MCP Support

The Model Context Protocol (MCP) has quickly become a standard in AI, acting like a universal "USB-C port" that connects AI models to tools, data sources, and services. MCP eliminates the need for fragmented, custom integrations by providing a standardized way for AI to interact with the digital world.

What makes Memex different is that you can both build and use MCP servers. Memex can even code an MCP server, then add it to its own context, test it, and iteratively debug and improve it — all autonomously.

Using MCPs

MCP is like an infinitely expanding toolset for Memex. You can customize the tools it has access to for a tailored experience for your use case. You can discover and manage available MCP tools under the ⚙️ Settings → MCP tab.

• Under Configured Servers you can see your available MCPs, turn them on/off, and manage their built-in tools.

• Under Server Directory you can find one-click installable MCP servers, including Neon, Netlify, GitHub, Context7, and more. Over time we’ll be curating this to servers that are tested with Memex and help with the most common tasks. But you can always add your own servers — and even ask Memex to install them for you!

• Under Add Customer Server you can install any MCP server.

Memex works with MCP servers in an intuitive way – often requiring no special instructions. Simply enable the MCP server, and Memex understands how to use it. For more specific applications, a gentle nudge like "use playwright to check this website" or "use context7 to look up the latest docs" is all it takes.

In the below example, Memex uses three MCP servers—Headless Terminal, TaskMaster, and Netlify— to create a landing page and deploy it all with a single prompt.

Building MCPs

You can ask Memex to build and deploy an MCP for you just by describing what you want.

In the below video, you can see a quick example of what it looks like to go from prompt-to-MCP server, fast.