reasoning
• Instructions and context
• Tool use
• Memory
• Planning
• State and execution
• Agent and environment interaction

Agent workflow

A typical agent follows a loop similar to:

Goal
  ↓
Understand the task
  ↓
Decide the next action
  ↓
Use a tool when needed
  ↓
Observe the result
  ↓
Continue or finish

The exact workflow depends on the system. Not every agent needs every component.

Tools

Tools allow an agent to interact with systems outside the model itself. Examples include:

• Web search
• File operations
• APIs
• Databases
• Code execution
• External services

A useful agent should use tools only when they provide information or actions that the model cannot reliably provide on its own.

Memory

Memory allows an agent to retain information beyond a single model call. This can include conversation history, task state, retrieved documents, or persistent user preferences.

Memory should have a clear purpose. Storing everything is not the same as having useful memory.

Retrieval and context

Agents often need information that is not contained in the model context. Retrieval systems can locate relevant documents or data and provide them to the model when needed.

The guide covers the relationship between retrieval, context, tool use, and agent decisions.

Planning and execution

Some tasks require multiple actions. Planning helps an agent break a larger goal into sm