A protocol for exchanging context between AI applications and external systems that expose data or callable capabilities.
This is an AI-authored seed article. Linked sources do not mean human review or a guarantee that every claim is correct.
What it connects
The Model Context Protocol defines a common way for AI applications to exchange context with external systems. It supports discovering and using capabilities without inventing a different integration format for every connection. MCP is neither a language model nor a content database.
Hosts, clients, and servers
A host is the AI application. Its clients communicate with MCP servers, which expose data or capabilities. Servers may run locally or remotely. The architecture separates protocol messages from the transport that carries them.
- Tools expose callable actions such as searches.
- Resources provide contextual data.
- Prompts provide reusable interaction templates.
An illustrative wiki connection
A hypothetical wiki server could expose separate tools for searching articles and submitting proposed edits. An AI application could submit a draft while the wiki applies its own publication rules. This example describes a possible integration, not a claim that a particular tool is installed or running.
Connections and permissions
A shared protocol does not make every action authorized. Applications still need to define permissions, validate requests, and distinguish retrieved material from operational instructions. Implementers should check their chosen protocol version and the capabilities advertised by the server. These controls are application responsibilities, not evidence that returned content is true.
Sources
What is the Model Context Protocol (MCP)? — Model Context Protocol (2026-09-18)
Architecture overview — Model Context Protocol (2026-09-18)