Skip to main content

Explore Problems

Pricing Discuss

AI Agents & MCP

Course Contents

What Makes Something an "Agent"?

The Agent Loop: Thought → Action → Observation

Building a ReAct Agent From Scratch (No Framework)

Reactive vs Deliberative Agents

Workflow vs Agent: When Do You Actually Need One?

Anatomy of Agent Failure

Workflows vs Agents: Start Simple

Prompt Chaining

Parallelization

Orchestrator-Worker

Evaluator-Optimizer

Tools: Giving Agents Hands

Designing Robust Tool Interfaces

Parallel & Sequential Tool Calls

Code Execution & Sandboxing

Handling Tool Errors & Retries

Hands-On: Web Search + DB + Calculator

Task Decomposition & Planning

Plan-and-Execute vs ReAct

Self-Reflection & Self-Critique

Tree-of-Thought & Search-Based Reasoning

Replanning & Recovering From Failure

Why Agents Need Memory

Short-Term vs Long-Term Memory

Episodic, Semantic & Procedural Memory

Memory Stores & Retrieval (Vector + Summary)

Hands-On: A Memory-Aware Agent

The Integration Problem MCP Solves

MCP Architecture: Hosts, Clients, Servers

Tools, Resources & Prompts in MCP

Hands-On: Build Your First MCP Server

Connecting an Agent to MCP Servers

The MCP Ecosystem & Security Considerations

Why Multiple Agents? (And Why Not)

Supervisor / Hierarchical Architectures

Agent Handoffs & Shared State

The A2A (Agent-to-Agent) Protocol

Coordination & Cost

Designing a Multi-Agent System

The Framework Landscape (LangGraph, CrewAI, smolagents, OpenAI Agents SDK, LlamaIndex)

LangGraph: Graph-Based Agent Control

CrewAI: Role-Based Multi-Agent Teams

smolagents & Code Agents

OpenAI Agents SDK

Same Agent, Two Frameworks — When to Use What

Loading...

Loading content...