LabscoMCP early access ↗
FradSerMCP SERVER

MCP Server To Markdown

Hand it a list of local file paths — PDFs, images, spreadsheets, HTML, XML, CSV — and get Markdown descriptions back through Cloudflare's tomarkdown API.

by FradSer ↗Document Conversion, PDF & TranslationCommunity

LABSCO SUMMARY

The spreadsheet coverage is the unusual part.

Alongside the expected PDF and image handling it accepts .numbers, .ods and the whole Excel family, which is the format range that usually forces a manual export step first. Files are sent to Cloudflare for conversion, so this is the wrong tool for documents that must not leave your machine — and the returned token count per file is worth reading before you paste a large batch into a conversation.

What it is

An MCP server that converts files into Markdown descriptions using Cloudflare AI's native tomarkdown API. It takes an array of file paths in one call and returns, per file, the filename, MIME type, the generated Markdown description and a token count — so batch conversion is a single tool call rather than one per document.

What you get

  • Several files converted in one call, each answered with its filename, MIME type, description and token count — the to-markdown tool, whose only required input is filePaths
  • Documents: .pdf
  • Images: .jpeg, .jpg, .png, .webp, .svg
  • Web and data formats: .html, .xml, .csv
  • Spreadsheets: .xlsx, .xlsm, .xlsb, .xls, .et, .ods, .numbers

Requirements

A Cloudflare account: both CLOUDFLARE_API_TOKEN and CLOUDFLARE_ACCOUNT_ID must be set, and conversion runs on Cloudflare's side rather than locally. Node.js 18 or later. The npm package has been withdrawn, so build it from the repository with npm install and npm run build, then point Cursor or Claude Desktop at dist/index.js with both Cloudflare variables in the config's env.

Setup effort

Build from source — git clone https://github.com/FradSer/mcp-server-to-markdown.git