What is MCP (Model Context Protocol)? Definition & Guide

Model Context Protocol (MCP) is an open standard that provides a unified way for AI agents to connect to external tools, data sources, and services. Developed by Anthropic and released as an open specification, MCP defines how an AI agent discovers available tools, understands their capabilities, and invokes them through a consistent interface -- regardless of what the tool does or who built it.

Before MCP, every tool integration required custom code. If you wanted your agent to use GitHub, Slack, and a database, you needed three separate integrations with different APIs, authentication flows, and data formats. MCP standardizes this into a single protocol, similar to how USB standardized hardware connections.

The protocol follows a client-server architecture. An MCP server exposes a set of tools (functions the agent can call), resources (data the agent can read), and prompts (templates for common tasks). An MCP client inside the agent connects to these servers, discovers what is available, and routes tool calls to the right server.

How MCP Works

An MCP server is a lightweight process that wraps an external service. It declares its capabilities using a standardized schema -- tool names, parameter types, descriptions. When an AI agent starts up, it connects to its configured MCP servers and learns what tools are available.

During a conversation, when the AI agent decides it needs to use a tool, it sends a structured request to the appropriate MCP server. The server executes the action (querying a database, posting to Slack, reading a file) and returns the result. The agent then incorporates the result into its reasoning.

Discovery -- The agent learns what tools exist and what they do
Schema validation -- Parameters are typed and validated before execution
Transport flexibility -- MCP works over stdio, HTTP, or WebSocket connections
Authentication -- The server handles auth with the underlying service

Why MCP Matters

MCP solves the N-times-M integration problem. Without it, every agent framework needs custom integrations for every tool. With MCP, a tool built once works with any compatible agent. This creates a growing ecosystem where developers can share tools and agents become more capable over time.

For teams, MCP means less custom development. Instead of writing glue code to connect your agent to internal systems, you can deploy an MCP server that wraps your API and any MCP-compatible agent can use it immediately.

How KiwiClaw Uses MCP

KiwiClaw runs OpenClaw, which natively supports MCP servers as skills. Users can install MCP-compatible tools from the skills marketplace with one click, or configure custom MCP servers in their agent settings. KiwiClaw handles the server lifecycle, connectivity, and sandboxing so MCP servers run securely within the tenant's isolated environment.

Related Terms

Frequently Asked Questions

What is Model Context Protocol (MCP)?

MCP is an open standard developed by Anthropic that provides a unified way for AI agents to connect to external tools, data sources, and services. It defines a common protocol so that any MCP-compatible tool can work with any MCP-compatible agent without custom integration code.

How does MCP differ from function calling?

Function calling is a model-level feature where the LLM outputs structured tool calls. MCP is an application-level protocol that standardizes how agents discover, connect to, and interact with external tool servers. MCP can use function calling under the hood but adds discovery, authentication, and a server specification layer on top.

Does KiwiClaw support MCP?

Yes. KiwiClaw runs OpenClaw, which supports MCP servers as skills. Users can install MCP-compatible tools from the skills marketplace or configure custom MCP servers in their agent settings.