---
title: "MCP server"
description: "An MCP server exposes tools, data and prompts to AI applications through the Model Context Protocol (MCP), an open standard, so an AI agent or assistant can discover and call those capabilities, for example sending a message, without custom integration code."
canonical: "https://omnichannels.ai/glossary/mcp-server/"
language: "en"
updated: "2026-10-10"
publisher: "BroadNet Technologies"
brand: "OmniChannels.ai"
---

# MCP server

Canonical: https://omnichannels.ai/glossary/mcp-server/
Language: en
Publisher: OmniChannels.ai by BroadNet Technologies

**Definition:** An MCP server exposes tools, data and prompts to AI applications through the Model Context Protocol (MCP), an open standard, so an AI agent or assistant can discover and call those capabilities, for example sending a message, without custom integration code.

MCP defines how an AI host (such as a desktop assistant, an IDE or an agent framework) connects to servers. Each server describes its tools with names, descriptions and input schemas; the model decides when to call them, and the host executes the calls and returns results.

For messaging, an MCP server lets an agent do things like send an SMS or WhatsApp template, look up delivery status or list templates in response to natural-language instructions, using the same account permissions as the regular API.

Security matters: authenticate the connection, give the agent the least privilege it needs, require confirmation for actions with cost or customer impact (such as bulk sends), and log every tool call. Treat content returned to the model as untrusted input.

MCP complements, rather than replaces, a REST API: the REST API serves conventional applications; the MCP server packages the same capabilities for AI agents.

OmniChannels.ai offers an MCP server for AI agents at https://mcp.broadnet.me/mcp.

## Sources

- [Model Context Protocol specification](https://modelcontextprotocol.io/specification/latest)