zapier
Connect your AI to thousands of apps with the Model Context Protocol.
claude mcp add --transport stdio zapier-zapier-mcp npx -y zapier-mcp \ --env ZAPIER_API_KEY="Your API key for Zapier MCP (optional if using OAuth)" \ --env ZAPIER_OAUTH_CLIENT_ID="OAuth client ID (optional)"
How to use
Zapier MCP acts as a remote MCP server that exposes Zapier’s vast library of app integrations to your AI clients. Once running, you can enable specific actions (like sending emails, creating tasks, or posting messages) in the Zapier MCP web console, and those actions become callable tools your AI can invoke with natural language prompts. The server provides a meta-tool get_configuration_url to let you discover and manage available actions, and each enabled action appears as a dedicated tool (for example, gmail_send_email) that your AI client can call with the required parameters. This setup lets your AI orchestrate thousands of apps without writing bespoke integrations for each service.
To get started, generate credentials for your server (API Key or OAuth), point your AI client at your Zapier MCP instance, and then browse the actions in the Zapier MCP portal to enable the tools you want your AI to use. The more actions you enable, the more capable your AI becomes at performing end-to-end tasks across apps like Slack, Google Sheets, Gmail, Asana, GitHub, and HubSpot, all through natural language commands.
How to install
Prerequisites:
- Node.js (LTS version recommended) and npm installed on your system
- Internet access to fetch the MCP server package
Option A: Run with npx (recommended for quick starts)
- Ensure Node.js and npm are installed:
- node -v
- npm -v
- Start the Zapier MCP server using npx: npm_config_user_agent="" npx -y zapier-mcp
- The server should boot and provide a URL or endpoint to connect your MCP clients. Follow the on-screen prompts in the Zapier MCP portal to configure credentials and actions.
Option B: Install via npm (local/project scope)
- Initialize your project (if needed): mkdir zapier-mcp-app && cd zapier-mcp-app npm init -y
- Install the Zapier MCP package locally: npm install zapier-mcp
- Run the server from your project script (example in package.json): "scripts": { "start": "node node_modules/zapier-mcp/dist/server.js" } npm run start
Option C: Docker (alternative)
- If a Docker image is available, you can run:
docker run -i <image-name>
-e ZAPIER_API_KEY=your_api_key \ -p 8080:8080 - Access the MCP server at http://localhost:8080 (or the mapped port) and configure actions via the portal.
Additional notes
Tips and common concerns:
- Security: Use API Key or OAuth as recommended by Zapier MCP to protect access to actions.
- Action discovery: Use the built-in get_configuration_url tool to list available actions and their parameters before enabling them.
- Environment variables: If you operate behind a proxy or require specific endpoints, configure relevant env vars (e.g., HTTP_PROXY, HTTPS_PROXY).
- Resource limits: For large action sets, consider enabling a subset of core actions first to validate your workflow before scaling.
- Updates: Keep the zapier-mcp package up to date to receive new actions and improvements; periodically re-browse the action catalog in the portal.
- Debugging: If actions aren’t appearing, check credentials, ensure the server is reachable by your client, and verify that the required scopes/permissions are granted for OAuth or API Key.
Related MCP Servers
Wax
Sub-Millisecond RAG on Apple Silicon. No Server. No API. One File. Pure Swift
wanaku
Wanaku MCP Router
mcp
Octopus Deploy Official MCP Server
mem0
✨ mem0 MCP Server: A memory system using mem0 for AI applications with model context protocl (MCP) integration. Enables long-term memory for AI agents as a drop-in MCP server.
furi
CLI & API for MCP management
Pare
Dev tools, optimized for agents. Structured, token-efficient MCP servers for git, test runners, npm, Docker, and more.