MCP Explained: The Protocol Powering AI Integrations

What is MCP

MCP (Model Context Protocol) is an open standard. It lets AI clients connect with external services and tools. It is a universal adapter between AI models and the real world.

Before MCP, every AI integration needed custom API code. Each integration required its own authentication flow and tool-specific configuration. MCP puts all of this into one protocol.

How MCP works

MCP uses a simple client-server architecture:

MCP Client: Your AI application, like Claude, Claude Code, or Codex.
MCP Server: The service providing tools, like Convika.
Tools: Individual capabilities the server exposes, such as create page, publish, and get analytics.

If you use subagents in Claude Code or Codex, see Codex vs Claude Code subagents for how they inherit or configure MCP servers, along with their limits and how to give a single subagent an MCP server. For skills, which Claude Code and Codex load only when a request calls for one, see Codex vs Claude Code skills: SKILL.md paths, /name vs $name.

When you connect an MCP client to an MCP server:

The server advertises its tools, with descriptions.
The AI client discovers what it can do.
You ask the AI for tasks in plain words.
The AI picks the right tools and runs them.

Why MCP matters

For Users

One protocol, many services: You connect your AI to any MCP-compatible service the same way.
Natural language interface: You have no new UIs or CLIs to learn. You talk to your AI.
Workflow integration: You stay in the AI client you already use instead of switching tools.

For Developers

Standard implementation: You build one MCP server, and it works with every MCP client.
Rich tool descriptions: AI clients understand your tools through detailed descriptions.
OAuth 2.1 authentication: You get secure, standards-based authentication right from the start.

MCP and Convika

Convika's pages, forms, analytics, and domains are MCP tools too, which means:

Create a landing page by asking Claude: "Create a landing page for my product"
Publish by saying: "Publish my page to production"
Check performance: "How many visitors did my page get this week?"

You do not need to leave the chat. The same pages, forms, and analytics are in the Convika dashboard too. Create a page with the dashboard's AI or upload an HTML file, and check the numbers there. MCP is optional.

Getting started

To connect to Convika over MCP, add the server URL to your AI client:

https://mcp.convika.com

On your first connection, you are asked to authorize access over OAuth. After that, your AI client has full access to Convika's tools.

To set up your client, see the Getting Started guide.

If Codex or Claude Code saved a page as a file on your computer instead of creating it in Convika, How to preview an HTML file from Codex or Claude Code shows how to open it and how to see it on a phone.