The AI Handbook Open in the app →

Model Context Protocol

Agents & Tools · Intermediate · 4 min read

What is it?

MCP is an open standard for connecting AI assistants to external tools and data sources through a common interface.

Explain like I'm 5

MCP is like a universal power outlet for AI: instead of a different custom plug for every tool, everything connects the same standard way.

Why was it created?

Connecting each AI app to each tool with custom code doesn't scale. MCP was created to standardize how models access tools and data.

Where is it used?

  • Connecting assistants to tools
  • Exposing company data to AI
  • Reusable tool servers
  • Agent integrations

Why should developers care?

MCP is a growing standard for AI integrations, so developers building tool-using assistants increasingly encounter it.

How does it work?

An MCP server exposes tools, data, or prompts in a standard format. An AI client (the assistant) connects to it and can discover and use those capabilities without custom, one-off integration code.

Real-world example

A team runs an MCP server exposing their database and ticket system; any MCP-compatible assistant can then query both through one standard interface.

Common use cases

  • Standardized tool access for AI
  • Sharing data sources with assistants
  • Reusable integrations across apps
  • Building agent capabilities

Advantages

  • One standard instead of custom glue
  • Reusable across AI clients
  • Decouples tools from specific apps
  • Open and interoperable

Disadvantages

  • Still maturing
  • Requires running/maintaining servers
  • Security must be handled carefully
  • Ecosystem still growing

When should you use it?

When you want to expose tools or data to AI assistants in a reusable, standard way.

When should you avoid it?

For a single quick, one-off integration where a standard adds unnecessary overhead.

Alternatives

Custom per-app integrationsDirect function-calling wiring

Related terms

Function CallingAI AgentAgentic AILarge Language Model

Interview questions

Beginner

  • What does MCP stand for?
  • What problem does it solve?

Intermediate

  • What is an MCP server versus client?
  • How does MCP relate to function calling?

Senior

  • What security concerns come with exposing tools via MCP?
  • Why standardize tool access instead of custom integrations?

Common misconceptions

  • "MCP is a model or an AI" — it's a protocol/standard for connecting models to tools, not a model itself.
  • "MCP replaces function calling" — it standardizes how tools are exposed; the model still calls them.

Fun facts

  • MCP stands for Model Context Protocol.
  • It aims to be a shared, open standard so the same tool server works with many AI clients.

Timeline

  • 2024 — MCP introduced as an open standard by Anthropic

Learning resources

Quick summary

MCP is an open standard that connects AI assistants to tools and data through a common interface, replacing custom one-off integrations.

Cheat sheet

  • Open standard for AI tool/data access
  • Servers expose tools; clients use them
  • Reusable across assistants
  • Complements function calling

If you remember only one thing

MCP is a universal standard for plugging AI assistants into tools and data, so integrations are reusable instead of custom.