Appendix C — Cheatsheet
Quarto, reproducible publishing, coding agents, Markdown, Python, EPUB, data analysis, k-means clustering
A one-page reference for the patterns this book uses, with the chapter that explains each.
Front matter (YAML)
One source, several formats (Chapter 7):
---
title: "Project Report"
author: "Author Name"
date: today
format:
html:
toc: true
format-links:
- docx
- {format: typst, text: PDF}
docx: default
typst:
papersize: a4
pptx: default
revealjs:
output-file: report-slides.html
epub:
toc: true
lang: en
---Code cells and inline code
Options go at the top of a cell, as #| comments (Section 7.4):
#| label: fig-namenumbers the figure and lets the text refer to it as@fig-name.#| echo: falsehides the code and shows the output.#| eval: falseshows the code without running it.#| fig-cap: "…"is the caption under a figure.#| fig-alt: "…"is the alt text screen readers announce.#| tbl-cap: "…"is the caption of a table.#| output: asistreats printed text as Markdown.
Inline code puts a computed value in a sentence (Section 7.3):
The dataset contains `{python} len(df)` rows across `{python} df['year'].nunique()` years.Dashboards
A level-1 heading starts a page, a level-2 heading a row, and each cell’s output becomes a card (Section 8.1):
---
title: "Operations dashboard"
format: dashboard
---
# Overview
## Row {height=30%}
::: {.valuebox icon="trophy" color="primary"}
Most visited
`{{python}} top_name`
:::
## Row
```{python}
#| title: Visits by attraction
fig = px.bar(attractions, x="visits_2025", y="attraction")
fig
```Assign Plotly figures and end the cell with fig; never call fig.show() in a dashboard.
Websites
_quarto.yml turns a folder into a website (Section 9.3, Section 23.4):
project:
type: website
execute-dir: project
pre-render: scripts/analysis.py
render:
- index.qmd
- report.qmd
website:
title: "My project"
navbar:
left:
- {text: Report, href: report.qmd}
- text: Downloads
menu:
- {text: "Report (PDF)", href: report.pdf}
execute:
freeze: falseCommands
| Command | Does |
|---|---|
uv init --bare |
starts a uv project in this folder |
uv add pandas matplotlib |
adds packages to pyproject.toml and uv.lock, and installs them |
uv run quarto preview file.qmd |
renders and serves a page, re-rendering on every save |
uv run quarto render file.qmd |
renders one file in all its formats; --to docx for one format |
uv run quarto render |
renders the whole project |
uv run quarto publish gh-pages |
the first publish to GitHub Pages, interactive |
uv run quarto publish gh-pages --no-prompt |
later updates, once the gh-pages branch exists |
uv run quarto check jupyter |
shows which Python and Jupyter Quarto will use |
Check and undo with Git
Run them in the project folder, before and after every change, yours or the agent’s (Section 5.4, Section 5.5, Section 9.1):
| Command | Does |
|---|---|
git init -b main |
starts a history in this folder, on a branch called main (once per project) |
git status |
lists the files changed, added or untracked since the last commit |
git diff |
shows the changes not yet added, line by line: - the old line, + the new |
git add . |
adds every change to the next commit; git add report.qmd adds one file |
git commit -m "What changed" |
records the added changes as a commit, with a message saying what changed |
git log --oneline |
lists the commits, newest first, one line each |
git restore data/visits.csv |
throws away the uncommitted changes to one file, back to its last commit |
Prompts
A reliable prompt (Section 2.6):
- Names the inputs and outputs: file names and column names, in backticks.
- Does one task.
- Ends with a check: “render it and tell me which file it wrote”, “check the PDF is one page”.
- Respects the data: “compute every number from the cited files; never invent one”.