Lesson 2 of 30, Module 1: The Whole Idea. You understand what MCP is, why it exists, and you have FELT the problem it solves.
What you can do after this lesson
A protocol is worth building once precisely because it turns an M-times-N wiring problem (every tool rewired by hand into every app) into an M-plus-N one: wire in once, any app that speaks the protocol can use it. That is the same shift a universal remote made for a room full of single-purpose remotes.
Downloads
- Module 1 checkpoint (scribe-checkpoint-m1.zip): working code as the module leaves it, so a broken session costs you nothing
- The course GitHub repo: every checkpoint, the README and the full lesson list
Prefer to read?
The written version of this part of the course is on Substack: https://genaiunplugged.substack.com/p/what-is-mcp-model-context-protocol.
Lesson transcript
Three letters nobody can expand
Almost everybody has heard the letters MCP by now. Very few people can tell you what they stand for.
Hello, and welcome to lesson 2 of the MCP Masterclass. Last lesson I showed you what we are going to build. Today we find out what this thing actually is. And we start with our map, as always.
We are still at 6 lit boxes on the diagram. You, 2 apps, the client inside them, the pipe, and your Google Drive folder. The middle is still empty.
The four remotes on your coffee table
Picture your living room about 15 years ago. The TV had a remote. The music system had another remote. The set top box had a third one, and the air cooler had a fourth.
Every remote had different buttons in different places. Every remote needed its own batteries. When a new gadget arrived, a new remote arrived with it.
Then somebody sold you a universal remote. One handset, and it drove all of them.
So here is the one line I want you to remember from this lesson.
MCP is nothing but a universal remote for AI.
That is the whole idea. Hold onto it, because we come back to that living room a few times in this course.
Model, Context, Protocol, one word at a time
Now let us take the name apart.
MCP stands for Model Context Protocol. 3 words, and each one is doing a job.
Model is the AI itself. Claude Opus 5 is a model, and so is Claude Sonnet 5, and so is GPT.
Context is everything the model needs to be useful to you, sitting outside its trained knowledge. Your documents, your folders, your accounts, your notes.
I call this external intelligence. And I will keep calling it that through the whole course, because it is a better name than context.
A model on its own knows a lot about the world and nothing about you. It has never seen your Google Drive data. External intelligence closes that gap.
Protocol is nothing but an agreement about how 2 programs talk to each other. Somebody writes the agreement down, and everybody follows the same one.
Put those 3 words back together. Model Context Protocol is an agreed way to hand your stuff to an AI, so it gets external intelligence.
Why an agreement matters this much
But you must be thinking, Dheeraj, why does an agreement matter so much?
Think about the living room again, and think about who was in pain.
The gadget makers were in pain, because each of them had to design a remote from nothing. The customer was in pain, because the coffee table had 6 remotes on it.
Software had the same problem. Somebody who built a useful tool had to wire it into Claude by hand. Then they wired it into a second app like ChatGPT by hand. Then a third app like Cursor again. Then a fourth.
Every link was written by hand, and every one of them was different.
MCP is the agreement that ended that. You wire your tool into the protocol one time. After that, any AI powered app that speaks the protocol can use it.
What you are actually going to build
So what does this look like for us?
Well, you are going to write a small program called a server. Your server offers Claude, or any AI, a short list of jobs it is allowed to do.
Claude reads that list, picks what it needs, and asks your server to do the work. Your server does the work and hands the answer back.
That is the whole loop. A list, a request, an answer.
Who owns MCP
However, who owns this thing?
This is a fair question, and it deserves a straight answer, because you are about to build on top of it.
Anthropic made MCP and gave it away in November, twenty twenty four. In December, twenty twenty five, they handed it to the Agentic AI Foundation, which sits under the Linux Foundation.
So no single company can quietly take it away from you. The rules are public, and anybody can write an MCP server or an AI app that speaks them.
Now, the official docs call MCP a USB C port for AI. You will see that line repeated all over the internet.
It is a good line, and it points at the same idea as the universal remote I explained a minute ago. One shape of plug, and everything fits it.
We stay with the remote in this course, because a remote does something. It sends commands, and that is closer to what your server will do.
What we skipped, and where the map stands
And there are things we did not cover today.
We have not looked inside MCP yet. There is a host, a client and a server, and they each have a separate job.
There are also 3 kinds of thing a server can offer. Those turn up in lesson 8.
For now, one sentence is enough. MCP is nothing but a universal remote for AI.
So where does our map stand at the end of this lesson?
Nothing new lit up. Still 6 boxes, and still a hole in the middle.
That hole is there on purpose, and it stays open for all of module 1. By the time we fill it, you will badly want it filled.
Next lesson is all about getting your computer ready, so that everything after this really runs.
Bye now, and I will see you in the next lesson.
