The short version
Before MCP, every AI application needed its own custom integration for every tool. MCP gives them a common language. An MCP client (inside an AI application) connects to an MCP server, which advertises tools, resources and prompts the AI can use.
What an MCP server provides
- Tools — actions the AI can call, such as “search files” or “create draft”.
- Resources — data the AI can read.
- Prompts — reusable instructions.
What MCP does not decide
MCP describes how an AI calls a capability. It does not, on its own, decide whether a particular person’s AI should be allowed to call it, on which data, with what approval — or keep a record across all your AI tools. Those are organisational questions.
Where Kroy fits
Kroy exposes an MCP endpoint to AI clients and sits in front of your business systems. Each call is checked against the person, the agent, the delegation and your policies before anything happens.