Intro to Model Context ProtocolModel Context Protocol
The module uses Model Context Protocol (MCP) to connect AI models to real-world tools and data. It covers the MCP concepts, builds a TypeScript server that exposes a weather tool and a resource, and then looks at transports, trust boundaries and clients. The server talks to its client over standard input and output rather than a web port, so testing it means running a client such as MCP Inspector, not opening a page.
These pages accompany the Model Context Protocol module of Scrimba's AI Engineer Path. They don't replace the lessons; they hold the setup detail that changes faster than a recorded scrim can, so it can be corrected here instead of re-recorded.
The pages
Running Model Context Protocol locally is for when you've downloaded the server project from the lesson menu and want it running on your own machine. It covers what to install first, how to install and repair the server, how to test it with MCP Inspector, how to connect a desktop MCP client to it, and the errors most learners hit on the way.
Looking for a different course?
Run an AI course project locally lists every course and project that has a setup guide.