MCP

MCP is a standard way for AI systems to connect to tools, data, and services in a consistent and predictable manner. It defines how an AI agent can discover what it is allowed to do and how to interact with external systems safely.

I use it within Agentic AI systems to give access to tools, which allow the LLM to invoke actions in the world.

MCP enables agentic AI to operate reliably inside real products and workflows rather than alongside them. It improves safety, maintainability, and interoperability, which is essential for building systems that are intended to run in production rather than remain as experiments.