format-citations
About
This skill formats academic citations and bibliographies for major styles like APA and IEEE using CSL processors and R tooling. It converts between citation styles, generates reference lists, and validates formatting within Quarto or R Markdown projects. Use it when rendering documents with citations or setting up citation infrastructure.
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/format-citationsCopy and paste this command in Claude Code to install this skill
Documentation
Format Citations
Format citations across academic styles using CSL (Citation Style Language) processors and R tooling. This skill covers turning BibTeX entries into properly formatted in-text citations and reference lists for APA 7, Chicago, Vancouver, IEEE, and custom styles. Uses Pandoc's citeproc, knitcitations package, and Quarto's native citation engine for reproducible document making.
When Use
- Rendering R Markdown or Quarto document with formatted citations
- Turning bibliography from one citation style to another
- Making standalone reference list from .bib file
- Check that in-text citations match specific style guide
- Setting up citation infra for multi-document project (book, thesis)
Inputs
- Required: .bib file (or other bibliography source known by Pandoc)
- Required: Target citation style (e.g.,
apa,chicago-author-date,ieee) - Optional: CSL file path (default: uses Pandoc built-in styles)
- Optional: Output format (
html,pdf,docx; default: guessed from document) - Optional: Locale for lang-specific format (default:
en-US)
Steps
Step 1: Verify Citation Infrastructure
# Check Pandoc availability (required for citeproc)
pandoc_path <- Sys.which("pandoc")
if (!nzchar(pandoc_path)) {
pandoc_path <- Sys.getenv("RSTUDIO_PANDOC")
}
stopifnot("Pandoc not found" = nzchar(pandoc_path))
message(sprintf("Pandoc: %s", system2(pandoc_path, "--version", stdout = TRUE)[1]))
# Check for citeproc support
citeproc_ok <- any(grepl("citeproc", system2(pandoc_path, "--list-extensions", stdout = TRUE)))
message(sprintf("Citeproc: %s", ifelse(citeproc_ok, "built-in", "external needed")))
Got: Pandoc version 2.11+ spotted with built-in citeproc support.
If fail: Install Pandoc or set RSTUDIO_PANDOC in .Renviron to point to RStudio-bundled Pandoc. Quarto also ships its own Pandoc.
Step 2: Configure Document YAML for Citations
For R Markdown:
---
title: "My Document"
bibliography: references.bib
csl: apa.csl
link-citations: true
output:
html_document:
pandoc_args: ["--citeproc"]
---
For Quarto:
---
title: "My Document"
bibliography: references.bib
csl: apa.csl
link-citations: true
cite-method: citeproc
---
Got: YAML header right refs .bib file and CSL style.
If fail: CSL file not found? Download it from CSL repo (see Step 3) and place it in project directory.
Step 3: Obtain CSL Style Files
# Common CSL styles and their repository names
csl_styles <- list(
apa = "apa.csl",
chicago = "chicago-author-date.csl",
vancouver = "vancouver.csl",
ieee = "ieee.csl",
nature = "nature.csl",
harvard = "harvard-cite-them-right.csl",
mla = "modern-language-association.csl"
)
download_csl <- function(style, dest_dir = ".") {
base_url <- "https://raw.githubusercontent.com/citation-style-language/styles/master"
filename <- csl_styles[[style]]
if (is.null(filename)) stop(sprintf("Unknown style: %s", style))
dest <- file.path(dest_dir, filename)
utils::download.file(
url = sprintf("%s/%s", base_url, filename),
destfile = dest, quiet = TRUE
)
message(sprintf("Downloaded %s to %s", filename, dest))
dest
}
# Download APA 7 style
download_csl("apa")
Got: CSL file downloaded to project directory.
If fail: Check network. CSL GitHub repo has 10,000+ styles. For offline use, bundle needed CSL files in project.
Step 4: Write In-Text Citations
Use Pandoc citation syntax in your document body:
<!-- Single citation -->
According to @Smith2020, the method improves accuracy.
<!-- Parenthetical citation -->
The method improves accuracy [@Smith2020].
<!-- Multiple citations -->
Several studies confirm this [@Smith2020; @Jones2021; @Lee2022].
<!-- Citation with page number -->
As noted by @Smith2020 [p. 42], the results are significant.
<!-- Suppress author name -->
The results are significant [-@Smith2020].
<!-- Citation with prefix -->
[see @Smith2020, pp. 42-45; also @Jones2021, ch. 3]
Got: Pandoc/Quarto renders these into properly formatted citations in target style (e.g., (Smith, 2020) for APA, (Smith 2020) for Chicago).
