MIDAS MCP Server Add-on: Prerequisites

Before you begin

The MIDAS MCP Server addon can be used with any AI assistant, agent, or platform that supports the Model Context Protocol (MCP). Support for MCP varies between platforms, and some require additional software to be installed on your computer before they can connect to a remote MCP server such as MIDAS.

This page summarizes what is currently required for the most popular AI platforms. Since MCP support is a fast-moving area, always check the AI provider's own documentation for the most up to date instructions, and see our Installation page for MIDAS-specific configuration.

What you'll need from MIDAS

• MIDAS v4.43+.
• The MIDAS MCP Server addon installed and enabled (see Installation).
• An MCP key - This is provided when you initially purchase the MCP Server addon, and may be rotated any time from your MIDAS MCP Server settings screen.
• If you have a self-hosted MIDAS system, your system must be accessible over HTTPS, since the MCP server requires a secure connection (All cloud-hosted MIDAS systems are accessible over HTTPS).

AI platform support

Claude (Anthropic)

Claude Desktop and Claude web/mobile apps support connecting to remote MCP servers such as MIDAS.

• Claude Desktop connects to remote MCP servers using a small bridging tool called mcp-remote, which requires Node.js to be installed on your computer. Node.js is free, and available for Windows, macOS, and Linux from nodejs.org.
• Claude.ai (web) and Claude for Work customers can add remote MCP servers directly as custom connectors, without needing Node.js.

ChatGPT (OpenAI)

ChatGPT supports connecting to remote MCP servers through its Connectors feature. This currently requires a ChatGPT Plus, Pro, Team, Enterprise, or Edu account, with Developer Mode enabled in your account settings. No additional software is required on your computer, as the connection is configured directly within ChatGPT.

Google Gemini

The Gemini CLI (a free command-line tool from Google) supports connecting to remote MCP servers directly. Gemini CLI requires Node.js to be installed. Support for remote MCP servers within the standard Gemini web/app interface is still developing, so check Google's current documentation if you plan to use Gemini outside of the CLI.

Other platforms

Many other AI platforms and developer tools support MCP, including a number of AI-powered code editors and agent platforms. Support and setup steps vary between them. If your preferred platform is not listed above, check its documentation for "MCP", "custom connector", or "remote tools" support. Provided it supports connecting to a remote MCP server over HTTPS with a custom authentication header, it should be able to connect to MIDAS in the same way as the platforms above.

What's next?

Once you have confirmed your AI platform's requirements above, head to the Installation page for step-by-step configuration instructions for each platform.