Choose your language

The default path uses VS Code + GitHub Copilot. Keep the guide open alongside VS Code; the example repository includes code, tests, and offline lessons.

Choose the language you already use. Its entry page leads to the detailed VS Code + Copilot setup, then the Next links take you through six lessons.

Python

Use VS Code + Copilot with Python and Matplotlib. Repair the plot, check its image, and review the diff.

Python 3.10–3.12

Start the Python path

R

Use VS Code + Copilot with base R. Repair the plot, check its image, and review the diff.

R 4.1 or later

Start the R path

You need VS Code, Git, your chosen language, and Copilot access. Setup explains installation, student access, cloning, and the Chat controls. Copilot runs the commands; you review their output and the resulting files. Prepare before a workshop or work at your own pace.

Before granting access, read permissions and independent work. Practise evaluating requests, keep the repair understandable, and retain the ability to run and maintain the code without AI assistance.

Prefer Claude Code? Use the short VS Code extension alternative. Prefer a terminal? Follow the complete CLI appendix for Python or R. The exercise and acceptance criteria are the same.

What you will do

  1. In Agent, request setup checks and a baseline PNG; open it in Explorer.
  2. In Ask, understand the plotting code and the overlapping bars and clipped labels.
  3. In Plan, agree on a repair against the local figure specifications.
  4. In Agent, implement the repair and rerun both checks against the latest source.
  5. Compare the plots and alternative text, then inspect every file in Source Control → Changes.
  6. Stage reviewed source, make a local commit, and leave a handoff with results and unfinished work.

Bookmark your path or current lesson. Change language returns to this choice.

Coming from Part 1

The Part 1 quickstart created a base-R function. Here you repair an existing plot using the exercise brief and local figure rules.

Use the supplied invented totals. Keep real records outside the workspace and conversations; read the lab data guidance before adapting this to a study.

Teaching a group? Use the 45-minute workshop, with each participant’s setup completed beforehand.