Module 1 · about 60 min
Why MCP?
AI is already capable, so why does it need a standard way to reach tools?
0 / 5 completed
AI 201Available
Model Context Protocol: Connecting AI to Tools & Trusted Data
From using AI to letting AI use tools and trusted data safely: understand MCP architecture, trace a tool call, compare MCP, APIs and RAG, and design an implementable MCP Tool Canvas.
My learning progress
Overall progress: 0% (0 / 32 items completed)
- Why MCP? (0 / 5)
- MCP Architecture (0 / 5)
- Tool Discovery & Tool Calls (0 / 6)
- MCP vs API vs RAG (0 / 5)
- Security, Grounding & Provenance (0 / 6)
- Practice A: Ask · Not done yet
- Practice B: Inspect · Not done yet
- Practice C: Connect · Not done yet
- AI 201 assessment · Not done yet
- MCP Tool Canvas · Not done yet
We suggest finishing AI 101 first. This is a suggestion only and never blocks you from starting.
Learning outcomes
- Explain what MCP is for and the roles of Host, Client, Server and Tool
- Trace a full tool call, from the request to a human-verified answer
- Choose between prompt-only, RAG, API and MCP tool for a given scenario
- Check a tool output for source, freshness and provenance
- Complete an MCP Tool Canvas covering schema, permission, risk and provenance
Prerequisites
- AI 101, or a basic grasp of generative AI and prompting
- You know AI output needs human verification and sensitive data must not be handed to AI casually
- No programming background required
Who it is for
- Staff and faculty who finished AI 101 and want AI to work with campus tools and data
- Coordinators and managers who plan or review AI tool adoption
- Students and researchers who want to understand how AI calls tools safely
Learning journey
Modules
Module 2 · about 60 min
MCP Architecture
How Host, Client, Server, Tool and the trusted data source fit together.
0 / 5 completed
Module 3 · about 60 min
Tool Discovery & Tool Calls
Trace how AI discovers a tool, prepares arguments, gets a result and answers.
0 / 6 completed
Module 4 · about 60 min
MCP vs API vs RAG
Pick the right pattern for the scenario, and stop treating MCP as an API replacement.
0 / 5 completed
Module 5 · about 60 min
Security, Grounding & Provenance
Permission, risk, source tracing and human confirmation: the core module of the course.
0 / 6 completed
Progress is stored only in this browser. It is never uploaded and does not identify you.