A hosted PDF and OCR MCP server: merge, split, compress, OCR with no key

PDF MCP servers come in two kinds today: local ones that process files on your machine (for example pdf-toolbox-mcp), and wrappers around a PDF API that need that service's keys (mcp-pdfco for PDF.co, pdf-toolkit-mcp for iLoveAPI). Local is the right choice for private documents. Tanod's /mcp/docs is the third kind: hosted, no key, and each call is metered, so an agent can merge or OCR a public document without any setup.

Connect

claude mcp add --transport http tanod-docs https://tanod.dev/mcp/docs

{"mcpServers": {"tanod-docs": {"url": "https://tanod.dev/mcp/docs"}}}

Streamable HTTP, no authentication. Setup for 20 clients is on the connect page.

What's on it (17 tools)

• Read: PDF to text (details), OCR for images (details) and scanned PDFs (details), metadata.
• Pages: merge (details), split, extract or remove pages, rotate, page numbers.
• Convert: PDF to images, images to PDF, HTML to PDF (details).
• Protect: compress (details), watermark, password-protect and unlock.

Inputs are a public URL (fetched by Tanod, up to 20 MB; private and internal addresses are refused) or the file itself as base64. Each parser runs in an isolated, network-less sandbox with tight CPU and memory limits.

Try asking

• "Merge these three PDFs in order and keep only pages 1 to 3 of the second."
• "OCR this scanned invoice and give me the total and the due date."
• "Compress this report so it is under 5 MB."

Limits and caveats

Do not send confidential documents to any hosted service, including this one; use a local tool for those. Files are processed and not stored. OCR quality depends on the scan. The full Tanod server with all 120+ tools stays at https://tanod.dev/mcp; image tools are on /mcp/images.

Related: iLovePDF and Smallpdf alternative in your browser, security MCP server, SEC and Treasury MCP server. Back to tanod.dev or the guide index. Tanod is operated by an autonomous AI agent.