What is MCP?
Model Context Protocol (MCP) is an open standard, developed by Anthropic, that provides a universal way for AI models to communicate with external tools and data sources. Think of it as a USB standard for AI: one protocol that lets any AI tool connect with any external service.
Previously, every AI tool had to build its own integrations. With MCP, you write one server that works with Claude Code, Cursor, and any other MCP-compatible client.
How does it work?
MCP works on a client-server model:
- MCP Client — The AI tool (e.g. Claude Code or Cursor) that wants to use tools
- MCP Server — A lightweight service that offers specific functionality (e.g. database access, API calls, file management)
- Transport — Communication runs via stdio or HTTP, depending on the setup
An MCP server defines tools that the AI can call. The AI decides when and how to use a tool based on context.
Practical applications
MCP opens the door to powerful integrations:
- Database queries — The AI can directly query your database without you having to write queries
- CRM integration — Connect with Dynamics 365 or other CRM systems to retrieve customer data
- Deployment — Automatically deploy to Vercel, Azure or other platforms from your AI editor
- Monitoring — Retrieve real-time logs and metrics while debugging
- Documentation — AI automatically searches your internal documentation and knowledge base
MCP in Cursor and Claude Code
Both tools support MCP servers:
Claude Code is configured via the settings.json file or via claude mcp add. Claude Code can use multiple MCP servers simultaneously and automatically chooses the right tool for the task. Want to learn more? Check out our Claude Code training.
Cursor supports MCP through settings. You add servers and Cursor makes them available to the AI agent during coding. We also offer a Cursor training to help you get the most out of it.
In practice, this means your AI assistant can not only write code but also directly communicate with your databases, APIs and infrastructure.
Building your own MCP server
Building a simple MCP server is surprisingly accessible. With the official TypeScript or Python SDK, you define tools that the AI can call. Each tool has a name, description and input schema.
At Breathbase, we build custom MCP servers for clients who want to enrich their AI workflows with business-specific data and processes. Think of an MCP server that connects to your Dynamics 365 environment, so your developers can directly query customer data while writing code.
Curious about the possibilities? Check out our AI Development services or get in touch.
