Make MCP Server
# Make MCP Server (legacy)
**A modern, cloud-based version of the Make MCP Server is now available. For most use cases, we recommend using [this new version](https://developers.make.com/mcp-server).**
A Model Context Protocol server that enables Make scenarios to be utilized as tools by AI assistants. This integration allows AI systems to trigger and interact with your Make automation workflows.
## How It Works
The MCP server:
- Connects to your Make account and identifies all scenarios configured with "On-Demand" scheduling
- Parses and resolves input parameters for each scenario, providing AI assistants with meaningful parameter descriptions
- Allows AI assistants to invoke scenarios with appropriate parameters
- Returns scenario output as structured JSON, enabling AI assistants to properly interpret the results
## Benefits
- Turn your Make scenarios into callable tools for AI assistants
- Maintain complex automation logic in Make while exposing functionality to AI systems
- Create bidirectional communication between your AI assistants and your existing automation workflows
## Usage with Claude Desktop
### Prerequisites
- NodeJS
- MCP Client (like Claude Desktop App)
- Make API Key with `scenarios:read` and `scenarios:run` scopes
### Installation
To use this server with the Claude Desktop app, add the following configuration to the "mcpServers" section of your `claude_desktop_config.json`:
```json
{
"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>"
}
}
}
}
```
- `MAKE_API_KEY` - You can generate an API key in your Make profile.
- `MAKE_ZONE` - The zone your organization is hosted in (e.g., `eu2.make.com`).
- `MAKE_TEAM` - You can find the ID in the URL of the Team page.
TDQS
Scored across 6 tools
The tool names are entirely opaque, consisting only of scenario IDs like 'run_scenario_11422' with no indication of their distinct purposes. From the descriptions, some tools appear to handle inventory operations while others deal with scenario inputs or testing, but the naming provides no disambiguation, making it impossible for an agent to reliably choose between them without guessing.
All tool names follow a perfectly consistent pattern: 'run_scenario_' followed by a numeric ID. While this pattern is uninformative, it is uniformly applied across all six tools, with no deviations in style or structure.
With 6 tools, the count is reasonable and not excessive. However, given the unclear domain inferred from the tool names and descriptions—which suggest a mix of inventory management, scenario testing, and input handling—it's borderline whether this number adequately covers the scope, as the tools seem fragmented rather than cohesive.
Inferred domain includes inventory management and scenario testing, but there are significant gaps. For inventory, only list and add operations are present, missing update and delete. For scenario inputs and testing, the tools are vague and incomplete, lacking clear CRUD or lifecycle coverage, which will likely cause agent failures in complex tasks.