DESK · THEORY
ExplainerIntermediate · June 2, 2026 · 4 min read
On this page

What are connectors in Claude Code?

The plug that lets Claude reach into the tools you already use, your email, Slack, your numbers, and do the work there instead of just talking about it.

"Draft replies to the five most important emails in my inbox and leave them in my drafts." You type that into Claude Code half-expecting a lecture about how it can't see your email. Instead it reads the five threads, writes five replies in your voice, and saves them as drafts. You switch to Gmail and they're sitting there, waiting for you to hit send.

It can do that because of one thing somebody set up once: a connector. Without it, Claude works inside the folder on your laptop. With it, Claude reaches the systems your business actually runs on.

What it is

A connector is the bridge between Claude and a tool you already use, your email, your calendar, Slack, a Google Doc, your customer database. Plug one in and Claude can read from that tool and act inside it, not just give you advice about it.

Under the hood, connectors run on an open standard called MCP, the Model Context Protocol that Anthropic published in late 2024 and most of the industry now speaks. That linked explainer covers what MCP is and why it should change how you buy software; here we'll stay on the practical side. The one thing to carry over: you will see "MCP server" and "connector" used for the same thing. A connector is an MCP server with a friendlier name and, usually, a one-click login.

You add one by typing claude mcp add and pasting the tool's address, or by picking it from Anthropic's connector directory and signing in. Inside a session, the /mcp command shows you what's connected and walks you through the login (the terminal tricks page has the mechanics). None of this is coding. You're authorizing an app, the same way you'd connect a new tool to your calendar.

Why it matters

Out of the box, Claude is brilliant at anything inside the folder you opened: your files, your notes, your drafts, and whatever you put in its CLAUDE.md brief. The moment your work depends on a system that lives somewhere else, a chat tab turns you into the courier. You copy the data out of the other tool, paste it in, copy the answer back, paste it where it belongs.

Connectors delete the courier. Claude reads the calendar itself, queries the numbers itself, posts to the channel itself. The felt difference is the one you get when you stop CC'ing yourself on tasks and just hand them to a capable assistant who already has the logins. I run Claude across a few companies, and the connectors that earn their keep are boring and load-bearing: the inbox, the calendar, the docs, the numbers. Once those are plugged in, "look into this and get back to me" becomes a thing you can say to software.

What a good setup looks like

Start with the tools you actually live in, and connect them one at a time:

Anthropic's directory lists hundreds of connectors, most built by the tool makers themselves. Use the official ones, and connect read-only first. You can always grant more later.

Common mistakes

Connecting a tool you don't trust. A connector can carry hidden instructions inside the data it returns, a trick called prompt injection. Anthropic says it plainly: a server that fetches outside content can try to steer Claude into doing something you never asked for. Only plug in connectors from a company you'd already trust with that data.

Granting write access on day one. Read-only is the safe default. Let Claude see your inbox before you let it send from it. Earn the trust in that order, the same way you would with a new hire.

Connecting everything at once. Five connectors you understand beat twenty you forgot you turned on. Each one is a door into your business; only open the ones you use.

Forgetting the perimeter. Connectors and permissions work together. The connector decides what Claude can reach; your permission settings decide what it can do without stopping to ask. Set both deliberately, not once you're already in a hurry.

Do this next

Pick the one tool you copy-paste in and out of most, usually your inbox, and connect it. Then read Claude Code tips for non-technical CEOs for the habits that turn all this reach into actual leverage, and the pillar on why a CEO belongs in the terminal if you still want the bigger case. The connectors are the reach. The habits are what you do with it.

The Thursday 3

Get three workflows like this every Thursday

The Thursday 3 is a free weekly email. Three workflows that put you in the top 1% of CEOs. 90-second read. Every card links back to a step-by-step guide like this one.

Get the newsletter →
The Desk Theory books

The architecture behind this workflow.

Two operator manuals for the same job, run two ways: OpenCLAW for the always-on harness, Claude Code for the focused-work CLI. Pick one, or get the bundle for $149.

Browse the books · $99 each

Want one workflow like this taken apart end-to-end every week? The Tuesday Pro Deep Dive · $39/mo.