Overview

What is Kubeflow MCP Server

The Kubeflow MCP Server exposes Kubeflow Trainer operations as Model Context Protocol (MCP) tools, built on top of the Kubeflow SDK’s TrainerClient. It lets AI agents (Claude, Cursor, Claude Code, or any custom agent) plan, submit, monitor, and manage training jobs conversationally — without users needing to learn Kubernetes or the Kubeflow SDK directly.

It is developed and released as a separate project from the SDK — see KEP-936 for the design proposal.

Key Benefits

• Agent-Native — Tools are auto-discovered via MCP, with no manual API wiring
• Guided Workflow — Phase ordering with next-step hints (Plan → Discover → Train → Monitor)
• Preview-Before-Submit — Every mutating operation requires explicit confirmation
• Security-First — Persona gating, namespace enforcement, input validation, bearer/JWT auth
• Multi-Platform — Auto-detects OpenShift, EKS, and GKE with platform-specific guidance
• Token-Efficient — Progressive/semantic modes compress 23 tools into 2–3 meta-tools
• Extensible — Plugin architecture for additional Kubeflow clients (optimizer and hub planned)

How It Works

1. Connect — The server loads Kubeflow SDK clients (trainer, with optimizer and hub planned) and registers their operations as MCP tools
2. Filter by persona — The active persona (readonly, data-scientist, ml-engineer, platform-admin) determines which tools are visible to the caller
3. Preview, then confirm — Mutating tools return a preview when called without confirmed=True, and only apply the change on a confirmed follow-up call
4. Guide by phase — Tool responses include next-step hints so agents follow the Plan → Discover → Train → Monitor workflow instead of guessing what to call next

Quick Start

Install the MCP server:

pip install kubeflow-mcp

Start the server (defaults to stdio transport):

kubeflow-mcp serve

For Claude Code, register it directly:

claude mcp add kubeflow -- kubeflow-mcp serve

For the HTTP transport (e.g. Docker image):

kubeflow-mcp serve --transport http --auth-token my-secret-token

Or run the pre-built multi-arch image published to GHCR:

docker run --rm -p 8000:8000 \
  -e KUBEFLOW_MCP_AUTH_TOKEN=my-secret-token \
  ghcr.io/kubeflow/mcp-server:latest

Example: Fine-Tune a Model via AI Agent

Once connected, your AI agent can run a complete training workflow through natural language:

User: "Fine-tune Llama 3.2 1B on the alpaca dataset"

Agent calls: check_compatibility()        → ✅ K8s 1.29, Trainer CRD installed
Agent calls: get_cluster_resources()      → 4x A100 GPUs available
Agent calls: estimate_resources(...)      → needs ~8GB GPU, 1x A100
Agent calls: list_runtimes()              → torchtune-llama3.2-1b, ...
Agent calls: fine_tune(                   → preview config (confirmed=False)
    model="hf://meta-llama/Llama-3.2-1B",
    dataset="hf://tatsu-lab/alpaca",
    runtime="torchtune-llama3.2-1b"
)
Agent calls: fine_tune(..., confirmed=True) → TrainJob "train-llama-abc" created
Agent calls: get_training_logs(...)        → training progress...

Every mutating tool requires confirmed=True — agents always preview before submitting.

Tool Catalog

The MCP Server organizes tools by workflow phase:

Next Steps

• Read the full documentation on the SDK website for configuration, authentication, and tool discovery modes
• Watch the demo walkthrough (OSS India)
• Visit the GitHub repository for source code, roadmap, and contributing guidelines