MCP Zenml

This project implements a server based on the Model Context Protocol (MCP) standard to connect the ZenML platform with AI applications. It provides access to core ZenML functions through a standardized protocol, including querying information about users, pipelines, run records, etc., and supports triggering new pipeline runs.

Developer tools Artificial intelligence chatbots #AI Connector #ZenML Interface #Data Protocol .Python

rating : 2.5 points

downloads : 27

update time : 2025-04-29

What is ZenML MCP Server?

The ZenML MCP Server acts as a bridge between AI applications and your ZenML platform, allowing AI models to query information about your ML pipelines, stacks, runs and other ZenML resources through a standardized protocol.

How does it work?

The server implements the Model Context Protocol (MCP) standard which works like a 'USB - C port for AI applications'. AI clients like Claude or Cursor can connect to your ZenML data through this standardized interface.

When to use it?

Use this when you want AI assistants to help analyze your ML pipeline status, investigate run results, or trigger new pipeline executions through natural language queries.

Key Features

ZenML Data AccessRead access to pipelines, runs, stacks, artifacts and other ZenML resources

Pipeline ControlTrigger new pipeline runs from existing templates

Log AccessView step execution logs for cloud - based runs

Standard ProtocolUses MCP standard for consistent AI application integration

Pros and Cons

Advantages

Provides AI assistants with real - time access to your ML pipeline data

Standardized interface works with multiple AI clients

No need to build custom integrations for each AI tool

Secure access through ZenML API keys

Limitations

Currently in beta - some features may change

Requires ZenML Cloud account

Read - only for most resources (except pipeline triggering)

Artifact metadata only (not actual data files)

Setup Guide

Install prerequisites

Install uv package manager and clone the repository

Create MCP config

Create a JSON configuration file with your ZenML server details and API key

Configure your AI client

Add the config file to your preferred AI client (Claude Desktop or Cursor)

Example Use Cases

Pipeline AnalysisAsk your AI assistant to analyze recent pipeline runs

Triggering RunsStart new pipeline executions through natural language

Frequently Asked Questions

Do I need ZenML Cloud to use this?

Which AI clients support MCP?

Is my data secure?

Additional Resources

MCP Protocol Documentation

Learn about the Model Context Protocol standard

ZenML Website

Official ZenML platform website

ZenML Cloud

🚀 ZenML's MCP Server

This project implements a Model Context Protocol (MCP) server for interacting with the ZenML API.

ZenML MCP Server

🚀 Quick Start

Project Introduction

This project aims to integrate with the ZenML platform through the MCP protocol. MCP is a protocol for connecting different tools and services, while ZenML is a machine learning operations platform.

MCP Protocol Explanation

MCP (Model Context Protocol) is an open standard that allows applications and tools to share model-related information. Through MCP, the ZenML server can interact with other tools and services that support this protocol.

ZenML Platform Introduction

ZenML is an open-source machine learning operations platform that provides a unified workflow engine, experiment tracking, model management, and other functions. It helps data scientists build and deploy machine learning models more efficiently.

Usage Guide

Prerequisites

Install Python 3.8 or a higher version.
Install the pip package management tool.
Ensure that the uv tool is installed on the system for running the server.

Configuration File Creation

You need to create a configuration file mcp.json and fill in the following content (modify according to the actual situation):

{
    "MCP Server": {
        "name": "ZenML MCP Server",
        "description": "An MCP server integrated with the ZenML platform.",
        "server_url": "http://localhost:8000",
        "auth_token": "your_auth_token_here"
    }
}

Installation and Running

Use uv to run the server:
```
uv mcp_server.py
```
After the server starts, you can access http://localhost:8000 in your browser to check the service status.

Integration Guide

Integration with Claude Desktop

Install and open Claude Desktop.
Go to the settings menu and select the "Developer" tab.
Click "Edit Configuration" to find or create the mcp.json file.
Fill in the above content in the file and save it.
Restart Claude Desktop.

Integration with Cursor

Create a .cursor folder in the project root directory.
Create a mcp.json file in this folder and fill in the relevant content.
Open the Cursor settings and enable the ZenML server configuration.

Optional Optimization Suggestions

To improve the readability of the output, you can add the following preference in the personal settings of Claude Desktop:

When using the ZenML tool and returning JSON results, consider presenting the results in a more readable format, such as a table.

This ensures that the output information is clearer and more understandable.

Summary

By integrating with the ZenML platform through the MCP protocol, you can conveniently use ZenML's functions in tools such as Claude Desktop and Cursor. After completing the configuration according to the above steps, you can directly call the ZenML tool in the chat window for model management and operations.