Model Context Protocol guide
What is MCP?
Model Context Protocol is an open standard that gives AI applications a consistent way to connect with external tools, data sources, resources, prompts, and workflows.
MCP in one sentence
An MCP client inside an AI application connects to an MCP server that exposes approved capabilities in a structured form.
Core architecture
The building blocks of MCP
MCP creates a shared protocol for discovering capabilities and exchanging structured requests, context, results, and errors.
Client
The part of an AI application that connects to an MCP server, discovers capabilities, and exchanges protocol messages.
Server
A local or remote system that exposes selected capabilities to compatible MCP clients.
Tools
Callable actions that can search, calculate, create, update, query an API, or trigger a workflow.
Resources
Context or data that a client can retrieve and provide to the model or user experience.
Prompts
Reusable interaction templates that an MCP server can make available to a compatible client.
Human control
Permissions, review, confirmation, and interface safeguards placed around sensitive capabilities.
Example interaction
How an MCP connection works
01
Connect
The AI application opens a connection with an approved MCP server.
02
Discover
The client learns which tools, resources, or prompts are available.
03
Request
The application selects a capability and supplies structured inputs.
04
Return
The server returns a result, resource, or error for the application to present.
MCP and websites
MCP can support a website without living inside the page.
A website, web application, assistant, or developer tool may use an MCP server for external data and actions. MCP is not limited to browser pages or visible website controls.
Where WebMCP fits
WebMCP focuses on the webpage currently open in a supported browser. It lets the page expose structured tools for visible, human-in-the-loop interaction with a browser agent.
Common questions
MCP FAQ
What does MCP stand for?+
MCP stands for Model Context Protocol. It is an open standard for connecting AI applications with external systems such as tools, data sources, files, databases, and workflows.
What is an MCP client?+
An MCP client is the component inside an AI application that connects to an MCP server, discovers available capabilities, and exchanges structured protocol messages.
What is an MCP server?+
An MCP server exposes approved capabilities to compatible clients. Depending on the implementation, those capabilities may include tools, resources, prompts, or access to external services.
What is an MCP tool?+
An MCP tool is a callable capability described with a name, purpose, and input schema. A compatible AI application can request it to query a system, call an API, perform a computation, or trigger a workflow.
Is MCP only for websites?+
No. MCP can connect AI applications with local software, cloud services, databases, files, developer tools, business systems, and many other external capabilities.
How is MCP different from WebMCP?+
MCP focuses on client-server connections with external systems. WebMCP focuses on structured tools exposed by a webpage inside a supported browser context. A product can use both.