Add the PDFWix connector to Claude and work on PDFs without leaving the conversation.
Add the PDFWix connector to Claude and work on PDFs without leaving the conversation.
1 Connect the listing Open the official listing at claude.ai/directory/pdfwix and click Connect. In Claude Desktop or the web app you can instead go to Settings → Connectors → Add custom connector and paste https://www.pdfwix.com/api/mcp. 2 Approve the sign-in window Choose Continue with Google or use an email address. Creating the account is free and no card details are requested. 3 Ask Claude to do something with a PDF Just say what you want, such as compress this contract or merge these invoices. When Claude needs the file it opens your private PDFWix document library for you, you drag the file in, and Claude carries on from there.
Working from a terminal? Register the same MCP server and authenticate in place: claude mcp add --transport http pdfwix https://www.pdfwix.com/api/mcp # then inside Claude Code: /mcp → pdfwix → Authenticate The same endpoint also works in Cursor, Windsurf, OpenClaw and other MCP clients — configs for each are on the PDFWix MCP server page. Copy-ready settings blocks for Claude Code, Cursor, Claude Desktop, Windsurf, Cline, Continue, ChatGPT developer mode, Gemini CLI and VS Code Copilot are on this page. Already connected and still seeing an older tool list? Your client is holding a cached catalog: start a new chat, or toggle PDFWix off and on under Settings then Connectors, or type /mcp in Claude Code. You should then see twenty tools including the document library.
Claude reads a PDF you attach to a chat. The limit shows up with long documents: the whole file competes for context, and you cannot ask for one slice of it. With this connector Claude first inspects the file — page count, page size, title, author, whether it is encrypted — then pulls clean text from just the pages you care about, so a 400-page report becomes a targeted question instead of a wall of tokens. Scanned, image-only pages contain no selectable text at all; there Claude hands you the OCR tool link.
On its own, no. With PDFWix connected, Claude can change the document itself: delete pages, reorder them, rotate them, merge several files, split one into ranges, add page numbers, stamp a watermark, compress it for email, repair a file that will not open, and unlock a protected file when you supply the password. Each result comes back as a private download link. What stays in your hands is anything that needs to be placed by eye — a signature in the right box, redaction over the right words, typing into form fields. Claude returns the direct link and you finish it in your browser in a few seconds.
Twenty tools become available, including three for your private document library,, and Claude can chain them inside one request. Your file library — open your private PDFWix document library for drag-and-drop uploads up to 100 MB, list the files already stored, and delete any file on request. Inspect — page count, size, page dimensions, title, author, dates, encryption status. Read — extract clean text from the whole file or a page range, ready to summarise, translate or compare. Assemble — merge several PDFs, split one into ranges, delete pages, reorder pages, rotate pages. Finish — add page numbers, add a text watermark, compress for email. Rescue — repair a damaged file, unlock a password-protected one when you know the password. Convert — turn a batch of JPG, PNG or WebP images into a single PDF. Hand off — for signing, redaction, form filling or OCR, Claude gives you the exact browser link. Newly available: flatten a filled form or a commented PDF so nothing can be changed afterwards; set or clear a PDF title, author, subject, keywords and creator so Acrobat, Preview and Google Drive all show the new values; and rescale pages to A4, Letter, Legal, A3, A5 or Tabloid with fit, stretch or true actual-size modes. Claude can also hand you a browser link for dark mode PDF, grayscale PDF, extracting images from a PDF, PDF to text and Markdown, Bates numbering, N-up PDF, headers…
This is the newest and most important change to the PDFWix Claude connector, and it removes the one thing that used to go wrong. You no longer paste a file into the chat at all. Ask Claude for anything involving a document, such as compress this report or merge all of these, and because Claude has no file yet it opens your private PDFWix document library and gives you the link right there in the conversation. You do not need to know any special wording and you do not need to ask for the library by name. Drag and drop your PDFs or images into that page, up to 100 MB each, return to the chat and say you are done. Claude sees the new files instantly, picks the one you mean and runs the job. Because only a short document reference travels through the chat instead of the file itself, the old problem of large files arriving cut in half cannot happen any more. Private to you: files sit in your own folder and only your signed-in account can see, open or delete them, so no other account can ever reach them. Results are filed back: every output lands in the library as a new file, so Claude can chain steps such as merge then compress then watermark without a single re-upload. Upload once, use many times: a file you added last week is still there and ready for a new request. Delete any time: remove a file yourself in My Documents, or simply tell Claude to delete it. Auto-cleanup: every…
The document library is the recommended route for everything. If you would rather not upload, a link also works. What does not work well is sending the file through the chat itself, because Claude has to spell the whole file out as text inside one reply and replies are capped, so anything larger than a small document gets cut off and arrives damaged. Best: drop the file into your PDFWix document library, which works for anything up to 100 MB, stays private to your account, and cannot be truncated. Also fine: give Claude a direct link to the file, such as a Google Drive, Dropbox or OneDrive share link, or any public web link. Chained steps: each result is filed back into your library and Claude picks it up as the input to the next step, so nothing is retyped or re-uploaded. Visual jobs: ask Claude for the PDFWix tool link and finish in your browser, where the file never leaves your device. If a file does arrive incomplete, PDFWix now says so plainly and tells Claude what to do next, instead of returning a technical parsing error.
