MCP Server for ZenML
An MCP server that connects an AI client such as Claude Desktop or Cursor to a running ZenML instance, giving it read access to ML and LLM pipeline definitions, runs and deployments.
The server exposes ZenML's core entities over MCP, including users, infrastructure stacks and their components, service connectors, pipeline definitions, individual pipeline runs and steps with their code and logs, schedules and artifact metadata. It also covers deployment and serving concepts such as pipeline snapshots, running deployments with their status, URL and logs, and model-serving services, plus organizational concepts such as projects, tags and builds, and the ZenML model registry with its versioned model artifacts.
Because these are read tools, an AI assistant connected through the server can answer questions about a team's ML pipelines, such as why a run failed or what a deployment's current status is, without a person switching to the ZenML dashboard. A set of compatibility endpoints supports older pipeline-run templates during migration to newer ZenML concepts.