What do I need to connect this MCP server to my ZenML instance?

You need two environment variables: ZENML_STORE_URL (your ZenML server URL) and ZENML_STORE_API_KEY (a service account API key for authentication). Both are required to establish a connection.

Can I trigger pipeline runs through this MCP server?

Yes. The server supports triggering pipeline runs via the ZenML API. You can initiate runs and monitor their status through the chat interface.

What information can I query about my pipelines and runs?

You can browse stacks, view pipelines, inspect runs and their status, and retrieve artifacts produced by those runs. This gives you full visibility into your ML workflow history and configuration.

How do I set up authentication securely?

Use a ZenML service account API key (not a personal user key) and pass it via the ZENML_STORE_API_KEY environment variable. This follows security best practices for programmatic access.

Does this server work with self-hosted or cloud ZenML deployments?

Yes. As long as you have a valid ZENML_STORE_URL and API key, the server works with any ZenML deployment—self-hosted, cloud-hosted, or hybrid.

MCP ServerSTDIOOfficialv1.5.1

Zenml IO MCP Zenml MCP Server

Browse ZenML stacks, pipelines, runs, and artifacts; trigger pipeline runs directly from your AI chat. For ML engineers and platform teams managing ZenML deployments.

io.github.zenml-io/mcp-zenml

Try MCP with AI models View source

Hosted URL

Local install

Transport

STDIO

Auth

2 required env vars

What the Zenml IO MCP Zenml MCP server does

How models use it and what it is built for.

Browse ZenML stacks, pipelines, runs, and artifacts; trigger pipeline runs directly from your AI chat. For ML engineers and platform teams managing ZenML deployments.

Connect to Zenml IO MCP Zenml

Local install — runs as a subprocess.

docker run docker.io/zenmldocker/mcp-zenml:1.5.1

Environment variables

Configuration this server reads at startup.

ZENML_STORE_URLRequired
Base URL of your ZenML server (e.g., https://<workspace-id>-zenml.cloudinfra.zenml.io).
ZENML_STORE_API_KEYRequiredSecret
API key used to authenticate with your ZenML server (ideally a service account key).

Resources

Where to find authoritative docs and source for Zenml IO MCP Zenml.

Source repository

Example prompts for Zenml IO MCP Zenml

Paste any of these into Agent Studio after connecting Zenml IO MCP Zenml.

List all pipeline runs from the last 7 days and their status
Show me the artifacts produced by the latest run of my training pipeline
Trigger a run of the data-preprocessing pipeline with default parameters
What stacks are available in my ZenML workspace and which is active

Zenml IO MCP Zenml MCP server — FAQ

Common questions about connecting and running Zenml IO MCP Zenml.

What do I need to connect this MCP server to my ZenML instance?
You need two environment variables: ZENML_STORE_URL (your ZenML server URL) and ZENML_STORE_API_KEY (a service account API key for authentication). Both are required to establish a connection.
Can I trigger pipeline runs through this MCP server?
Yes. The server supports triggering pipeline runs via the ZenML API. You can initiate runs and monitor their status through the chat interface.
What information can I query about my pipelines and runs?
You can browse stacks, view pipelines, inspect runs and their status, and retrieve artifacts produced by those runs. This gives you full visibility into your ML workflow history and configuration.
How do I set up authentication securely?
Use a ZenML service account API key (not a personal user key) and pass it via the ZENML_STORE_API_KEY environment variable. This follows security best practices for programmatic access.
Does this server work with self-hosted or cloud ZenML deployments?
Yes. As long as you have a valid ZENML_STORE_URL and API key, the server works with any ZenML deployment—self-hosted, cloud-hosted, or hybrid.

Skip the local setup — run MCP in your browser

MCP Playground runs 10,000+ hosted MCP servers — GitHub, Linear, Notion, Stripe, Sentry and more — across Claude, GPT, Gemini, DeepSeek and 60+ AI models. Compare model answers side-by-side, save agent presets, share runs. Zero install.

Open Agent Studio