zapier mcp setup

How to Set Up Zapier for MCP Integration

As AI agents move beyond answering questions and start taking real actions, sending emails, updating spreadsheets, creating tasks, or scheduling events, they need a reliable way to interact with external tools. This is where automation platforms like Zapier become an important part of agent architectures. Zapier allows you to connect thousands of applications through a consistent interface. MCP (Model Context Protocol) provides a standardized way for AI systems to discover and invoke those capabilities safely. Together, they form a practical bridge between language models and real-world software systems.

Before setting up a Zapier MCP server, it helps to understand what Zapier provides, what MCP is designed to solve, and how the two fit together in a typical agent workflow. This context makes the setup steps—and the design decisions behind them—much easier to reason about.

If you are already familiar with Zapier and MCP, you can skip ahead to the step-by-step tutorial. Otherwise, the next sections establish the conceptual foundation.

What Is Zapier?
What Is MCP, in Practical Terms?
MCP Clients and MCP Servers
Setting Up Zapier for MCP Integration (Step-by-Step Tutorial)
What Comes Next

What Is Zapier?

Zapier is an automation platform that connects applications and services without requiring custom integrations for each one. At a high level, Zapier allows you to trigger actions in one application based on events in another, automate repetitive workflows across different tools, and interact with SaaS products through standardized actions, enabling integrations without writing custom backend logic for each connection.

Zapier supports thousands of applications, including email, calendars, spreadsheets, project management tools, databases, and collaboration platforms. Instead of writing and maintaining individual API integrations, developers and teams can rely on Zapier’s pre-built connectors.

In the context of AI systems, Zapier acts as a universal action layer. Rather than teaching an agent how to speak to Gmail, Google Sheets, Notion, and Slack individually, you expose those capabilities through a single, consistent interface.

What Is MCP, in Practical Terms?

Model Context Protocol Before vs. After

MCP (Model Context Protocol) is a standard that allows AI systems to interact with external tools in a structured, discoverable, and controlled way. It separates responsibilities into two roles:

MCP Server: Exposes tools and actions
MCP Client: Connects to the server and invokes those tools

Instead of hardcoding tool logic directly into an application, MCP enables an AI system to dynamically discover and query its environment by asking what tools are available, what actions those tools can perform, and what inputs or parameters they require, making integrations more flexible and extensible. This makes tool usage more dynamic, modular, and easier to evolve over time.

MCP Clients and MCP Servers

MCP Client & Server

An MCP server is responsible for defining and exposing capabilities. In this setup, Zapier acts as the MCP server by exposing actions like “send email,” “create spreadsheet row,” or “create calendar event.”

An MCP client is the application that connects to an MCP server and invokes the tools it exposes. This client can take many forms, including Python applications, AI chatbots, developer tools, or IDE integrations—any system that needs to discover, call, and coordinate external capabilities through a standardized interface.

Zapier currently supports multiple MCP clients, such as:

Claude (Claude Code)
ChatGPT
Cursor
Custom Python-based agents

In this guide, we use a Python MCP client, but the Zapier-side setup is largely the same regardless of which client you use. The key idea is that the MCP server (Zapier) remains stable, while clients can change.

Setting Up Zapier for MCP Integration (Step-by-Step Tutorial)

With the conceptual foundation in place, we can move into the practical setup. In this section, you will create a Zapier MCP server, configure authentication, and add tools that can be used by an MCP-compatible client. The goal is to create a reusable, general-purpose MCP server that can support a wide range of automations through Zapier.

Prerequisites

Before you begin, make sure you have:

A Zapier account (a free plan is sufficient to get started). You can sign up on Zapier.

Step 1: Create an MCP Server in Zapier

1. Open the Zapier MCP Portal

Navigate to Zapier MCP.

Sign in using your Zapier account credentials. This portal is where all MCP-related configuration happens.

