What is MCP?

MCP is a newly established standardized protocol designed for communication between LLMs and external APIs or data stores.

Who created it ? : Anthropic

Technical Overview

An SDK is available for various programming languages, enabling the implementation of an MCP server in the preferred language. This allows LLMs to perform function calls seamlessly.

Explain in simple way :

A middleware solution acts as a bridge between LLMs and different data sources like APIs, SDKs, and databases. This middle layer helps LLMs talk to these data sources, allowing them to get data and make changes or perform actions based on specific requests.

Git Repo :  https://github.com/modelcontextprotocol

Components

MCP Server: This server provides an MCP endpoint accessible to a Large Language Model (LLM) in a manner akin to a REST API or GraphQL server.

MCP Client: This refers to a client capable of directly accessing the MCP server.

How it works :

This approach separates data from the LLM, allowing it to process user queries, convert them into function calls, and execute actions in a standardized way. The LLM retrieves data through REST APIs or database queries and receives responses in a consistent format.

Consider the following examples:

Example 1: Calendar Integration

A calendar application, like Google Calendar, provides APIs for creating, updating, and canceling invites. Since the LLM does not inherently understand these APIs, an MCP server acts as a wrapper. It defines tools for invite creation, updating, and cancellation. When the LLM receives a related query, it calls the MCP server, which executes the appropriate function, interacts with the calendar API, and returns the response.

Example 2: SQL Database Integration

In a database storing user names and email addresses, an MCP server can expose functions for adding users and retrieving emails. The LLM's query is translated into an SQL command, executed on the database, and the response is returned to the LLM.

Who is using : https://opentools.com/registry

Advantages

• Standardization – Provides a unified approach for LLM tools and data exposure.

• Faster Implementation – Reduces the time needed to build LLM-based workflows.

• Enhanced Dynamism – Enables more dynamic and agentic LLM applications through external integrations.

Use Cases

• Allow users to connect their preferred LLM to your tool’s data via MCP.

• Integrate products with multiple tools to streamline workflows, provided MCP is supported.

Maturity : Early

Summary

MCP is a promising concept that acts as a standardized wrapper for APIs, CLIs, and SDKs. It ensures a unified authorization layer for LLM tools. While still in its early stages, it is gaining adoption and has strong potential.