Content Island MCP

In this series of tutorials we are going to discover how Content Island MCP works: A tool that connects your favorite AI with the content of your Content Island project.

What is this for? Very simple: it allows us to develop complete projects with the help of an AI agent, making the most of our content.

The practical example

Imagine that we want to create a simple application to visualize the documentation of a project.

In addition, we want a person with the role of content manager to be able to modify it without depending on a front end developer.

The user interface could look something like this:

Mock baja fidelidad portal simple doc, sidebar izquierda con árbol de carpetas y páginas, y derecha el contenido de la página seleccionada

Our data model in Content Island

In Content Island we already have the following schema defined:

Root: the main entity that groups all the folders.
Folder: each folder contains its items.
Item: the pages, with their name, slug (friendly URL) and content in Markdown.

Folder con campos title, slug y items. Item con campos title, slug, description, content, Root con campo folders

And we have already entered data, we can move to the content tab.

The plan

We want to develop a rapid prototype to validate the idea and gather early feedback.

For this we are going to use Copilot in agent mode, directly from VSCode.

Copilot serves as an orchestrator: by interacting with it we can generate the project.

The interesting thing is that, behind Copilot, you choose which AI engine to use: it can be Claude Sonnet, GPT or the one you prefer.

This way, we usually get a working prototype in a very short time.

But... one important piece is missing.

Nos falta la integracación de copilot con Content Island

Even if Copilot can orchestrate the AI, it doesn't know how to access Content Island to extract the content.

And that's where Content Island MCP comes in, the bridge that connects the two parties.

Content Island se integra con el MCP de Content Island

With this we can now develop a working prototype.

If you want to go ahead and see the final result:

In this sandbox you can find it: https://stackblitz.com/github/content-island/sanbox-agent-mcp-doc-portal-prototype
The repository: https://github.com/content-island/sanbox-agent-mcp-doc-portal-prototype

Conclusion

We now have all the pieces to start.

In the next video we will see how to install and configure Content Island MCP step by step.

Tutorial: Content Island MCP