What is this?
The Zapier MCP Server is a purpose-built connector that lets your AI agents and large language models orchestrate workflows across over 6,000 popular SaaS apps via Zapier’s automation platform. By exposing Zapier workflows, triggers, and actions through the Model Context Protocol, it removes the need to build custom integrations for each service.
Under the hood, it implements the MCP 1.0 protocol over HTTP/REST, translating tool calls into Zapier REST API requests, managing OAuth2 credentials securely, and injecting conversational context into each action. This means your AI assistant can discover available Zaps, trigger them dynamically, and listen for real-world events—all with no additional developer overhead.
Quick Start
Install the server using npm:
npm install @modelcontextprotocol/server-zapier
Then add it to your MCP client configuration:
{
"mcpServers": {
"zapier-mcp-server": {
"command": "npx",
"args": ["-y", "@modelcontextprotocol/server-zapier"],
"env": {
"API_KEY": "your-api-key-here"
}
}
}
}
Key Features
Integration with 6,000+ Apps: Instantly tap into any service supported by Zapier—from Gmail and Slack to Salesforce and GitHub—without writing custom code.
No-Code Automation: Leverage Zapier’s visual workflow builder under the hood to trigger pre-built automations or build new ones dynamically through MCP.
Secure Credential Management: Uses Zapier’s encrypted vault and MCP-level ACLs to store tokens and enforce scoped access for agents.
Example Usage
For example, you can trigger a Slack notification directly from your AI assistant:
// Example code
const result = await client.callTool({
name: "triggerZap",
arguments: {
zapId: "slack_send_message",
data: { channel: "#general", text: "Hello from AI Assistant!" }
}
});
This call executes the specified Zap and returns a run ID and status to confirm the message delivery.
Configuration
The server accepts the following environment variables:
API_KEY – Your Zapier API key or OAuth client credentials obtained from the Zapier Developer Platform.
MCP_SERVER_PORT (optional) – Port number for the MCP HTTP server (default: 4000).
Available Tools/Resources
List Resources: Discover available Zaps and inspect their input/output schemas.
triggerZap: Execute a specific Zap run by its ID with provided parameters.
pollResource: Poll or subscribe to trigger events (e.g., new email) for real-time processing.
getPrompts: Fetch system prompts that guide LLMs on valid tool invocations and error handling.
Who Should Use This?
This server is perfect for:
Use case 1: AI developers who need to automate SaaS workflows without building custom connectors.
Use case 2: Product teams prototyping conversation-driven automations quickly.
Use case 3: Enterprises seeking secure, scalable orchestration across multiple applications.
Conclusion
With the Zapier MCP Server, you can empower your AI agents to orchestrate complex workflows across thousands of apps in minutes—no custom integration work required. Try it today to streamline your automations and focus on building smarter AI experiences.
Check out the GitHub repository for more information and to contribute.