pdfmux MCP server

pdfmux is a helper that reads PDF files and turns them into text your AI can actually use. It also checks its own work, so if a page comes out blank or scrambled, it tries again with a different method instead of quietly losing that page. It is handy for anyone who deals with lots of PDFs, like reports, invoices or scanned documents.

What this MCP server does

You ask your AI something about a PDF, like to pull out the text or check a document. Your AI sends the file to pdfmux through this helper. pdfmux looks at each page, picks the reading method that fits best, and checks whether the result looks right. If a page looks bad, it tries again with a stronger method. Then it hands the clean text back to your AI, along with a confidence score so you know how much to trust it.

Click to zoom

What you can do with it

• Turn a PDF into clean Markdown text
• Extract text from scanned pages using OCR
• Split a long PDF into smaller chunks for your AI
• Pull structured data out of invoices or receipts
• Check which pages an extractor silently skipped
• Compare two extractions of the same document
• Get a confidence score for each page

Try asking your AI

• “Read this PDF and give me the main points”
• “Extract the table on page 3 of report.pdf as text”
• “Turn this scanned invoice into structured data”
• “Check whether this extractor dropped any pages from my PDF”

What it gives back to you

You get back the text of the PDF, usually as Markdown or as a structured list of fields. You also get a confidence score for the document and, if any pages looked bad, a note about which ones. In the chat, this shows up as readable text or a small table you can copy. If you asked it to check another tool's work, it tells you which pages were silently dropped.

Before you start

Who it's for

Anyone who works with lots of PDFs and wants their AI to read them properly, like analysts, researchers, accountants or operations staff.