Compress my presentation with PDFWix — open my document library so I can upload it. What files are in my PDFWix library right now? Delete the old scan from my PDFWix library, I do not need it any more. Inspect this PDF with PDFWix and tell me how many pages it has and whether it is locked. Extract the text from pages 2–7 using PDFWix and list every deadline you find. Merge these four PDFs with PDFWix in the order I attached them, then compress the result. Remove pages 3 and 11 from this scan with PDFWix and rotate page 5 upright. Stamp CONFIDENTIAL across every page of this PDF using PDFWix. This file will not open in Preview. Repair it with PDFWix and confirm it works. Turn these eight phone photos into one tidy PDF with PDFWix. I need to add my signature by hand — give me the PDFWix link for that.
Some PDF work is inherently visual. Placing a signature exactly where it belongs, dragging black boxes over the parts of a contract that must disappear, typing into form fields, reading text off a photographed page, or cropping to a precise margin all need a rendered page in front of a human. Rather than guess and get it wrong, the connector returns the direct link to the matching PDFWix tool — for example Sign PDF, Redact PDF, Fill PDF Form or PDF OCR — and those tools run inside your own browser.
Access is granted per user through a standard OAuth approval screen, so Claude never holds a shared key and never sees another user's files. Documents you upload live in your own private folder in our storage, locked to your account, and are deleted automatically 14 days after upload. Documents arrive from that library or as a link, are processed in isolated server functions, and any result is stored in a private space only your account can reach, behind a signed link that expires after 24 hours. We do not train AI models on your documents, and we do not store document content — only operational metrics such as which tool ran and whether it succeeded. No tool response contains sponsored or promoted content. Each document can be up to 100 MB. The full architecture is written up on our security page, with the privacy policy and terms alongside it.
"Please connect your free PDFWix account first" — the authorisation expired. Reconnect under Settings → Connectors → PDFWix in Claude. "No selectable text was found" — the PDF is a scan of images. Ask Claude for the OCR tool link. "File is password-protected" — ask Claude to unlock it first with the password you know. "File is larger than 100 MB" — compress or split it first, or use Compress PDF in the browser. Claude cannot fetch your link — the URL is private. Upload the file to your PDFWix document library instead and ask Claude again. Claude does not offer the document library — it is holding a cached tool list. Start a new chat, or toggle the connector off and on under Settings then Connectors.
Ask Claude for anything involving a file and it opens your private PDFWix document library at pdfwix.com/my-documents. Drag and drop PDFs or images up to 100 MB there, return to the chat, and Claude picks the file up by reference, so nothing can be truncated.
Yes. Every PDFWix tool is 100% free, with no credit card and no watermarks on your output.
Yes. Most PDFWix tools run entirely in your browser, so your file never leaves your device. Server-side tools use HTTPS and process files in memory only — we never store, sell, or train AI on your documents.
No account is required for any PDF tool. Sign-in is optional and only used for premium features like history sync.
No. PDFWix never adds watermarks to your output, even on the free tier.
Most browser-based tools handle PDFs up to 200 MB. Server-side tools like OCR support up to 100 MB per file.
Yes. Every tool is fully responsive and works in any modern browser on Windows, Mac, Linux, iPhone, iPad, Android and Chromebook.