MCP server

MCP server: the explanation

The server receives messages from an MCP client and responds using the capabilities it provides. These can include executable tools, readable resources, and prompt templates. A tool might search for documents, while a resource supplies the content of an identified file. The server may perform operations itself or pass requests to another system. It does not need to contain a language model, because its role is to provide an interface rather than necessarily make decisions about your overall task.

This component makes a service available to different compatible applications without rebuilding the entire integration for each one. It can wrap an existing API, provide access to local files, or expose organization-specific functions. Here, server describes a software role rather than necessarily a machine in a data center. A process running on your computer can perform that role, just as a remotely hosted network service can. Hosting arrangements affect operation, but do not define the basic purpose.

The risks depend on accessible data, exposed functions, and the permissions held by the process. A local server may have extensive rights on your device, while a remote server may receive information needed to answer requests. A tool description alone does not establish that its implementation behaves correctly. Examine the server’s source, configuration, and write capabilities. Updates, activity records, and the ability to revoke access also matter when operating the connection over time rather than merely testing its initial setup.

A concrete example

You connect an assistant to an MCP server that searches documents in a working folder. The server exposes a search function and returns matching passages with their source information. You configure read access to the folder you need. The assistant can then prepare a summary without having a deletion function available. You still check which documents were consulted and whether the written summary accurately reflects the retrieved material.

Why it matters

Choosing an MCP server means trusting software with some of your information or actions. Check who maintains it, where it runs, and which permissions it uses. Distinguish reading operations from possible modifications. Also assess error handling, service availability, and ways to withdraw access when the task ends or your organization changes. These operational details matter even when connecting the server requires only a simple configuration step.

Common misconceptions

• An MCP server does not have to be a cloud service: it can run as a local process on your device, although some of its functions may still contact remote systems to obtain the information you request.
• An MCP server is not automatically an intelligent agent: it may simply expose deterministic functions or documents, while the application and its model decide how to use those capabilities when carrying out your requested task.

Related terms: mcp protocol

MCP (Model Context Protocol)

MCP, or Model Context Protocol, is a protocol for exchanges between an AI application and services that provide data or functions.

MCP client

An MCP client is the component of an application that communicates with a server using the Model Context Protocol.

API

An API, or application programming interface, defines how software can use another component’s functions or data.

SDK

SDK stands for software development kit.

webhook

A webhook is a mechanism that lets a service automatically send a notification to a web address when an event occurs.

structured output

Structured output is a response arranged according to a defined format, with expected fields and data types.

A2A protocol

The A2A protocol is an open set of rules that lets independent AI agents communicate and delegate tasks.

Where this site covers it

The term appears on 3,548 pages of this site. A few of them:

• Choosing an MCP server: trust, maintenance, permissions, and local or remote hosting Guide
• Securing an AI agent: permissions, secrets, sandboxing, human approval, and logging Guide
• Getting started with an AI agent: choosing tools, setup, first tasks, and safeguards Guide
• Connect an AI agent to Telegram, WhatsApp, Slack, or Discord: methods and risks Guide
• Claude Code reference guide: installation, subscriptions and API access, commands, CLAUDE.md, MCP, subagents, and best practices Guide
• Cursor reference guide: the AI code editor, modes, rules, pricing, and limits Guide
• Evaluating an AI agent: test sets, metrics, public benchmarks, and production monitoring Guide
• AI automation in Make and Zapier: features, agents, limits, and pricing Guide