GlossaryAI search
What is the Model Context Protocol (MCP)?
Also called: MCP, MCP server, Model Context Protocol server
Definition
The Model Context Protocol (MCP) is an open standard for connecting AI applications to external data sources, tools and workflows, so an assistant can read data or take actions in another product through one common interface.
Model Context Protocol (MCP), explained
Anthropic introduced and open-sourced MCP on November 25, 2024, describing it as a standard for connecting AI assistants to the systems where data lives. The architecture is simple: products expose data and actions through MCP servers, and AI applications act as MCP clients that connect to them. The project's own docs compare it to a USB-C port for AI applications, and note that AI apps like Claude or ChatGPT can use it to reach data, tools and workflows.
Before MCP, every integration between an AI app and another product was custom. MCP lets a product build one server and be usable from any client that supports the protocol: chat assistants, coding tools, IDEs and agent frameworks.
For a startup, an MCP server is a distribution decision as much as a technical one. If your users spend their day inside an AI assistant or coding tool, an MCP server lets them use your product from there: query their data, create records, trigger jobs. Developer-tool companies in particular now list MCP support next to their API and SDKs, and MCP server directories have become another place to be listed.
Security deserves real attention. An MCP server can expose sensitive data and actions to a model, so scope permissions tightly, require proper authentication, log what's called, and treat anything the model sends as untrusted input.
MCP doesn't replace a good API, docs or SEO. It's one more surface where buyers can find and use you, and the product still has to be worth connecting.
Why it matters for founders
If your buyers work inside AI assistants, an MCP server puts your product where they already are. It's a new integration channel, and early, well-documented servers get noticed.
Example
An analytics startup ships an MCP server with read-only tools for querying dashboards. Users ask their assistant "what were signups last week?" and get the answer from the startup's data, with a link back to the dashboard.
Common mistakes
- Exposing write actions without strict scopes and authentication.
- Shipping a server with no docs, examples or listing anywhere.
- Treating MCP as a replacement for a public API.
Sources
- Anthropic: Introducing the Model Context Protocol (November 2024)
- Model Context Protocol: What is MCP?
Checked
Related terms
- AI agentAn AI agent is a system in which a language model decides its own next steps and uses tools, such as web browsing, search, code or APIs, to complete a task with limited human direction.
- Marketplace listingA marketplace listing is a page for your product inside another platform's app store or integration directory, such as the Shopify App Store, Chrome Web Store or Slack Marketplace, where that platform's users find and install it.
- llms.txtllms.txt is a proposed markdown file at a site's root, /llms.txt, that summarizes the site and links to its most useful pages for AI agents. It's voluntary, grants no access, and Google Search ignores it.
- Large language model (LLM)A large language model (LLM) is an AI model trained on very large amounts of text to predict the next piece of text. That one skill lets it answer questions, summarize, write and reason in plain language.