WindyFlo – Zapier MCP Server Integration Guide
This document explains how to create a Zapier Model Context Protocol (MCP) server, retrieve the server-specific URL, and integrate it with the Custom MCP node in WindyFlo to perform various external automation actions.
Step 1: Create a Zapier MCP Server
Go to the Zapier MCP Portal and log in.
Click the + New MCP Server button.
In the MCP Client dropdown, select Other.
Enter a name in the Name field for your MCP server.
Click Create MCP Server to finalize the setup.
Step 2: Add Tools (Actions)
In the left sidebar, select the newly created MCP server.
Navigate to the Configure tab.
Click the + Add tool button.
Search and select the app you want to use (e.g., Gmail, Google Sheets).
Choose the action (e.g., Send Email, Create Spreadsheet Row, etc.).
For each required field, choose how it should be populated:
Manually set by the user
Automatically selected by AI (recommended)
From a predefined list of values
Use a default value
If required, connect your account (OAuth/API key).
Click Save to register the tool.
💡 Tip: You can add multiple tools to a single MCP server and execute them based on incoming data.
Step 3: Copy the MCP Server URL
Go to the Connect tab.
Find the field labeled Connect with server-specific URL.
Click Copy URL to copy the endpoint.
Example:
⚠️ Caution: This is a unique server endpoint. Treat it as sensitive information—do not expose it publicly.
Step 4: Configure the MCP Node in WindyFlo (Actual Usage)
In WindyFlo, add a Custom MCP node to your canvas.
In the MCP Server Config field, input the copied URL in JSON format as shown below:
Replace the value with the actual URL copied from Zapier.
Make sure it follows correct JSON syntax:
"url": "..."
In the Available Actions dropdown, click the expand icon (∨) to load available tools you previously configured in Zapier.
Example actions include:
GMAIL_SEND_EMAIL– Send an emailGMAIL_CREATE_DRAFT– Create an email draftGMAIL_ARCHIVE_EMAIL– Archive a messageGMAIL_ADD_LABEL_TO_EMAIL– Add a label to an email
Select one or more actions to define the operations this node will trigger.
✅ You can also link a preceding LLM node (e.g., GPT) to dynamically generate JSON payloads for execution based on input text.
If you'd like a pipeline template or pre-configured demo, let me know. I’d be happy to help set it up for your use case.
Last updated