Skip to content
Shared AI context What is MCP?

Explained 3 min read

What is MCP? A plain-language explanation

MCP stands for Model Context Protocol. It is a common way for AI apps to connect to information and tools outside themselves. You do not need to be a developer to understand it, or to use it.

How MCP connects

  1. ClientThe AI appChatGPT, Claude, Codex or another
  2. ConnectorAn address and a sign-inYou add it once
  3. ServerYour VisionApproved Direction to read
An illustration. Claude, ChatGPT, Gemini, Codex and Obsidian are trademarks of their respective owners.

In short

  1. MCP is a shared standard for connecting AI apps to outside information.
  2. A server offers information or actions, and the AI app connects to it as a client.
  3. Before you connect, check what it can read, what it can change, and how to revoke it.

The short version

MCP in one sentence

MCP is an open standard that lets an AI app, such as ChatGPT, Claude or Codex, ask an outside source for information or ask it to do something, in a way every supporting app understands.

Anthropic introduced it in 2024, and many AI products now support it.

The parts

Server, client, connector

  • Server

    The outside source. It offers information or actions, such as your Vision.

  • Client

    The AI app that connects to it, such as ChatGPT, Claude or Codex.

  • Connector

    What some apps call the connection you add. In practice it is an address and a sign-in.

What it does

What an MCP connection can and cannot do

It can

  • Let an AI read information you choose to share
  • Let an AI use tools the server offers
  • Keep the source in one place, outside any one app

It cannot

  • Make an AI trustworthy by itself
  • Decide what is correct or current
  • Replace your own review of the results

VisionBoard

How VisionBoard uses it

VisionBoard offers your approved Direction at one address: https://visionboard.si/api/mcp. You add it to a tool that supports MCP, sign in once and choose which Vision it may see.

The tool reads approved material and cites the exact version. If you let it save work, that work arrives as a Draft with its name on it. The Connect AI page has the steps for each tool. Not every AI product supports MCP, so it does not work with every tool.

Before you connect

Four things to check before you connect anything

  • What can it read?

    Look for a clear limit, such as one Vision, and approved material only.

  • What can it change?

    Prefer read-only by default, with changes saved as reviewable drafts.

  • Who runs it?

    Check who operates the server and what they say about your data.

  • Can you revoke it?

    You should be able to remove access at any time.

Useful questions.

Is MCP only for developers?

No. Developers build servers, but anyone can add a connection to an app that supports it, usually with an address and a sign-in.

Is an MCP server the same as an API?

They are related. An API is how software talks to a service. MCP is a common way for AI apps to find and use what a service offers, so each app does not need its own custom integration.

Which apps support MCP?

Many do. The Connect AI page lists the tools we have seen working with VisionBoard, with an honest status for each.

One source

Give every AI
the same direction.

Start your Vision

One active Vision free. No credit card required.

Claude, ChatGPT, Gemini, Codex and Obsidian are trademarks of their respective owners.