What you’ll know by the end of this check

What connectors are and how they change what Claude can do for you
The difference between web connectors and desktop extensions — and which to set up first
The security model: what Claude can see, and what it can’t

The upgrade connectors unlock

Without connectors, Claude knows what you tell it in the chat. With connectors, it knows what’s actually happening in your tools — your Slack threads, your Drive docs, your Jira tickets, your calendar. You stop copy-pasting context into every conversation and start just asking questions.

This is the shift from “assistant that answers questions” to “informed collaborator.” The connector model lets Claude read your actual data, and in many cases take actions — creating tasks, drafting replies, pulling records — without you leaving the conversation.

The underlying standard is called MCP (Model Context Protocol) — think of it as USB-C for AI integrations. One standard, many tools. Because it’s open, any tool can build a connector and it works with Claude. The directory of available connectors is at claude.ai/directory.

Two types of connectors

Web connectors link Claude to cloud services: Google Drive, Gmail, Slack, Notion, Asana, Linear, Stripe, Confluence, and many more. You connect them from the claude.ai browser interface — find the connector, click connect, authenticate with your existing credentials, grant the permissions you’re comfortable with. Done.

Desktop extensions are connectors that run locally through the Claude desktop app. These let Claude reach things that live on your machine: local files, native apps like Figma, browser automation. They require the desktop app installed.

Start with web connectors. They cover 80% of what most people need and take under 5 minutes to set up.

What to connect first

Pick the two or three tools where you spend the most time. Common high-value starting points:

Google Drive / Notion / Confluence — ask questions about your actual docs without uploading them
Slack — surface what’s been decided, find threads from last month, prep for meetings
Gmail — draft replies, find email threads, summarize what’s outstanding
Linear / Asana / Jira — query your actual task list, create tasks, check status

One connector that’s not available yet and matters to know: GitHub is not in the Enterprise Search connector list. If your team’s knowledge lives in repos, you need a different approach (like a curated knowledge base project) to make that accessible.

The security model

Three things worth knowing:

Claude only sees what you see. Connecting your work email doesn’t give Claude access to your CEO’s inbox. It sees your data, within your permissions, nothing more.
Scoped access. You can review exactly what permissions each connector requests before granting them. Most connectors ask for read access only; action-capable connectors (creating tasks, drafting messages) require broader permissions.
Revocable any time. You can disconnect a service through Claude’s settings or directly through the third-party service’s security/OAuth settings.

Treat connector setup the same way you’d treat granting any OAuth app access to your accounts — read the permission list, grant what makes sense for how you work.

Things to try right now (5 minutes)

Go to claude.ai/directory and connect one tool you use every day. Then ask Claude a question you’d normally need to go dig up yourself: “What are my open Asana tasks due this week?” or “Find the Slack thread where we decided on the Q3 roadmap.” See what comes back when it has real context.

The canonical version

Full official lesson at anthropic.skilljar.com/claude-101/383397 — includes the connector directory walkthrough and practical use-case examples by tool category.

Ready to verify this check?

Finish the official lesson, then come back and mark this check verified on your flight log.