What is the Model Context Protocol (MCP)?
The Model Context Protocol (MCP) is an open standard that lets AI applications connect to external tools, data sources and prompts through one uniform interface. Instead of hand-wiring every model to every system, an app speaks MCP to any number of MCP servers, each of which exposes its capabilities in a predictable, machine-readable way. MCP was introduced by Anthropic and is now implemented across many AI clients and platforms.
Why MCP exists
Before MCP, every integration between an AI model and an external system was bespoke: a custom function, a custom schema, a custom auth flow. That does not scale — N models times M systems is N×M pieces of glue. MCP turns that into N+M: each client speaks MCP once, each system exposes MCP once, and they interoperate. It is often described as "a USB-C port for AI" — a single connector standard.
The core concepts
An MCP server can expose three kinds of capability, and a client discovers them at connection time:
- Tools — actions the model can invoke (query a database, create a ticket, send a message). Each tool has a name, a description and a JSON-Schema for its inputs.
- Resources — read-only data the model can pull in as context (a file, a record, a document).
- Prompts — reusable, parameterized prompt templates the server offers to the client.
Communication is JSON-RPC 2.0. A client connects, negotiates capabilities, lists what the server offers, and then calls tools or reads resources on demand.
How MCP is transported
MCP runs over a few transports, chosen by how the server is deployed:
- stdio — the server runs as a local subprocess; messages go over standard input/output. Common for local, single-user tools.
- Streamable HTTP — the current standard for remote servers: a single HTTP endpoint, with streaming for long-running responses.
- SSE (Server-Sent Events) — an earlier remote transport still widely deployed.
Authentication and security
Remote MCP servers that expose sensitive systems need authorization. MCP defines an OAuth 2.1-based authorization flow, so a client obtains a token and presents it to the server. In an organization, that raises the real questions this Learn section covers: who is allowed to reach which server and which tools, how tokens are issued and stored, and how every call is governed and audited. See MCP Authentication, MCP Authorization and MCP Security.
MCP in an organization
A single developer can point their AI client at one MCP server and be productive in minutes. An organization has a harder problem: dozens of servers, many users and teams, corporate identity, compliance rules, and data that cannot leave the network. That is where an MCP Gateway comes in — a single governed surface that fronts many upstream MCP servers, applies identity and policy, and keeps an audit trail.
How Kravn fits
Kravn is a self-hostable MCP gateway, registry and proxy. It connects to upstream MCP servers, imports their tools, resources and prompts into one registry, and re-exposes them behind governed endpoints — integrated with your identity stack, with no data leaving your perimeter. If you are adopting MCP beyond a single laptop, that is the layer you will want.