| name | quarto_publishing |
| description | Complete skill for architecting, writing, and compiling scientific documents, websites, dashboards, and books using Quarto. Includes guidelines for generating interactive Observable JS blocks and automated CI/CD deployment pipelines. |
Quarto Scientific Publishing Skill
This skill equips you with the knowledge and curated templates necessary to build high-fidelity data narratives, websites, and dashboards using Quarto.
1. System Requirements & CLI Usage
Quarto relies on the quarto CLI. When executing this skill, first verify if Quarto is installed:
quarto --version
If not installed, instruct the user to install it from https://quarto.org/docs/get-started/ or use standard package managers (e.g. winget install JanDeDobbeleer.OhMyPosh wait, winget install quarto.quarto, brew install --cask quarto).
Common Commands
quarto render document.qmd : Compiles a single file.quarto preview document.qmd : Starts a live-reloading local server.quarto create project <type> <name> : Scaffolds a new project (type can be default, website, blog, book, confluence).quarto publish gh-pages : Automates deployment to GitHub pages (requires git tree).
2. Anatomy of a .qmd File
Quarto Markdown (.qmd) uses YAML frontmatter to define the document's properties, format, and behavior, followed by standard markdown interspersed with executable code blocks.
---
title: "Global Temperature Analysis"
author: "AI Analyst"
format:
html:
theme: cosmo
toc: true
code-fold: true
execute:
echo: false
warning: false
---
## Introduction
Here is an analysis of the temperature dataset...
```python
import pandas as pd
import matplotlib.pyplot as plt
# The code will be executed and the plot will be embedded directly in the output!
plt.plot([1, 2, 3], [10, 20, 25])
plt.show()
## 3. High-Value Capabilities (Curated AI Workflows)
As an AI, you can leverage Quarto to provide massive value to the user:
### A. Instant Interactive Dashboards
Quarto can generate highly polished Dashboards using purely markdown and Observable JS (or Python/R).
To create a dashboard, set `format: dashboard` in the YAML frontmatter.
```markdown
---
title: "Live Metrics Dashboard"
format: dashboard
logo: logo.png
nav-buttons:
- icon: github
href: https://github.com/
---
# Row {height=30%}
## Column
```markdown
**Total Revenue**
# $45,000
Column
**Active Users**
# 1,250
Row {height=70%}
...
### B. Observable JS (OJS)
When writing client-side interactive documents, prefer using `{ojs}` code blocks. Observable JS is reactive by default and requires zero backend server configuration.
```markdown
```{ojs}
// Reactive slider
viewof num = Inputs.range([1, 10], {step: 1, value: 5, label: "Multiplier"})
// Reactive calculation
num * 10
### C. Automated GitHub Pages CI/CD
To deploy a Quarto website or book, you can utilize the curated GitHub Actions template bundled in this skill at `templates/quarto-publish.yml`.
When a user asks to deploy their Quarto project:
1. Ensure the root directory has a `_quarto.yml` file.
2. Copy `.skills/quarto/templates/quarto-publish.yml` into the user's `.github/workflows/` directory.
3. Commit and push. GitHub Actions will automatically render and publish the Quarto site to the `gh-pages` branch.
## 4. Execution Rules
* **Never** assume Python or R dependencies are installed. Always write out a `requirements.txt` or execute `pip install` when generating computational `.qmd` files that rely on external libraries.
* **Always** use `{ojs}` blocks for lightweight interactive visualizations unless the user specifically requests a Python/R server-backed Shiny application.
* **Keep it Beautiful**: Quarto supports Bootstrap 5 themes. Always inject a premium theme (e.g., `theme: flatly`, `lumen`, `zephyr`, `lux`, or `slate` for dark mode) in the frontmatter rather than leaving it default.
