Glossary · last updated 2026-06-08
MCP (Model Context Protocol)
Also known as: Model Context Protocol
An open standard (introduced by Anthropic in late 2024) for connecting AI models to external tools and data sources through a common interface. Relevant to AI search as the plumbing that lets agents fetch, query, and act on content rather than only reading crawled pages.
The Model Context Protocol is an open standard for how AI models and the applications around them connect to external tools, data, and services. Instead of every AI product building bespoke integrations, MCP defines a common interface: a model (or the agent wrapping it) can discover and call MCP "servers" that expose data sources, APIs, and actions. Anthropic introduced it in late 2024 and adoption has spread across the ecosystem since.
For most publishers, MCP is not a direct optimization surface the way llms.txt or schema is. Its relevance to AI search is forward-looking and structural: as agentic search matures, the agents acting on a user's behalf increasingly reach content and capabilities through protocols like MCP rather than by reading a crawled HTML page. An organization that exposes its data or product through an MCP server is making itself consumable by agents in a way a static page cannot match.
The practical near-term implication is to watch the space rather than over-invest. The durable AI-visibility work is still content and entity authority. But teams whose product or data could be useful to an AI agent (pricing, availability, structured reference data) should track MCP as the emerging interface through which agents will increasingly consume and act, the same way the web learned to expose APIs alongside pages.
Related