>

Learn MCP and Agent Infrastructure

Model Context Protocol, agentic frameworks and the plumbing behind AI agents — how tools connect to models, and the security that connection demands.

Understand MCP and why standardised tool access matters
Choose between agentic frameworks for a given problem
Design agent systems that fail safely rather than expensively
Take prompt injection seriously when agents have tool access
Free foreverNo signup requiredProgress saved locally15 lessons

Three tracks — pick your level. Progress is saved automatically in your browser.

BeginnerStart here. No prior knowledge assumed.
0 of 5 complete
Lesson 1
What is Agentic AI?
Agents versus assistants, and what changes when a model can act.
Lesson 2
What is MCP?
Servers, clients and hosts, and the integration problem the protocol solves.
Lesson 3
RAG
Giving agents grounded knowledge rather than relying on training data.
Lesson 4
Embeddings and Vector Databases
The retrieval layer most agent systems depend on.
Lesson 5
Prompt Injection and AI Security
Why tool access turns a content problem into a security problem.

What problem does MCP solve?

Why does tool access raise the security stakes?

IntermediateBuild on the basics.
0 of 5 complete
Lesson 1
LangChain
The framework, its abstractions, and when they help versus obscure.
Lesson 2
LlamaIndex
Retrieval-focused framework for document-heavy agent systems.
Lesson 3
CrewAI
Multi-agent orchestration and where role-based agents fit.
Lesson 4
AutoGen
Conversational multi-agent patterns.
Lesson 5
Manus
A general autonomous agent in practice, and verifying its output.

What is the distinction between MCP tools and resources?

When do framework abstractions hurt?

AdvancedDepth, edge cases and strategy.
0 of 5 complete
Lesson 1
Reasoning Models
When extra inference compute improves agent reliability.
Lesson 2
AI Evals and Benchmarks
Evaluating agent systems, where failure is multi-step and compounding.
Lesson 3
Token Economics
Agentic loops consume tokens fast — controlling the cost.
Lesson 4
Devin
Autonomous coding agents, and the honest gap between demos and completion rates.
Lesson 5
AI Regulation
Governance obligations when an AI system takes consequential actions.

What is the most reliable prompt injection mitigation?

Why do agent failures compound?

Ask an AI about this page

Opens your assistant with this page as the source, and a question rather than a summary. It will ask what you are building before it answers.

ChatGPTClaudeGeminiPerplexityGrok

Nothing is sent from here. The link carries only this page’s title and address.