How is this different from just uploading a PDF to a standard translator?

Standard PDF translators often scramble mathematical formulas and tables. By using the LaTeX source code instead of the rendered PDF, this skill ensures that complex equations, citations, and numbering remain perfectly intact and professionally formatted.

Which languages are supported for translation?

The skill supports a wide range of global languages including Chinese, Japanese, Korean, French, German, Spanish, and English. It intelligently adjusts fonts, line breaks, and labels (like 'Figure' or 'Table') to match the typographic standards of your chosen language.

Can I use this skill for documents that are not on arXiv?

While optimized for arXiv papers, the skill can process any PDF, DOCX, or EPUB file. It utilizes Calibre to convert and prepare non-LaTeX documents, though results are most precise when the source LaTeX is available.

What file formats are provided at the end of the translation process?

Yes, the skill generates several formats including a print-ready PDF, an editable DOCX, and mobile-friendly EPUB and HTML files, allowing you to read your translated paper on any device.

What happens if the translation process is interrupted?

This skill uses parallel subagents to speed up processing and includes a manifest-check feature. If a translation is interrupted by a connection error or rate limit, it can resume exactly where it left off without starting over.

Translate Book: arXiv Papers into Your Language

Example session with this skill installed

translate ~/papers/2609.11801.pdf to Korean

Read your context and instructions
Compiled the translate book

Summary

Translated arXiv:2609.11801, Thinking with Looped Flows (Suleymanzade et al.), into Korean from its LaTeX source. The book is 29 pages. The paper is CC BY 4.0, so the translation may be shared with attribution.

Translated title: 루프 플로우로 사고하기

Output	Size
`book.pdf` (print-ready, 29 pages)	2.56 MB
`book.epub`	1.07 MB
`book.docx`	1.00 MB
`book.html`	238 KB

All in 2609.11801_temp/. All four formats built.

What was checked

Check	Result
Chunks translated	10 of 10
Chunk verification	10 passed, 0 failed
Section numbering	kept from the paper: 1 서론 (Introduction) ... 6 결론 (Conclusion), appendix sections by letter
Figures	7, rasterised from the source PDFs, captions translated as `그림 N (Fig. N)`
Reference list	copied, not translated

Next steps

Open book.pdf and compare a page with the original: equations, table values and ± signs are the paper's own.
Before sharing a translation of another paper, check its licence on arXiv: CC BY papers may be republished with attribution, papers under arXiv's default licence may not.

Connects securely to your tools. The creator never sees your data.

About this skill

Translating a paper from its PDF means translating what a PDF reader can recover from it, and formulas are the first thing lost. This skill fetches the LaTeX the authors wrote instead: it recognises an arXiv preprint from its first page and asks you before downloading the source.

Pick the target language with one flag: Chinese, English, Japanese, Korean, French, German, Spanish, and more. Each language gets its own labels (Figure 1, 그림 1, 図 1, Abbildung 1), fonts and line breaking, while every equation, number and citation stays the paper's own.

What it does

Resolves the paper's own macros and reads equation, theorem, section and float numbers from the source, so the book keeps the paper's numbering.
Rasterises figures from the original vector PDFs and keeps their captions.
Copies the reference list instead of translating it.
Translates in parallel with subagents, checks every chunk against a manifest, and resumes an interrupted run.
Builds HTML, DOCX, EPUB and a print-ready PDF.
Also takes any PDF, DOCX or EPUB book through Calibre, not only arXiv papers.

Free and open source under the MIT licence. The latest version, issues and releases are on GitHub: https://github.com/kcy4334-lgtm/translate-book-arxiv

How to install

Works the same in every agent - Claude, Cursor, Codex, Copilot and 20+ more.

~30 seconds

1
Download the ZIP
Free skills download straight away. Paid skills unlock right after purchase.
2
Unzip into your skills folder
Every agent reads skills from one folder on your machine. Drop the unzipped folder in there.
3
Ask your agent to use it
Restart the agent if it was already running. It picks the skill up automatically - no config needed.

Skills folder by agent

Click the path to copy it. Create the folder if it does not exist yet.

Reviews

No reviews yet

Be one of the first to try it. Every listed skill passes our trust checks below.

Security scanned

Passed our 8-point scan before listing

Fresh listing

Recently published to Agensi

Free forever

No account required to browse

Translate Book: arXiv Papers into Your Language