Some of our users asked, “The AI Integration directions list code generators, not AI agents; why?” Or “Why must I install a code generator to use the MCP?”

A Model Context Protocol (MCP) helps an AI agent work with another system — like the LRS — so it can answer questions and perform operations. The AI agent doesn’t have eyes or fingers, so it can’t use the regular Graphical User Interface (GUI) you use. Instead, the AI agent generates code to do what you ask and sends it to your LRS via an Application Program Interface (API). The MCP helps the AI agent write better queries and tasks in the code that it sends via the API. 

Most AI agents keep their code-generator in a separate application. The thinking part of the agent — the “Model” — is remote; but the code-generator application is local: installed on your computer. This configuration lets you enhance the code generator by also installing programming languages for it to use, like node.js and Python. Also, any files the AI agent generates, it backs-up to your computer, which offers some privacy and security. For our purposes, it also lets you customize the application to include the MCP interface.

You can usually find the MCP configuration file in the settings for the AI code-generating application in your Users directory: e.g., C: > Windows > Users > yourname > AI_application > config > mcp.json.