optuna-mcp

Overview

The Optuna MCP Server provides a comprehensive interface to Optuna's optimization framework through the Model Context Protocol. It enables automated hyperparameter tuning, interactive analysis of optimization results, and integration with other MCP tools. The server offers a rich set of tools for study management, trial execution, and visualization capabilities.

Try asking AI

After installing, here are 3 things you can ask your AI assistant:

• create_studycreate_study(study_name: string, directions: list[string])

  Create a new Optuna study with the given study_name and directions.

  Try:Create an Optuna study named 'my-study' for minimizationTry:Create a new study named 'optimize-hyperparameters' with directions minimize and maximize
• get_all_study_names

  Get all study names from the storage.

  Try:List all available study namesTry:Show me all the studies
• askask(search_space: dictionary)

  Suggest new parameters using Optuna.

  Try:Suggest parameters for x in [0,1] and y in [0,10]Try:Get new trial parameters for learning_rate in [0.001,0.1] and batch_size in [16,256]
• telltell(trial_number: integer, values: float | list[float])

  Report the result of a trial.

  Try:Report trial 1 result: 0.85Try:Tell trial 5 the values are 0.75 and 0.92
• plot_optimization_historyplot_optimization_history(target: integer, target_name: string)

  Return the optimization history plot as an image.

  Try:Plot the optimization historyTry:Show me the optimization history plot for target 0
• plot_pareto_frontplot_pareto_front(target_names: list[string], include_dominated_trials: boolean, targets: list[integer])

  Return the Pareto front plot as an image for multi-objective optimization.

  Try:Plot the Pareto frontTry:Show me the Pareto front plot for objectives 'accuracy' and 'time'
• plot_contourplot_contour(params: list[string], target: integer, target_name: string)

  Return the contour plot as an image.

  Try:Create a contour plot for parameters x and yTry:Plot a contour visualization showing the effect of learning_rate and batch_size
• plot_param_importancesplot_param_importances(params: list[string], target: integer, target_name: string)

  Return the parameter importances plot as an image.

  Try:Show me parameter importancesTry:Plot which parameters are most important for the objective
• best_trial

  Get the best trial.

  Try:Show me the best trialTry:What are the parameters and value of the best trial?
• set_trial_user_attrset_trial_user_attr(trial_number: integer, key: string, value: any)

  Set user attributes for a trial.

  Try:Set trial 3 user attribute model_type to 'resnet'Try:Add comment to trial 5: 'initial test with high learning rate'
• launch_optuna_dashboardlaunch_optuna_dashboard(port: integer)

  Launch the Optuna dashboard.

  Try:Start the Optuna dashboardTry:Launch the dashboard on port 58081
• plot_sliceplot_slice(params: list[string], target: integer, target_name: string)

  Return the slice plot as an image.

  Try:Create a slice plot for parameters x and yTry:Show me slice plots for hyperparameters

Comparable tools

Installation

{
  "mcpServers": {
    "Optuna": {
      "command": "/path/to/uvx",
      "args": [
        "optuna-mcp"
      ]
    }
  }
}

{
  "mcpServers": {
    "Optuna": {
      "command": "docker",
      "args": [
        "run",
        "-i",
        "--rm",
        "--net=host",
        "-v",
        "/PATH/TO/LOCAL/DIRECTORY/WHICH/INCLUDES/DB/FILE:/app/workspace",
        "ghcr.io/optuna/optuna-mcp:latest",
        "--storage",
        "sqlite:////app/workspace/optuna.db"
      ]
    }
  }
}

Compare optuna-mcp with