Step 5: Generate Standalone Reference Lists with R
# Using RefManageR to print formatted references
library(RefManageR)
BibOptions(style = "text", bib.style = "authoryear", sorting = "nyt")
bib <- ReadBib("references.bib", check = FALSE)
# Print all entries in text format
print(bib)
# Format specific entries
print(bib[author = "Smith"])
# Generate markdown reference list programmatically
format_reference_list <- function(bib, style = "apa") {
BibOptions(style = "text", bib.style = "authoryear")
entries <- capture.output(print(bib))
entries <- entries[nzchar(trimws(entries))]
paste(sprintf("- %s", entries), collapse = "\n")
}
cat(format_reference_list(bib))
Got: Formatted reference list printed to console or caught as char vector for more handle.
Step 6: Convert Between Citation Styles
# Render the same document in different styles
styles <- c("apa", "chicago", "ieee")
for (style in styles) {
csl_file <- download_csl(style)
output_file <- sprintf("output_%s.html", style)
rmarkdown::render(
input = "document.Rmd",
output_file = output_file,
params = list(csl = csl_file),
quiet = TRUE
)
message(sprintf("Rendered %s with %s style", output_file, style))
}
For Quarto:
quarto render document.qmd --metadata csl:apa.csl -o output_apa.html
quarto render document.qmd --metadata csl:ieee.csl -o output_ieee.html
Got: Many output files, each with same content formatted in different citation style.
If fail: Rendering fails? Check all citation keys in document body exist in .bib file. Missing keys make warnings but may break format.
Step 7: Validate Citation Formatting
# Check for undefined citations in rendered output
validate_citations <- function(rmd_file, bib_file) {
# Extract citation keys from document
doc_text <- readLines(rmd_file, warn = FALSE)
doc_keys <- unique(unlist(regmatches(
doc_text,
gregexpr("@([A-Za-z][A-Za-z0-9_:.#$%&+-?<>~/]*)", doc_text)
)))
doc_keys <- gsub("^@", "", doc_keys)
# Remove false positives (email-like patterns)
doc_keys <- doc_keys[!grepl("\\.", doc_keys)]
# Extract keys from .bib file
bib <- RefManageR::ReadBib(bib_file, check = FALSE)
bib_keys <- names(bib)
# Find mismatches
undefined <- setdiff(doc_keys, bib_keys)
unused <- setdiff(bib_keys, doc_keys)
list(
undefined = undefined,
unused = unused,
cited = intersect(doc_keys, bib_keys)
)
}
result <- validate_citations("document.Rmd", "references.bib")
if (length(result$undefined) > 0) {
warning(sprintf("Undefined citation keys: %s",
paste(result$undefined, collapse = ", ")))
}
if (length(result$unused) > 0) {
message(sprintf("Unused .bib entries: %s",
paste(result$unused, collapse = ", ")))
}
Got: Report of undefined keys (cited but not in .bib), unused entries (in .bib but never cited), and valid citations.
If fail: False positives may show up with email addresses or code with @. Refine regex or by hand review flagged keys.
Validation
- Document renders with no citation warnings from Pandoc/citeproc
- All
@keyrefs in document resolve to .bib entries - Reference list shows at end of document (or in
div#refs) - In-text citations match target style format
- Citation sorting follows style rules (alphabetical for APA, numbered for IEEE)
- Hyperlinks from in-text citations to reference list entries work (if
link-citations: true)
Pitfalls
- Missing CSL file: Pandoc falls back to Chicago author-date if no CSL is set. Always set
csl:clear for style consistency - Citation key typos: Misspelled key like
@Smtih2020silent renders as literal text. Turn on Pandoc warnings with--verboseto catch these - Locale-dependent format: APA needs "and" between authors in English but "und" in German. Set
lang:in YAML header to match - nocite for uncited entries: To add entries in reference list with no cite in text, add
nocite: '@*'(all) ornocite: '@key1, @key2'to YAML - CSL version mismatch: Some older CSL 0.8 files clash with modern Pandoc. Always use CSL 1.0+ from official repo
- Quarto vs R Markdown diffs: Quarto uses
cite-method: citeprocby default; R Markdown may need clearpandoc_args: ["--citeproc"]
See Also
manage-bibliography- make and keep .bib files this skill consumesvalidate-references- check .bib entry completeness before format../reporting/format-apa-report- full APA report format beyond citations../reporting/create-quarto-report- Quarto document setup with citation support
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.
