Glossary
What is MCP (Model Context Protocol)?
MCP, short for Model Context Protocol, is an open standard that lets AI applications connect to tools and data through small programs called MCP servers. It gives an AI agent one common way to read files, query systems and take actions, instead of a custom link for every tool.
By the AIBound team · Updated 15 November 2026
Key takeaways
- MCP is an open standard that gives AI agents one common way to reach tools and data.
- Each MCP server is a small program that sits between an agent and a tool, so every server is a new connection to check.
- Employees can add MCP servers to their own devices in minutes, so a company needs to find them before it can set rules.
Why MCP matters
Before MCP, every AI tool needed its own custom connection to each system. MCP gives agents one common way to connect, so they can do far more with far less setup. That is why teams adopt it quickly. It also means every new server is a new door into files, data and systems, and many are added without review.
How MCP works
MCP has three parts. The AI app the employee uses is the host. It runs a client for each MCP server it connects to, and each server offers one or more tools. The agent asks, the server does the work and sends back the result.
- 1 The AI app connects to an MCP server, which can run on the employee device or elsewhere.
- 2 The server tells the agent which tools it offers, such as reading a file or searching a database.
- 3 The agent asks the server to run a tool and uses the result to carry on with its task.
MCP vs API
An API is an interface a service publishes for software to call. MCP is a common way for an AI agent to find and use tools, often by calling APIs on its behalf.
| Question | MCP | API |
|---|---|---|
| What is it? | A shared standard for connecting AI agents to tools and data | An interface a service exposes for software to call |
| Who uses it? | An AI agent, through an MCP server | A known application or a developer's code |
| What changes for security? | The agent decides when to call a tool, so the servers it can reach matter | The service controls who may call it and how often |
Sources
- Model Context Protocol: Introduction · modelcontextprotocol.io
- Model Context Protocol: Architecture overview · modelcontextprotocol.io
Sources checked 19 November 2026
Examples
MCP examples
FAQ
Frequently asked questions
What does MCP stand for?
MCP stands for Model Context Protocol. It is an open standard for connecting AI applications to tools and data. An AI agent uses it to find out what a tool can do and to ask the tool to act, without a custom link for each system.
Is MCP the same as an API?
No. An API is an interface a service publishes. MCP is a shared way for an AI agent to discover and use tools, and an MCP server often calls APIs on the agent's behalf. The agent decides when to call it, so the risk sits with the agent as well.
Is MCP safe to use at work?
MCP itself is a standard, so the risk depends on which servers are installed and what each one can reach. A server that nobody reviewed, with wide access to files or systems, is the usual problem. Companies need to find the servers and set rules for them.
What is an MCP server?
An MCP server is a small program that offers tools to an AI agent, such as reading a file or searching a system. It sits between the agent and the tool. Employees can install one on their own device in minutes, which is why security teams track them.