write-vignette
About
This skill helps R developers create package vignettes using R Markdown or Quarto for tutorials and complex workflow documentation. It covers setup, YAML configuration, building, and CRAN requirements. Use it when you need user-facing guides, getting started tutorials, or documentation for CRAN submission.
Quick Install
Claude Code
Recommendednpx skills add pjt222/agent-almanac -a claude-code/plugin add https://github.com/pjt222/agent-almanacgit clone https://github.com/pjt222/agent-almanac.git ~/.claude/skills/write-vignetteCopy and paste this command in Claude Code to install this skill
Documentation
Write Vignette
Create long-form documentation vignettes for R packages.
When Use
- Adding "Getting Started" tutorial for package
- Documenting complex workflows that span multiple functions
- Creating domain-specific guides (e.g., statistical methodology)
- CRAN submission needs user-facing documentation beyond function help
Inputs
- Required: R package with functions to document
- Required: Vignette title and topic
- Optional: Format (R Markdown or Quarto, default: R Markdown)
- Optional: Whether vignette needs external data or APIs
Steps
Step 1: Create Vignette File
usethis::use_vignette("getting-started", title = "Getting Started with packagename")
Got: vignettes/getting-started.Rmd created with YAML frontmatter. knitr and rmarkdown added to DESCRIPTION Suggests field. vignettes/ directory exists.
If err: usethis::use_vignette() fails? Verify working directory is package root (contains DESCRIPTION). knitr not installed? Run install.packages("knitr") first. For manual creation, create vignettes/ directory and file by hand. Ensure YAML frontmatter includes all three %\Vignette* entries.
Step 2: Write Vignette Content
---
title: "Getting Started with packagename"
output: rmarkdown::html_vignette
vignette: >
%\VignetteIndexEntry{Getting Started with packagename}
%\VignetteEngine{knitr::rmarkdown}
%\VignetteEncoding{UTF-8}
---
## Introduction
Brief overview of what the package does and who it's for.
## Installation
```r
install.packages("packagename")
library(packagename)
Basic Usage
Walk through the primary workflow:
# Load example data
data <- example_data()
# Process
result <- main_function(data, option = "default")
# Inspect
summary(result)
Advanced Features
Cover optional or advanced functionality.
Conclusion
Summarize and point to other vignettes or resources.
**Got:** Vignette Rmd file contains Introduction, Installation, Basic Usage, Advanced Features, Conclusion sections. Code examples use package's exported functions and produce visible output.
**If err:** Examples fail to run? Verify package installed with `devtools::install()`. Ensure examples use package name in `library()` calls (not `devtools::load_all()`). For functions requiring external resources, use `eval=FALSE` to show code without execution.
### Step 3: Configure Code Chunks
Use chunk options for different purposes:
```r
# Standard evaluated chunk
{r example-basic}
result <- compute_something(1:10)
result
# Show code but don't run (for illustrative purposes)
{r api-example, eval=FALSE}
connect_to_api(key = "your_key_here")
# Run but hide code (show only output)
{r hidden-setup, echo=FALSE}
library(packagename)
# Set global options
{r setup, include=FALSE}
knitr::opts_chunk$set(
collapse = TRUE,
comment = "#>",
fig.width = 7,
fig.height = 5
)
Got: Setup chunk with include=FALSE sets global options (collapse, comment, fig.width, fig.height). Chunks configured appropriate: eval=FALSE for illustrative code, echo=FALSE for hidden setup, standard chunks for interactive examples.
If err: Chunk options not taking effect? Verify syntax uses {r chunk-name, option=value} format (comma-separated, no quotes around logical values). Check setup chunk runs first by placing at top of document.
