Skip to content
NextiaNextia Learning

MCP: Connect AI Applications to Tools and Data

Connect an AI application to tools and data once, with a standard protocol, and keep control of what it may do. You build one MCP server for Larkfield's support knowledge, inspect it, connect an AI application to it, and protect it with sign-in, permissions and a person's approval.

Start course See the curriculum

Read every lesson as a guest, or sign in to enroll and track your progress.

What you'll learn

By the end of this course, you will be able to:

  • Explain MCP host, client and server responsibilities.
  • Expose tools, resources and prompts with clear contracts.
  • Connect and inspect a local server and plan remote access.
  • Enforce permissions and test compatibility and failures.

About this course

Larkfield’s support knowledge is needed in more than one AI application: the help desk’s assistant, an editor assistant that the staff use, and soon a partner’s app. Writing a custom integration for each one is slow, and every copy can go wrong in its own way.

In this course, you build one MCP server, support-mcp, that offers the support knowledge once: a search over the policies, a ticket lookup, the policies themselves, and a prompt to draft a reply. Any AI application with an MCP client can use it. You design its capabilities first, read the real messages that travel between the programs, and inspect the server with MCP Inspector.

Then you connect a small AI application to it, first with a mock model and then, if you want, with a real one. You make the server available over the network, and you check who the caller is, in which organization and with which rights, on the server, for every request.

Finally you protect it. A ticket that tries to trick the model cannot make the server do more than the caller may do, and the one change it offers, a refund proposal, waits for a person’s approval. Everything runs on your computer, every message in the lessons comes from a real run, and the lessons say which protocol revision and SDK version they use.

Before you start

You completed Building Reliable Applications with LLM APIs, Build and Deploy Your First API, Docker and Cloud Essentials for AI Developers, Authentication, Permissions and Multi-Tenant Applications, and AI Security, Privacy and Responsible Design, or you have the same skills. Modules 1 to 4 build mainly on the first two, and use the organizations, roles and scopes of the authentication course. You do not need an account or money. The server, the AI application with a mock model and the practice sign-in all run on your computer; a real model is optional.

Courses to take first: Building Reliable Applications with LLM APIs, Build and Deploy Your First API, Docker and Cloud Essentials for AI Developers, Authentication, Permissions and Multi-Tenant Applications, AI Security, Privacy and Responsible Design.

Who this is for

People who have built an application on a model API with tool calling and an API with sign-in and permissions, as in Building Reliable Applications with LLM APIs, Build and Deploy Your First API and Authentication, Permissions and Multi-Tenant Applications, and who now want several AI applications to share the same tools and data safely. Developers and AI engineers.

How we made this course

Experienced AI practitioners designed this course. AI helped us draft the content; people chose the teaching approach and the examples, and reviewed every lesson. The outputs that you see are the actual results of running the code. How we make our courses

Inside the course

Module 1Understand the integration problem

3 lessons

See the problem MCP solves. Compare a direct integration, tool calling and a shared protocol, meet the host, the client and the server, and write down the versions you build on.

  1. Where MCP fits50 min
  2. Participants and boundaries50 min
  3. Version awareness45 min

Module 2Design useful server capabilities

3 lessons

Design before you build. Choose narrow tools, readable resources with stable addresses, and a prompt that a person starts, each with a clear contract and bounded results.

  1. Tools55 min
  2. Resources50 min
  3. Prompts45 min

Module 5Remote access and authorization

3 lessons

Open the server to remote clients without opening it to everyone. Serve it over Streamable HTTP, check every token and membership on the server, and follow the official authorization flow.

  1. Remote transport50 min
  2. Identity and permission55 min
  3. OAuth integration55 min

After the case studiesFinal assignment

Use every module in one practice project. You collect evidence, and check it against the course outcomes.

  1. Final assignment180 min

Case studies

Realistic projects from start to finish, on real data, with every output explained. Each one makes a portfolio piece. Complete at least one for the certificate; the others are optional.

  1. An MCP server over a real open dataset3 hr

    Expose a permissively licensed public dataset through read-only tools and resources with bounded results, stable URIs and contract tests.

  2. Security review of an MCP integration3 hr

    Find and fix the weaknesses of a deliberately weak local MCP server with tests: a poisoned tool description, a passed-through token, a cross-tenant read and an unapproved write.

Reviews

Loading the reviews…

    Took this course? Sign in to review it. Learners can write a review after they complete 25% of the lessons.

    Share this course