2. Create a New MCP Server

From the MCP dashboard:

Click + New MCP Server to begin creating a new MCP server that will expose Zapier actions to external AI clients

For MCP Client, select Python, which configures the server for Python-based agents and applications and determines the connection examples shown by Zapier

Once these options are selected, Zapier creates the MCP server and redirects you to its configuration page, where you can generate credentials, add tools, and manage how MCP clients connect and authenticate.

3. Generate Credentials and Connection Details

On the server configuration page:

Open the Connect tab to access connection details and authentication settings for this MCP server

Click Generate an API key to create a credential that allows MCP clients to authenticate and make authorized requests to the server

Once you do that, you will be prompted to Generate a connection token. This token contains the secret that will authenticate requests to your server.

Once the token is generated, copy and save it somewhere safe before closing the dialog box.

Scroll down and copy the server URL shown in the example code (This is important). Make sure you copy the server URL exactly as shown. This URL is what MCP clients will use to connect to Zapier.

4. Store Your Credentials Securely

At this point, you should have the MCP server URL and the Zapier MCP API key. Treat these credentials like any other sensitive secret. Do not commit them to source control or expose them publicly.

Step 2: Add Tools to Your MCP Server

Once the server is created, the next step is to define what actions it can perform.

1. Start Adding Tools

Go back to the Configure tab in your MCP server dashboard, where tools and server capabilities are managed

Click + Add tool to begin defining which actions and integrations this MCP server will expose to connected clients

This opens a dialog where you define what capabilities the MCP server exposes.

2. Choose an Application

Use the search bar to find any supported Zapier app. Few examples are:

Gmail: Email automation
Google Calendar: Schedule management
Google Sheets: Spreadsheet operations
Google Docs: Document creation
Google Forms: Form management
Notion: Notes and databases
Google Drive: File management
Airtable: Database operations

Each app represents a category of actions your AI system can perform. In this example, we are using Google Sheets.

3. Select Actions for the App

After selecting an app (for example, Google Sheets here), you will see available action categories:

Find data: For example, in Google Sheets App 'Lookup Spreadsheet Rows' is a find data tool
Take action: 'Create Spreadsheet Column' is an example of a take action tool in Google Sheets
Or you can choose to scroll though 'All' the options

For quick setup, you can choose 'All', which exposes every supported action for that app. This is often the easiest option when experimenting or building general-purpose systems. Alternatively, you can select individual actions if you want tighter control. Once you have selected your required tools, you can click the 'Connect' button.

4. Connect Your Account

If the selected app requires authentication, Zapier will prompt you to connect an account before proceeding. You can either choose from an existing connected account or initiate a new connection through the OAuth flow, which securely authorizes Zapier to access the necessary permissions on your behalf.

Once connected, Zapier can perform actions on behalf of that account.

5. Add the Tools

After selecting actions and connecting accounts:

Confirm your selection

Zapier automatically adds the tools to your MCP server. These tools are now available to any MCP client that connects to the server.

Recommended Starter Tools

For general-purpose MCP setups, the following categories are commonly useful:

Communication

Gmail: Send Email, Find Email
Slack: Send Channel Message

Productivity

Google Sheets: Create Row, Lookup Rows
Google Calendar: Create Event, Find Event
Notion: Create Page, Update Page

Task Management

Todoist: Create Task, Find Task
Asana: Create Task, Update Task

You can always add or remove tools later as your use cases evolve.

What Comes Next

At this point, your Zapier MCP server is fully configured and ready for integration. You now have a running MCP server backed by Zapier, an authenticated server URL with its corresponding API key, and a set of real-world tools exposed through MCP that can be programmatically discovered and invoked by compatible clients.

The next steps move from setup to usage: connecting an MCP client, testing basic tool calls, and designing workflows where an AI system can reason about when and how to invoke these actions. With this foundation in place, Zapier becomes a powerful execution layer for AI-driven systems—allowing agents to move from reasoning to action without tight coupling to individual application APIs.