Main menu
Definition

Model Context Protocol (MCP)

Model Context Protocol (MCP) is an open protocol that lets AI applications connect to external tools and data sources in a standard way. Learn how MCP clients and servers work, and what to watch for when an AI model can act on your systems.

DTDevolutions Team·November 12, 2026·3 min read

What is Model Context Protocol?

Model Context Protocol (MCP) is an open standard, introduced by Anthropic in 2024, that defines how AI applications discover and call external tools, read data, and use prompt templates. Instead of building a custom integration for every model and every tool, a developer exposes capabilities once through an MCP server, and any compatible MCP client can use them.

How does MCP work?

  1. An AI application, the host, starts or connects to one or more MCP servers through its MCP client.
  2. Each server advertises the tools, resources, and prompts it offers.
  3. When a user asks for something, the model chooses which tool to call and with which arguments.
  4. The server runs the action with its own credentials and permissions, and returns the result to the model.

Servers can run locally over standard input and output, or remotely over HTTP.

MCP security considerations

An MCP server acts with whatever access it is given, so it is effectively a non-human identity. Common risks include over-broad tool permissions, secrets returned to the model, prompt injection through untrusted content, and unverified third-party servers. Limit each server to the tools a task needs, prefer read-only modes, keep secret values out of the model's context, and log tool calls.

Remote Desktop Manager includes an experimental MCP server. Read the setup and the security disclaimer in our documentation: MCP server in Remote Desktop Manager. Devolutions CLI can also run as an MCP server: Connect an AI client with MCP.

Why does MCP matter?

MCP is quickly becoming the common way AI assistants and agents reach real systems. For IT teams, that means AI tools can now read vaults, open sessions, or change configuration. Treating MCP servers like any other privileged integration, with scoped access and audit, keeps that convenience from becoming a new blind spot.

Frequently asked questions

What is Model Context Protocol?

Model Context Protocol (MCP) is an open standard that lets AI applications connect to external tools and data sources through MCP servers, so one integration can work with many compatible AI clients.

What is an MCP server?

An MCP server is a program that exposes tools, resources, or prompts to AI applications using the Model Context Protocol. The AI client calls the server, and the server performs the action with its own permissions.

Is MCP secure?

MCP itself is a protocol; security depends on how servers are built and configured. Limit tool permissions, avoid returning secrets to the model, vet third-party servers, and keep a log of tool calls.

Bring AI into remote connection management, carefully

Manage connections, credentials, and access from one platform built for IT teams.

Explore Remote Desktop Manager

Related terms

Non-human identity (NHI)

A digital identity used by software rather than a person, such as a service account, API key, workload, bot, or AI agent.

Read now →

Secrets management

The practice of securely storing, distributing, rotating, and auditing machine credentials such as API keys, tokens, certificates, and database passwords.

Read now →

Least privilege

Granting users only the access needed to carry out their responsibilities.

Read now →