PDFMathTranslate MCP

Translates scientific PDFs while keeping formulas, tables and layout intact, as a CLI, Docker image and MCP server for agents

  • 38,408 ★ on the repository
  • Last commit
  • AGPL-3.0
MCP serverOfficial

Extended access

Why this level

  • Sends document content to an external translation service
  • Some translation services require a paid API key
All reasons and checks

pdfmathtranslate/pdfmathtranslate

Install

Manual install

pip install uv
uv tool install --python 3.12 pdf2zh

Install via uv, then translate with pdf2zh document.pdf

Access and checks

How we review

Extended accessThe tool runs code or makes network calls and only reads external data.

  • Sends document content to an external translation service
  • Some translation services require a paid API key

This is third-party code. Review the repository files before installing.

What it does

The tool translates scientific PDFs while keeping formulas, charts, the table of contents and annotations in place. It runs as a command line tool (pdf2zh document.pdf), a web UI, a Docker image, and an MCP server: the --mcp flag starts stdio mode, --mcp --sse starts SSE mode. Several translation services can be picked via the -s flag. The MCP tool translate_pdf currently hardcodes the Google service regardless of what service the plain CLI mode would use.

Who it is for. For researchers, students and technical editors who need a bilingual version of a scientific paper without losing the layout.

Good fit when

  • You need to translate a scientific PDF while keeping formulas and page structure
  • You want an agent to translate PDFs on demand through MCP
  • You need batch translation via CLI or Docker without manual UI work

Not a fit when

  • The document is not scientific and layout does not matter, a plain text translator is simpler
  • No supported translation service is reachable and there is no time to set up a local option

Example request

Translate this scientific PDF to Russian, keeping the formulas and layout

MCP

Transport
stdio, sse
Authentication
not required

Limitations

Requires Python 3.11 through 3.12. Translation needs a configured translation service, and some require a paid API key. The AGPL-3.0 license imposes obligations on derivative works. The MCP mode currently always translates through Google, even if a different service is selected for the CLI.

How to disable. Remove the package: uv tool uninstall pdf2zh or pip uninstall pdf2zh, and remove the MCP server entry from the agent's configuration.

FAQ

Which translation services are supported?

The README lists several services selectable via the -s flag, but the translate_pdf MCP tool currently always uses Google regardless.

Does it work without a VPN from Russia?

It depends on the chosen translation service; Google, which the MCP mode hardcodes, is typically unreachable from Russia without a VPN.

How is PDFMathTranslate different from ordinary PDF translators?

It translates scientific PDFs with formulas, tables and diagrams kept in place, preserving the table of contents and annotations. It runs as a command line, web UI, Docker image and MCP server for agents.

How do I set up PDFMathTranslate with a local model?

Install with pip install uv and uv tool install --python 3.12 pdf2zh, then pick a translation service in the settings. The supported services and launch options are in the install block and the README.

Discussion

No questions or reviews yet.

Official

MCP PDF from ComPDF: the agent converts PDFs and images to Word, Excel, Markdown and JSON and merges, splits, protects and compares files via a cloud API

MCP serverBroad access98Repository stars

MinerU PDF Translate turns scientific PDFs into translated PDFs in another language with formulas and figures preserved.

SkillExtended access100Repository stars

MCP PDF for large documents: the agent searches by meaning and keyword across one file or a folder and reads only the pages, tables and scans it needs

MCP serverExtended access171Repository stars
Official

An MCP server built on Docling: converts PDFs into structured documents, builds new ones and feeds them to RAG

MCP serverExtended access791Repository stars

An MCP server that turns PDF, DOCX, XLSX, PPTX, images, audio, web pages and YouTube transcripts into Markdown

MCP serverExtended access3KRepository stars

PDF extraction with self-checking: unreadable pages get a flag and never vanish silently, plus an MCP server for Claude

MCP serverExtended access86Repository stars

README in short

The README describes the tool as scientific PDF translation that preserves formulas, charts, the table of contents and annotations, available as a CLI, a web UI, and Docker. The CLI options table separately lists --mcp for stdio mode and --mcp --sse for SSE mode. The project was accepted as a demo at EMNLP 2025, and a free public online service runs at pdf2zh.com. AGPL-3.0 licensed.

README badge

Report a problem