paperkit

Write a research paper once — get submission-ready LaTeX for every major publisher.

You write the paper as plain Markdown (paper.md) with a BibTeX file and a figures/ folder. paperkit renders it into the official LaTeX templates of IEEE, Springer Nature, Springer LNCS, Elsevier, MDPI, ACM, Taylor & Francis and arXiv. It checks every output and, if Tectonic is installed, compiles every variant to PDF to prove it works. Text, numbers and references are identical in every format, so you can submit the same paper anywhere and re-target it after a rejection in minutes.

Publisher	Template	Variants
IEEE	IEEEtran	conference, journal (Transactions), single-column review copy
Springer Nature journals	sn-jnl (v3.1, Dec 2024)	all 9 reference styles + referee (double-spaced) + line numbers
Springer proceedings	llncs	LNCS / CCIS / LNAI
Elsevier	elsarticle	preprint & review (numbered), preprint (author-year), 1p, 3p and 5p journal layouts
Elsevier (newer journals)	cas-sc / cas-dc	single and double column
MDPI	mdpi.cls (official)	Drones, Remote Sensing, Sensors, Applied Sciences, Electronics, J. Imaging, AI, Computers, Information, BDCC (any MDPI journal by changing one option)
ACM	acmart	sigconf, acmsmall (journals), anonymous review manuscript
Taylor & Francis	interact	Interact layout, NLM references
Preprint	article	arXiv / TechRxiv / SSRN (standard packages only)

Each publisher's back matter is mapped to what that publisher asks for: Springer Declarations, Elsevier CRediT + competing interest + generative-AI statements and highlights, MDPI's statement commands (including the DURC statement for Drones), the IEEE/ACM acknowledgement with AI disclosure, and the T&F disclosure, funding and data statements.

Quick start

pip install git+https://github.com/Apurba94/paperkit        # or: pip install -e path/to/paperkit
python -m paperkit new my-paper                               # creates paper.md, references.bib, figures/
# ... write the paper in my-paper/paper.md ...
python -m paperkit fetch                                      # once: Springer Nature + MDPI class files
python -m paperkit build my-paper --compile --zip             # -> my-paper/journals/

Open my-paper/journals/README.md to see every variant with its compile status, then upload the folder (or its zip) for your target journal to Overleaf or compile it locally.

Writing paper.md

The YAML header holds the metadata (title, authors with affiliations/ORCID/e-mail, abstract, keywords, Elsevier highlights, generative-AI disclosure and the declarations). The body is Markdown:

Element	Syntax
Section / subsection / subsubsection	# Title {#sec:label}, ##, ###
Appendix	# Appendix: Title {#sec:appA}
Citation	[@key], [@key1; @key2]
Cross reference	@fig:label, @tab:label, @eq:label, @sec:label
Emphasis, code, math	**bold**, *italic*, `code`, $x^2$
Display equation	$$ ... $$ {#eq:label}
Figure	{#fig:label width=0.7} (add wide to span two columns)
Table	Table: Caption. {#tab:label cols=Xrrr wide} followed by a pipe table (X = wrapping column)
Lists	- item, 1. item

See paperkit/starter/paper.md for a complete example. Unknown declarations should stay [Author to complete.]; paperkit lists them in the build README so nothing is submitted with a placeholder by accident.

Publisher class files

IEEE, Springer LNCS, Elsevier, ACM and the arXiv template use classes that ship with every TeX distribution and with Overleaf. Springer Nature, MDPI and Taylor & Francis distribute their classes as downloads; python -m paperkit fetch downloads the Springer Nature and MDPI packages from the publishers into ~/.paperkit/publishers/. Taylor & Francis protects its site with a bot check, so download the template zip from your journal's Instructions for authors page and run python -m paperkit import-files taylor-francis interact.zip. These files are copied next to the generated .tex files but are never committed to your repositories.

Compile testing

With Tectonic installed (on PATH, at ~/.paperkit/bin/tectonic(.exe), or via PAPERKIT_TECTONIC), --compile compiles every variant (BibTeX included) and reports page counts, undefined citations/references and overfull lines. Tectonic uses the XeTeX engine; journals compile with pdfLaTeX, which these ASCII-only sources also support.

Python API

from paperkit import read_paper, build
paper = read_paper("my-paper")
build(paper, "my-paper/journals", only=["mdpi", "elsevier"], compile=True)

Projects that generate their paper programmatically (e.g. numbers computed from experiment results) can construct paperkit.Paper directly; see paperkit/model.py for the block types.

License

MIT. Publisher templates remain the property of their publishers.