What Is the Model Context Protocol?
Also called MCP, MCP server
Short Answer
The Model Context Protocol (MCP) is an open standard for connecting AI applications to tools and data sources. A tool or data source is wrapped once as an MCP server, and any MCP-compatible application can then use it. Anthropic released MCP in November 2024; it is now widely adopted across the industry.
- Taught in lesson 118
- In 1 real job ad
Why It Matters
MCP gives AI applications a standard way to reach tools and data, so a company can connect a system once and use it from many AI tools. Teams building agents and internal AI platforms increasingly use it, and understanding it also means understanding its security risks.
How It Works
MCP is often compared to a USB-C port for AI: one standard plug instead of a custom integration for every pair of model and tool. An MCP server can expose tools (actions), resources (data) and prompts. OpenAI, Google and Microsoft added support in 2025, and in December 2025 Anthropic donated MCP to the Agentic AI Foundation under the Linux Foundation.
Because an MCP server gives a model access to real systems, security matters: limit each server's permissions, trust only known servers and watch for prompt injection hidden in the content tools return.
An MCP server exposes tools, resources and prompts; an MCP client inside an AI application discovers them and offers them to the model. Servers can run locally or over the network. Security is part of the design: servers should require authentication, grant the smallest permissions needed, validate inputs and log actions, because content returned by tools can contain hidden instructions that try to steer the model.
Model Context Protocol in Practice
Everyday Example
A developer installs a GitHub MCP server once, and their AI coding assistant can then read issues and open pull requests in GitHub.
Industry Example: Pharma
A research team wraps its compound database as an MCP server so approved AI assistants can look up compound data through one controlled, logged access point.
Where You Meet It
- AI assistants that connect to company tools.
- Coding tools that read repositories and issue trackers.
- Internal platforms that offer approved tools to many teams.
- Job ads for agent and AI platform engineers.
Key Ideas Around It
- MCP server
- A program that offers tools and data through MCP.
- MCP client
- The part of an AI app that talks to servers.
- Resource
- Data a server lets the model read.
- Least privilege
- Giving each tool only the access it needs.
A Common Mix-Up
MCP does not make a tool safe by itself. It standardises the connection; permissions, validation and monitoring are still your job.
Check YourselfWhat is the main benefit of MCP for a company with many AI tools?
A system is wrapped once as an MCP server and every compatible AI application can use it, instead of building a custom integration for each.
Seen in Real Job Ads
Lines quoted word for word from live job postings.
Novartis
Scientific AI Platform Expert, Disease Area X, Cambridge, MA
Experience with omics analysis pipelines (i.e. genomics, transcriptomics), MCP, LLM patterns, biological foundational models, agent orchestration, and deep learning frameworks (i.e. Pytorch) is preferred.
Learn It on ai2hired
S10 · #118MCP Explained
Coming in Season 10, Agents and Tool Use. Every ai2hired lesson is free on YouTube.
Questions People Ask
Who created MCP?
Anthropic released it as an open standard in November 2024 and donated it to the Agentic AI Foundation, under the Linux Foundation, in December 2025.
What is an MCP server?
A small program that exposes a tool or data source through the protocol, so any MCP client, such as an AI assistant or agent, can use it.
Related Terms
Learn AI From Zero to Hired
Free lessons on YouTube, each one built from what real AI job ads ask for.