← Videos
MCP Explained

AI from the Ground Up8 Nov 2026 · 3:41

MCP Explained

How does an AI assistant connect to a knowledge library, call a tool and return an answer you can check?

From the episode description

How does an AI assistant connect to a knowledge library, call a tool and return an answer you can check? This practical introduction to the Model Context Protocol follows the host, client and server through tool discovery, a request and a result. We then inspect the code of a small local MCP server before returning to an illustrated ResearchNode lookup. In this episode: • The roles of the host, client and MCP server. • How a tool’s name, input schema and callback fit together. • A locally executed example that searches two invented notes. • Successful, empty and invalid-input results from that example. • How source passages, dates and saved revisions help you check an answer. The teaching server was run locally. The ResearchNode retrieval sequence is an illustrated walkthrough using a saved public MCP source, not a fresh screen recording or live API demonstration. It is separate from the fictional Workshop Notes examples elsewhere in this series. This episode focuses on the tool workflow, not every MCP feature. Input validation does not establish factual truth, and connecting a tool does not remove the need for application and server access controls. ResearchNode: AI from the ground up • Episode 08 of 10. Previous: From Chatbot to Agent: Giving AI Tools Next: Can a Document Trick an AI Assistant? Presented by Michelle, an AI-generated presenter.

Descriptions and chapter markers are supplied by the channel. They are not a transcript or an independent check of every claim in the video.