L027 · Under the hood · 1 min
MCP: a universal plug for AI
MCP is a shared standard that lets any AI assistant connect to apps and data, like a USB-C port for AI.
01 · The idea
For an assistant to read your calendar, search your company wiki or update a spreadsheet, it needs a connection to that app. Building each one by hand was slow.
The Model Context Protocol, or MCP, is an open standard created in 2024 and now supported by most major AI companies and many apps. An app offers an MCP server once, and many AI tools can plug in.
For users, this shows up as connectors or integrations: "connect Google Drive", "connect Notion", "connect GitHub". The AI can then read or act inside that app with your permission.
Only connect apps you trust, and give the least access needed. A connection that can read your email can also be tricked by a bad email.
02 · Think of it like this
Before USB, every device had its own cable. MCP is like one plug shape that fits every AI and every app.
03 · Do it now
04 · Quick check
Words in this lesson
See also
L163Under the hood · 3 min
MCP safety: what can go wrong with connectors
Every connector is a door between the AI and your data, so check who made it, what it can touch and what text it lets in.
L187Under the hood · 2 min
Browser agents: AI that uses websites for you
A browser agent looks at web pages, decides where to click and type, and repeats until a task is done, a handy trick that is slower and riskier than it looks.
L052Staying safe · 3 min
Prompt injection: hidden orders in the text
Prompt injection is when a web page, email or file contains hidden instructions that an AI reads and follows as if you had written them.
L155AI at work · 3 min
AI in your browser, and staying safe with it
A browser helper like Claude in Chrome can read the page you are on and click for you, so it needs the same caution you would give anyone using your logged-in browser.