mlflowAgent

Name: mlflowAgent
Author: iRahulPandey

@iRahulPandey

Repository

About mlflowAgent

mcp server for mlflow

#mlflow #mlflow-agent #mcp-server

Basic information

Config

Add this server to your MCP-compatible client using the configuration below.

{
  "mcpServers": {
    "mlflowAgent": {
      "command": "npx",
      "args": [
        "-y",
        "@smithery/cli",
        "install",
        "@iRahulPandey/mlflowMCPServer",
        "--client",
        "claude"
      ]
    }
  }
}

Tools

No tools detected

We auto-extract tools from the README. The maintainer can list them under a ## Tools heading to populate this section.

Overview

What is mlflowAgent?

mlflowAgent is a natural language interface for MLflow built on the Model Context Protocol (MCP). It connects to your MLflow tracking server and lets you query experiments, models, and system information using plain English. It consists of an MCP server (mlflow_server.py) that exposes MLflow functionality via MCP and an MCP client (mlflow_client.py) that uses an LLM (OpenAI) to translate natural language queries into tool calls.

How to use mlflowAgent?

Install via Smithery (npx -y @smithery/cli install @iRahulPandey/mlflowMCPServer --client claude) or manually (clone repo, create venv, install dependencies). Set OPENAI_API_KEY and optionally MLFLOW_TRACKING_URI. Start the MCP server with python mlflow_server.py, then send queries like python mlflow_client.py "Show me all registered models in MLflow".

Key features of mlflowAgent

Natural language queries to your MLflow tracking server
List and explore experiments and runs
Get details about registered models in the registry
Retrieve MLflow system status and metadata

Use cases of mlflowAgent

Ask "What models do I have registered?" without remembering MLflow CLI commands
Quickly list all experiments from a conversation interface
Get model details (e.g. for "iris-classifier") by typing a simple question
Check whether your MLflow server is running and responsive

FAQ from mlflowAgent

What are the prerequisites for running mlflowAgent?

Python 3.8+, an MLflow tracking server (default http://localhost:8080), and an OpenAI API key for the LLM component.

How do I configure the MLflow tracking server URI?

Set the environment variable MLFLOW_TRACKING_URI (default: http://localhost:8080). You can also set MODEL_NAME, MLFLOW_SERVER_SCRIPT, and LOG_LEVEL via environment variables.

What MLflow operations does mlflowAgent support?

It supports listing models (list_models), listing experiments (list_experiments), getting model details (get_model_details), and getting system info (get_system_info). It currently covers a subset of MLflow functionality.

Does mlflowAgent require internet access?

Yes, the client uses OpenAI models, so internet access is needed for the natural language processing component.

Are there any known limitations?

The tool supports only a subset of MLflow features. Error handling may be limited for complex operations. Future improvements may add model predictions, visualization, and run management.