Step 4: Handle External Dependencies
For vignettes that need network access or optional packages:
{r check-available, include=FALSE}
has_suggested <- requireNamespace("optionalpkg", quietly = TRUE)
{r use-suggested, eval=has_suggested}
optionalpkg::special_function()
For long-running computations, pre-compute and save results:
# Save pre-computed results to vignettes/
saveRDS(expensive_result, "vignettes/precomputed.rds")
# Load in vignette
{r load-precomputed}
result <- readRDS("precomputed.rds")
Got: External dependencies handled gracefully — optional packages conditionally loaded with requireNamespace(), network-dependent code uses eval=FALSE or tryCatch(), expensive computations use pre-computed .rds files.
If err: Vignette fails on CRAN due to unavailable optional packages? Wrap those sections with conditional variable (e.g., eval=has_suggested). For pre-computed results, ensure .rds file included in vignettes/ directory and referenced with relative path.
Step 5: Build and Test Vignette
# Build single vignette
devtools::build_vignettes()
# Build and check (catches vignette issues)
devtools::check()
Got: Vignette builds without errors. HTML output readable.
If err:
- Missing pandoc: Set
RSTUDIO_PANDOCin.Renviron - Package not installed: Run
devtools::install()first - Missing Suggests: Install packages listed in DESCRIPTION Suggests
Step 6: Verify in Package Check
devtools::check()
Vignette-related checks: builds correctly, doesn't take too long, no errors.
Got: devtools::check() passes with no vignette-related errors or warnings. Vignette builds within CRAN time limits (typically under 60 seconds).
If err: Vignette causes check failures? Common fixes: add missing Suggests packages to DESCRIPTION, reduce build time with eval=FALSE on slow chunks, ensure VignetteIndexEntry matches title. Run devtools::build_vignettes() separately to isolate vignette-specific errors.
Check
- Vignette builds without errors via
devtools::build_vignettes() - All code chunks execute correctly
- VignetteIndexEntry matches title
-
devtools::check()passes with no vignette warnings - Vignette appears in pkgdown site articles (if applicable)
- Build time reasonable (< 60 seconds for CRAN)
Pitfalls
- VignetteIndexEntry mismatch: Index entry in YAML must match what you want users to see in
vignette(package = "pkg") - Missing
vignetteYAML block: All three%\Vignette*lines required - Vignette too slow for CRAN: Pre-compute results or use
eval=FALSEfor expensive operations - Pandoc not found: Ensure
RSTUDIO_PANDOCenvironment variable set - Self-referencing package: Use
library(packagename)notdevtools::load_all()in vignettes
See Also
write-roxygen-docs- function-level docs complement vignette tutorialsbuild-pkgdown-site- vignettes appear as articles on pkgdown sitesubmit-to-cran- CRAN has specific vignette requirementscreate-quarto-report- Quarto as alternative to R Markdown vignettes
GitHub Repository
Related Skills
content-collections
MetaThis skill provides a production-tested setup for Content Collections, a TypeScript-first tool that transforms Markdown/MDX files into type-safe data collections with Zod validation. Use it when building blogs, documentation sites, or content-heavy Vite + React applications to ensure type safety and automatic content validation. It covers everything from Vite plugin configuration and MDX compilation to deployment optimization and schema validation.
polymarket
MetaThis skill enables developers to build applications with the Polymarket prediction markets platform, including API integration for trading and market data. It also provides real-time data streaming via WebSocket to monitor live trades and market activity. Use it for implementing trading strategies or creating tools that process live market updates.
creating-opencode-plugins
MetaThis skill helps developers create OpenCode plugins that hook into 25+ event types like commands, files, and LSP operations. It provides the plugin structure, event API specifications, and implementation patterns for JavaScript/TypeScript modules. Use it when you need to intercept, monitor, or extend the OpenCode AI assistant's lifecycle with custom event-driven logic.
sglang
MetaSGLang is a high-performance LLM serving framework that specializes in fast, structured generation for JSON, regex, and agentic workflows using its RadixAttention prefix caching. It delivers significantly faster inference, especially for tasks with repeated prefixes, making it ideal for complex, structured outputs and multi-turn conversations. Choose SGLang over alternatives like vLLM when you need constrained decoding or are building applications with extensive prefix sharing.
