Skip to content
vibespell

.github/copilot-instructions.md

Copilot instructions for Python ML

GitHub Copilot reads `.github/copilot-instructions.md` and applies it across the repository. It is the place for build commands, review rules, and stack conventions Copilot should not invent. This copy is written for a Python ML project. This repository is a Python data or machine-learning project. Training code does not see the test split. Notebooks are for exploration. Reusable logic lives in importable modules.

The example uses a balanced, readable style, pragmatic tests, and comments only where the intent is not obvious. Open the generator to make the agent stricter, switch to test-first, quiet the comments, or pick a different file.

The example assumes Python for data or ML work. It does not assume PyTorch versus scikit-learn, or a particular experiment tracker. Use the libraries already in the environment.

Customize in the generatorBrowse the library

Example file

Balanced strictness, readable style, pragmatic tests. Download name: copilot-instructions.md

# GitHub Copilot instructions

These instructions apply to the whole Python ML repository. Copilot reads them from `.github/copilot-instructions.md`.
Prefer the commands and layout below over guesses.

## Stack

This repository is a Python data or machine-learning project. Training code does not see the test split. Notebooks are for exploration. Reusable logic lives in importable modules.

## Layout

- Importable code lives in a package. Notebooks live in a notebooks folder and stay thin.
- Data paths come from config. Do not hard-code a laptop path.
- Experiment outputs go to a gitignored directory.

## Commands

- Tests: `pytest`
- Lint: `ruff check .`
- Train: `Run the project's training entry point, not a notebook cell`

Use the script the repo already defines when it differs from the command above.

## Conventions

- Fit scalers, vocabularies, and feature selection on training data only.
- Pin seeds where the library supports it, and record them.
- Keep a single source of metric computation. Do not reimplement accuracy in a notebook.
- Type the public functions. Validate dataframe assumptions at the boundary.
- Do not commit large data files or credentials.
- Prefer the project's config format for hyperparameters.

## Testing

Add or update tests when behavior changes.

Test the split and the metric with a tiny synthetic frame. Do not download a dataset in the test.

## Style

- Optimize for the next reader over the shortest diff.
- Comment only when the why is non-obvious.
- Name things the way neighboring files already do.
- Do not reformat code you did not change.

## Errors

- Fail at the boundary. Do not swallow errors.
- Raise a clear error when a required column is missing or a split is empty. Do not fill target values to hide it.

## Scope

- Do the requested task. Do not refactor unrelated code.
- Match the rigor already in the repo. Fix what you touch.
- Ask one clarifying question only when the task cannot be finished without it. Otherwise state the assumption and continue.

## Avoid

- Do not tune on the test set.
- Do not leak the label through an id, a timestamp, or a target-encoded feature fit on all rows.
- Do not call the network or a paid API from an import.
- Do not leave a GPU tensor on the wrong device and catch the error with a bare except.

## Done when

- The requested behavior works.
- `pytest` passes and the training entry point still runs the unchanged path, or you say why it cannot.
- The diff stays inside the task.
- You list anything you could not run.

## Review checklist

- The change matches the commands and layout in this file.
- No new dependency was added without a reason already present in the repo.
- Do not commit dataset credentials or cloud keys.
- Treat downloaded data as untrusted input. Validate schema before training.
- Do not log prompts or rows that contain personal data.

Add it to the project

  1. 01

    Copy a starting file

    Copy the example Copilot instructions on this page, or open the generator if you want different testing, strictness, or comment rules.

  2. 02

    Save it at .github/copilot-instructions.md

    Save it as `.github/copilot-instructions.md` in the Python ML repo. Create the `.github` folder if it is missing. Commit the file so Copilot sees it for every collaborator.

  3. 03

    Start a fresh agent session

    Rules files are read when a session starts. A chat that is already open may still be using the old instructions.

  4. 04

    Give it one small task

    Ask for one change a Python ML agent often gets wrong, such as a split that fits preprocessing on the training fold only. If it ignores the file, confirm the path and start a new session.

Questions about this file

Where do I put Copilot instructions in a Python ML repo?

Save it as `.github/copilot-instructions.md` in the Python ML repo. Create the `.github` folder if it is missing. Commit the file so Copilot sees it for every collaborator.

Which tools read this Copilot instructions?

GitHub Copilot in VS Code, Visual Studio, and Copilot cloud agents read this file. Other agents will not load it automatically, so keep an AGENTS.md too if more than one tool edits the repo.

What does the Python ML example assume?

The example assumes Python for data or ML work. It does not assume PyTorch versus scikit-learn, or a particular experiment tracker. Use the libraries already in the environment.

Does VibeSpell upload my repository?

No. Recipe pages are static, and the generator builds the file in your browser from the options you pick. Your source code never leaves the machine.

Other files for Python ML

Copilot instructions for other stacks