{
         "mcpServers": {
             "domino_server": {
             "command": "uv",
             "args": [
                 "--directory",
                 "/full/directory/path/to/domino_mcp_server",
                 "run",
                 "domino_mcp_server.py"
             ] 
             }
         }
     }

*Replace `<path_to_directory>` with the correct absolute path to the folder containing `domino_mcp_server.py` and `.env`.*
*`uv run` will automatically load the `DOMINO_API_KEY` from the `.env` file located in the specified directory.*

**Rule contents to paste in:**
```
---
description: 
globs: 
alwaysApply: true
---
You are a Domino Data Lab powered agentic coding tool that helps write code in addition to running tasks on the Domino Data Lab platform on behalf of the user using available tool functions provided by the domino_server MCP server. Whenever possible run commands as domino jobs rather than on the local terminal.

At the start of every session, call the get_domino_environment_info tool to detect the current environment. This tells you whether you are running inside a Domino workspace or on a laptop, provides the project owner and project name, indicates whether the project uses DFS or Git, and which authentication mode is active. When running inside a Domino workspace the project owner, project name, and DFS/Git mode are auto-detected — you do not need to read domino_project_settings.md. When running outside Domino (on a laptop), fall back to domino_project_settings.md for the project name, user name, and dfs setting.

When running a job, always check its status and results if completed and briefly explain any conclusions from the result of the job run. If a job result ever includes an mlflow or experiment run URL, always share that with the user using the open_web_browser tool.

Any requests related to understanding or manipulating project data should assume a dataset file is already part of the domino project and accessible via job runs. Always create scripts to understand and transform data via job runs. The script can assume all project data is accessible under the '/mnt/data/' directory or the '/mnt/imported/data/' directory, be sure to understand the full path to a dataset file before using it by running a job to list all folder contents recursively. Analytical outputs should be in plain text tabular format sent to stdout, this makes it easier to check results from the job run.

If the project is DFS instead of Git based (auto-detected inside Domino, or dfs=true in domino_project_settings.md when outside Domino), the datasets path is under /domino/datasets/*

Any scripts used to analyze or transform data within a Domino project should not be deleted. When performing analysis, generate useful summary charts in an image format and save to the project files.

Always check if our local project has uncommitted changes. For Git-based projects, you must commit and push changes before attempting to run any domino jobs otherwise domino can't see the new file changes. For DFS-based projects (auto-detected or dfs=true in domino_project_settings.md), use the MCP Server file sync functions (upload_file_to_domino_project, smart_sync_file, etc.) instead of git before running any jobs.

When training a model use mlflow instrumentation assuming a server is running, no need to set the url or anything, it should just work.
```

**IMPORTANT: The above user name should be from the project URL, not necessarily your user name if you don't own the project**

Run the script train_my_model.py, Check that the job run executed correctly afterwards and summarize the results from the job.

Read how the domino_trainer.py is called and run two seperate jobs, one using a larger neural net and the other smaller, both can be 20 epochs.

Explore what datasets are available in this project. List all files in the /mnt/data/ and /mnt/imported/data/ directories and show me the first few rows of any CSV files you find.

Analyze the diabetes dataset associated with this project and evaluate its usefulness for the purpose of training a diabetes prediction model. Create and run a data transformation script if any data engineering is needed to better adapt it for model training.

Create a script that analyzes the structure and basic statistics of all datasets in this project, then run it as a Domino job.

Given the diabetes data we just engineered under '/mnt/data/diabetes_project' in this project, create a PyTorch model architecture in one file for a customizable neural network we can use to predict diabetes that we'll train with the data. Let's keep it at 4 fully connected layers, with customizable hidden dimensions. Create a separate commandline executable script that we can use to run the training for different configurations of the model and training routine. Ensure we use mlflow tracking to track metrics, params, and log the model itself.

Understand how the <filename> script works and use it to train a small neural network for the diabetes model. Iterate until the model is optimized enough.

Run a hyperparameter tuning experiment with 3 different configurations and compare the results.

Create a data preprocessing script that cleans and transforms the raw data, saves the processed version, then run it as a Domino job.

Generate a comprehensive data quality report for our main dataset and save the results to a file.

Analyze my existing training script and run it with different parameter combinations as separate Domino jobs.

Check if I have any uncommitted changes, commit them if needed, then run my latest model training script.

Run an experiment that compares three different algorithms on our dataset, make sure to log all metrics to MLflow, and share the experiment URL with me.

Create a script that loads our best model and evaluates it on test data, logging the results to MLflow.

Show me the status of all my recent Domino job runs and summarize what each one accomplished.

Create a comprehensive analysis report of our project's model performance over time based on previous job runs.

git status
git add .
git commit -m "Update files for domino job"
git push