Back to MCP Servers

Make MCP Server

Trigger Make automation scenarios with parameters from an AI assistant and receive structured JSON output back.

Automation by Make (Integromat) API Key active
Overview

The Make MCP Server connects an MCP client (such as Claude Desktop) to your Make account and exposes every scenario configured with on-demand scheduling as a callable tool. The server parses each scenario's input parameters and descriptions so the LLM can pick the right scenario, fill in arguments, run it, and consume the structured JSON the scenario returns. This lets you turn any Make workflow, across 2,000+ Make app integrations, into a tool an AI agent can call.

This repository (integromat/make-mcp-server) is the official, open-source legacy implementation maintained by Make. It is a Node.js stdio server distributed as the npm package @makehq/mcp-server. The README notes that Make now also offers a cloud-hosted MCP server at https://mcp.make.com with OAuth authentication and broader management tools, which Make recommends for most new use cases. The self-hosted version remains useful when you need to run the bridge locally or only want scenario-run capabilities.

Authentication uses a Make API key with the scenarios:read and scenarios:run scopes, plus a zone (e.g. eu2.make.com) and a team ID. Scenarios marked as on-demand are auto-discovered and surfaced as tools; their input/output schemas are taken from the scenario's interface definitions.

Tools

Tool Description
run_scenario (dynamic per scenario) Each Make scenario configured with on-demand scheduling is registered as its own tool. The tool name, description, and parameter schema are derived from the scenario's name, description, and Input/Output interface definitions in Make.
Setup Guide

Prerequisites

  • Node.js installed locally
  • An MCP-compatible client (e.g. Claude Desktop)
  • A Make API key with scopes scenarios:read and scenarios:run (create one in Make profile settings)
  • Your Make zone (e.g. eu2.make.com, us1.make.com) and your Team ID (from the Team page URL)
  • One or more Make scenarios with scheduling set to On-Demand, with Input and Output interfaces defined

Configuration

Add the server to claude_desktop_config.json (or your MCP client's equivalent):

{
  "mcpServers": {
    "make": {
      "command": "npx",
      "args": ["-y", "@makehq/mcp-server"],
      "env": {
        "MAKE_API_KEY": "<your-api-key>",
        "MAKE_ZONE": "<your-zone>",
        "MAKE_TEAM": "<your-team-id>"
      }
    }
  }
}

Restart your MCP client. Each on-demand scenario in the specified team will appear as a tool. Give scenarios clear names, descriptions, and parameter labels so the LLM can call them reliably.

Cloud alternative

Make also hosts a cloud MCP server at https://mcp.make.com (OAuth) which adds management tools for scenarios, connections, webhooks, and data stores. See the Make Developer Hub for details.

Use Cases
  • Expose existing Make scenarios (Slack alerts, HubSpot updates, Airtable writes, file processing) as callable tools for Claude or other MCP-aware agents.
  • Let an AI agent orchestrate multi-step business workflows across 2,000+ apps without writing custom integrations, by chaining on-demand scenarios.
  • Bridge LLM output into systems Make already connects to (CRMs, databases, email, messaging) and stream back structured JSON for further reasoning.
  • Build internal "agent + automation" tools where the LLM decides which scenario to run and Make handles the connectors, retries, and error logic.
  • Prototype agent capabilities quickly: create a scenario in Make's visual builder, mark it on-demand, and it instantly becomes an LLM tool.
Example Prompts
  • "Run the create_hubspot_contact scenario for jane@acme.com with company 'Acme Corp' and source 'webinar'."
  • "Trigger the daily revenue digest scenario and summarize the JSON it returns."
  • "Use the send_slack_alert scenario to notify #ops that the nightly ETL completed with 412 rows processed."
  • "Call the enrich_lead scenario for these 5 email addresses and tell me which ones returned a LinkedIn URL."
  • "List the Make scenarios available to you and describe what each one does."
Pros
  • Official Make-maintained server (Integromat org) with a published npm package, @makehq/mcp-server.
  • Turns any Make scenario into an LLM tool automatically using the scenario's existing Input/Output interface, no extra schema work.
  • Inherits Make's 2,000+ app connectors, retries, error handling, and execution history.
  • Open-source (MIT) so you can self-host and inspect the bridge code.
Limitations
  • Officially marked as the legacy version; Make recommends the cloud-hosted server at https://mcp.make.com for most new deployments.
  • Only surfaces scenarios with on-demand scheduling; scheduled or webhook-only scenarios are not exposed.
  • No built-in scenario management (create, edit, enable) in this self-hosted version; those tools live in the cloud server and require a paid plan.
  • Tool quality depends heavily on the descriptions you write inside Make; sparse scenario metadata leads to poor LLM tool selection.
Alternatives
  • Make Cloud MCP Server (https://mcp.make.com), the OAuth cloud-hosted successor from Make itself with management tools.
  • Zapier MCP for similar "expose automation workflows as tools" coverage across Zapier's connector library.
  • n8n MCP server integrations for self-hosted, open-source workflow automation as MCP tools.