MCP Servers

Learn more about Model Context Protocol and it's use cases.

What is MCP?

Model Context Protocol (MCP) is an open-source standard developed by Anthropic in November 2024. MCP defines a shared communication protocol between an AI client (such as a model or assistant) and one or more servers (which provide capabilities, data, or APIs).

MCP servers are backend services that extend AI models like ChatGPT, Claude, Cursor with custom tools, data, or functionality. The goal is to make it possible for AI models to call functions, retrieve data, and work with resources without custom integrations or private APIs. The AI model acts as an MCP client, and includes built-in support for the MCP client interface. When you connect your AI client to an MCP server (such as the MTA MCP Actions server) this allows the model to securely fetch information, run computations, or interact with an external API.

Learn more about Model Context Protocol

Click here to see a list of applications that support MCP integrations.

How is MCP Useful When Working with AI?

When building or extending AI applications, one of the key challenges is connecting the model’s reasoning layer with real-world data and services. LLMs train on data and context (usually human-readable text) but can’t easily gain knowledge from external services.

Before MCP, custom integrations to specific tools were created to give the AI access to external features, but these integrations were specific to that LLM and tool. So an open standard was created so that all AI tools could communicate with services in the same way.

Ways MCP enhances AI workflows

• Standardized Integration: Instead of building custom APIs or plugins for each data source or service, you can connect via MCP. Different AI systems (such as Claude, ChatGPT, or Cursor) can all interact with the same MCP server.

• Security and Context Control: MCP supports permissions and clear data boundaries, ensuring AI models can access only the intended resources.

• Dynamic Context Enrichment: AI models can request and use live data—from databases, code repositories, monitoring systems, or alerts—directly in context, improving output accuracy and reducing hallucination.

• Local and Remote Flexibility: MCP works both with locally installed servers (for secure or offline data) and remote-hosted servers (for distributed or shared services).

Types of MCP Servers

MCP servers can be deployed locally or remotely, depending on your use case, security needs, and environment.

Remote-Hosted Servers

These servers are hosted externally, usually as web services or API endpoints, and made accessible via standard MCP protocol connections.

Typical use cases

• Integrating with SaaS products (e.g., analytics dashboards, notification systems, CRM data).

• Providing organization-wide AI access to shared tools or datasets.

• Allows multiple AI clients to use the same MCP endpoint.

Benefits

• Centralized maintenance and version control.

• Easier sharing across teams or projects.

• Scalable infrastructure (e.g., using cloud hosting).

Example

Mobile Text Alerts MCP Actions server provides access to the Mobile Text Alerts API, so any AI model configured with it can send messages (and other actions) securely.

Local Package Servers

Local MCP servers are installed as packages or modules directly on a developer’s system or inside a controlled environment. They typically run alongside an AI client and expose capabilities locally.

Typical use cases

• Interacting with local files, codebases, or development tools.

• Testing or prototyping custom MCP capabilities.

• Keeping sensitive data fully offline.

Benefits

• Data never leaves the local machine.

• Low-latency performance for local context access.

• Easy to customize or extend for specialized workflows.

Example

A developer might install a local MCP package that lets an AI assistant read and refactor code in the current project directory or execute local scripts securely.

Mobile Text Alerts MCP Servers

Developer Center

This Developer Center GitBook site includes a Model Context Protocol (MCP) server. See Developer Center MCP Server to learn more.

Mobile Text Alerts Actions MCP Server

This server hosted by Mobile Text Alerts enables users using AI assistants to send SMS right from within their AI workflows. See MTA Actions MCP Server to learn more.

Summary

By standardizing communication between AI clients and external tools, MCP simplifies integration, strengthens security, and unlocks more capable and context-aware AI applications.

For full technical details and protocol specifications, visit modelcontextprotocol.io.