What Is an MCP Server (And What Actually Happens Behind the Scenes)

So far, we’ve covered:

• why MCP exists
• what MCP is
• what tools are

Now let’s answer a key question:

When the model decides to use a tool… who actually runs it?

---

🧠 Simple Definition

An MCP server is:

The component that exposes tools and executes them.

---

⚠️ Important Clarification

An MCP server is not just your backend.

It is:

• a layer on top of your backend
• designed specifically for LLM interaction

---

🧩 What the MCP Server Does

It has three main responsibilities:

---

1. Expose Tools

It tells the system:

• what tools exist
• what they do
• what inputs they need

This is what the model “sees”.

---

2. Validate Inputs

Before running anything, it checks:

• are required fields present?
• are types correct?

This prevents bad or unsafe execution.

---

3. Execute Logic

This is where real work happens:

• database queries
• API calls
• business logic

---

🔄 What Happens During Execution

Let’s walk through a simple flow.

User asks:

“Show my last 3 orders”

---

Step 1 — Model decides

It generates:

{
  "tool": "get_user_orders",
  "arguments": {
    "user_id": "123",
    "limit": 3
  }
}

---

Step 2 — Request reaches the server

The MCP server receives this request.

---

Step 3 — Validation

It checks:

• is user_id present?
• is limit a number?

---

Step 4 — Execution

It runs something like:

SELECT * FROM orders WHERE user_id = 123 LIMIT 3

---

Step 5 — Response

It sends structured data back.

---

Step 6 — Model formats response

The model turns it into:

“Here are your last 3 orders…”

---

🧠 Key Insight

The model never:

• touches your database
• calls APIs directly
• runs any code

👉 The MCP server does all of that.

---

🔥 Why This Layer Matters

This separation gives you:

• better security
• cleaner architecture
• controlled execution
• reusable systems

---

🧭 Mental Model

Think of it like a restaurant:

• Model → decides what to order
• MCP server → kitchen that prepares it
• Tools → items on the menu

The model doesn’t cook.
The server does.

---

⚠️ Common Mistakes

---

Treating MCP server like a normal backend

It needs to be LLM-friendly, not just functional.

---

Exposing raw APIs directly

LLMs need:

• clear names
• structured inputs
• simple actions

---

Skipping validation

This can lead to:

• crashes
• incorrect actions
• security issues

---

🧠 What the MCP Server Does NOT Do

It does NOT:

• decide which tool to use
• understand user intent
• generate responses

👉 That’s the model’s job.

---

🧭 Why This Is Important

This separation is the foundation of MCP:

• model decides
• server executes

---

🧭 What’s Next

Now that we understand the execution layer,
there’s one missing piece:

Who connects the model and the server?

That’s the MCP client — the component that makes everything work together.