In Course 1, Claude worked with files in your local folder — CSVs, Markdown notes, text documents. Those files were a snapshot: data you exported at a point in time, manually saved to the right folder.

MCP changes that. With an MCP connection, Claude has live access to your tools and data — not a snapshot, but the real, current state. No exporting. No copying. No manual prep.

What MCP Is

MCP stands for Model Context Protocol. It is an open standard — built by Anthropic and now supported across the AI industry — that lets AI models connect to external tools and data sources.

Think of it as a plug. Claude is the socket. An MCP server is the plug that connects Claude to a specific tool: your Amazon data, your accounting software, your calendar, your project management system.

When an MCP server is connected, Claude can query that tool directly. Instead of “here is a CSV from last week,” it becomes “here is your live Amazon data right now.”

What MCP Is Not

MCP is not something you build. It is something you install and configure — the same way you install a VS Code extension or a phone app. Someone else already built the MCP server. You connect to it.

There is no coding involved. If the MCP server exists, you configure it once and Claude can use it in every session.

Examples of What MCP Enables

Without MCP (Level 2)	With MCP (Level 3+)
Export PPC bulk file, save to /reports, ask Claude to analyze it	Ask Claude “What are my worst-performing keywords this week?” — Claude queries live ad data directly
Export inventory report, save CSV, ask Claude to check reorder needs	Claude checks live inventory levels on a schedule and alerts you automatically
Manually pull order data and paste into a summary request	Claude generates a weekly order digest from live data without any manual steps
Check Asana manually to see what tasks are due	Ask Claude “What is on my plate this week?” — it queries Asana directly
Open each customer email, look up the order, write a reply from scratch	Claude reads the inbox, looks up the order via Amazon MCP, and saves a ready-to-send draft — you just review and click send

MCPs Relevant to Amazon Sellers

Here are the most useful MCP connections for Amazon seller workflows:

Seller Labs Amazon MCP (most important)

Gives Claude direct access to your Amazon Seller Central data — orders, inventory, advertising performance, profit data, and more — via the Seller Labs Data Hub. This is the primary data connection for every Amazon workflow in Course 2.

Learn more and get access at sellerlabs.com/amazon-mcp

QuickBooks MCP

Connects Claude to your QuickBooks accounting data. Useful for generating P&L summaries, categorizing expenses, and pulling financial context into AI workflows.

Google Calendar MCP

Lets Claude check your calendar — useful for scheduling-aware workflows and briefings.

Asana / Jira MCP

Connects Claude to your project management system. Useful for task-aware workflows: “What is on my Asana this week?” or “Update the task status after this analysis is done.”

Gmail / Email MCP

Connects Claude to your Gmail inbox. This one is particularly powerful for Amazon sellers because customer email — buyer inquiries, Amazon case correspondence, supplier communications — is a high-volume, time-consuming part of the job that AI handles well.

With an email MCP connected, Claude can:

• Read your inbox — scan for customer inquiries, flag priority messages, summarize threads you have not had time to read
• Classify incoming email — distinguish buyer questions, refund requests, supplier updates, and Amazon case notifications automatically
• Research before responding — when a buyer asks about order status, Claude checks the order data via your Amazon MCP, then drafts a specific, accurate reply — not a generic template
• Draft responses — write replies that match your brand voice from CLAUDE.md, with the relevant order or product details already filled in
• Triage on a schedule — run every few hours, flag anything urgent, summarize everything else so you can process it in one focused session

Example workflow: Every 3 hours, Claude scans your inbox for new buyer emails. For each one, it looks up the relevant order in your Amazon MCP, drafts a reply with the actual order details, and saves the draft for your review. You open Gmail once, review 5 ready-to-send drafts instead of researching and writing each response from scratch.

Security note: Email is user-generated content from sources you do not control — which means it carries prompt injection risk (covered in Module 5b). Always configure email workflows to draft for your review, never to send autonomously. The draft-and-review pattern gives you the time savings without the risk of Claude sending something unintended.

Google Workspace MCP and Gmail MCP are both available for Claude Code. Google Workspace MCP covers Gmail, Google Calendar, and Google Drive in one connection.

How to Add an MCP to VS Code

Each MCP is configured in a file called .mcp.json or inside VS Code’s settings. The exact steps vary by MCP, but the general process is:

1. Get the MCP connection details from the provider (Seller Labs, Anthropic, etc.)
2. Add the MCP configuration to your VS Code Claude Code settings
3. Authenticate (usually via an API key or OAuth login)
4. Restart Claude Code
5. Test by asking Claude a question that requires the connected tool

For the Seller Labs Amazon MCP specifically, the setup process is covered in Module 4 with a dedicated walkthrough video.

Further Learning

For a deeper understanding of how MCP works technically — including how to evaluate and compare different MCP servers — Anthropic’s free course is the best resource:

Introduction to Model Context Protocol (Anthropic Academy) — covers MCP architecture, available MCPs, and how to connect them to Claude Code.