Skip to content

The Best MCP Servers for Document Processing in 2026

6 min read | Last updated April 3, 2026 | PdfBroker.io

The MCP ecosystem has grown to over 4,000 servers, and document processing is one of the most practical use cases. AI agents that can generate PDFs, convert between formats, and manipulate existing documents save developers from writing boilerplate integration code.

This guide covers the MCP servers worth installing for document workflows — from cloud PDF generation to local file manipulation. Each entry includes what it does well, where it falls short, and how to set it up.

What to Look For

When evaluating document processing MCP servers, the key differentiators are:

Cloud vs. local rendering. Cloud servers handle the rendering infrastructure for you — no local dependencies, no font management, no container setup. Local servers give you more control but require you to maintain the rendering stack on every machine.

Output quality. PDF generation quality varies dramatically. Markdown-to-PDF tools produce functional but plain documents. HTML-to-PDF engines with CSS Paged Media support produce print-ready, professional output with headers, footers, and page numbers.

Compliance support. If you need PDF/A (archival) or PDF/UA (accessibility) compliance — particularly relevant since the European Accessibility Act — verify the server supports it. Most local PDF tools don't.

Cloud-Based PDF Generation

PdfBroker.io

What it does: Cloud-based HTML-to-PDF generation with two rendering engines (WeasyPrint and wkhtmltopdf), PDF merging, PDF-to-image conversion, and text overlay on existing PDFs.

Why it stands out: PdfBroker.io is currently the only cloud-based PDF generation service available as an MCP server. The WeasyPrint engine supports full CSS Paged Media (headers, footers, page numbers via @page margin boxes) and compliance standards (PDF/A-1b, PDF/A-3b, PDF/UA-1). The server returns time-limited download URLs instead of raw PDF bytes, keeping the AI agent's context window clean.

Tools exposed:

  • html_to_pdf — WeasyPrint rendering with compliance options
  • html_to_pdf_wk — wkhtmltopdf with JavaScript support
  • merge_pdfs — Combine multiple PDFs
  • pdf_to_image — Convert pages to PNG/JPEG
  • write_text_on_pdf — Overlay text on existing documents

Setup (Claude Code):

claude mcp add-json pdfbroker '{
  "type": "http",
  "url": "https://mcp.pdfbroker.io/",
  "headers": {
    "X-PdfBroker-ClientId": "YOUR_CLIENT_ID",
    "X-PdfBroker-ClientSecret": "YOUR_CLIENT_SECRET"
  }
}'

Setup (Claude Desktop, Cursor, VS Code):

{
  "mcpServers": {
    "pdfbroker": {
      "command": "npx",
      "args": ["-y", "@pdfbroker/mcp-server"],
      "env": {
        "PDFBROKER_CLIENT_ID": "YOUR_CLIENT_ID",
        "PDFBROKER_CLIENT_SECRET": "YOUR_CLIENT_SECRET"
      }
    }
  }
}

Best for: Professional documents (invoices, contracts, reports), compliance-sensitive documents (PDF/A, PDF/UA), and any workflow where you don't want to manage local PDF rendering infrastructure.

Pricing: Free tier (200 req/month, wkhtmltopdf only), paid plans from €8.95/month for all engines.

Website: pdfbroker.io

Pdfcrowd MCP Server

What it does: Cloud HTML-to-PDF conversion via the Pdfcrowd API.

Why consider it: Established PDF API with its own rendering engine. Good for high-volume document generation.

Limitations: No PDF/A or PDF/UA compliance. No PDF merging or manipulation tools. Requires a Pdfcrowd subscription.

Best for: High-volume HTML-to-PDF conversion where compliance isn't required.

Local PDF Generation

Markdown2PDF

What it does: Converts Markdown content to PDF using a bundled Chrome instance (Puppeteer).

Why consider it: Zero-config for Markdown-to-PDF conversion. Supports syntax highlighting, custom styling, watermarks, and page numbers.

Limitations: Markdown input only (no raw HTML/CSS control). Bundles a full Chrome installation (~300MB). No compliance standards. Quality is limited compared to dedicated PDF engines.

Setup:

{
  "mcpServers": {
    "markdown2pdf": {
      "command": "npx",
      "args": ["-y", "markdown2pdf-mcp"],
      "env": {
        "M2P_OUTPUT_DIR": "/path/to/output"
      }
    }
  }
}

Best for: Quick document generation from Markdown notes, README files, or AI-generated Markdown content.

PDFKit MCP Server (mcp-z)

What it does: Generates PDFs programmatically using PDFKit — the AI agent specifies text, headings, images, dividers, and positioning directly.

Why consider it: No HTML required. The AI agent works with a simple document API (add heading, add paragraph, add image). Good for structured, data-driven documents.

Limitations: Layout control is limited compared to CSS. No CSS Paged Media, no print stylesheets. No compliance standards. Local-only.

Best for: Simple structured documents where the AI agent knows the exact layout — data tables, simple reports, form letters.

Playwright MCP (PDF capability)

What it does: The Playwright MCP server includes a browser_pdf_save capability that captures the current browser page as a PDF.

Why consider it: Useful if you're already using Playwright MCP for browser automation. Captures any web page as a PDF, including JavaScript-rendered content.

Limitations: Requires navigating to a URL first — not a standalone PDF generation tool. Quality depends on the page's print stylesheet. No compliance standards.

Best for: Capturing web pages or web app screens as PDF snapshots within a broader browser automation workflow.

PDF Reading & Manipulation

PDF Reader MCP (Sylphx)

What it does: Reads, extracts text, and processes PDF files. Supports parallel processing for large document sets.

Why consider it: Production-grade PDF reading with 94%+ test coverage. Supports both local files and URLs. HTTP transport available for remote access.

Limitations: Read-only — no PDF generation or modification.

Best for: Document analysis, text extraction, and processing existing PDFs.

PDF Tools MCP (hanweg)

What it does: Merges PDFs and extracts pages. Can find and merge related PDFs based on text pattern matching.

Why consider it: The pattern-matching merge feature is unique — find all PDFs containing a specific invoice number and merge them.

Limitations: Basic feature set. No generation, no compliance, no text overlay.

Best for: PDF file management — merging, splitting, and organizing existing documents.

Comparison Table

Server Type HTML→PDF Markdown→PDF PDF/A PDF/UA Merge JS Support Free Tier
PdfBroker.io Cloud WeasyPrint + wkhtmltopdf — Yes Yes Yes wkhtmltopdf 200 req/mo
Pdfcrowd Cloud Yes — No No No Yes Trial
Markdown2PDF Local — Yes No No No — Open source
PDFKit MCP Local — — No No No — Open source
Playwright Local Page capture — No No No Yes Open source
PDF Reader Local — — — — — — Open source
PDF Tools Local — — No No Yes — Open source

Recommendations

For professional document generation: PdfBroker.io. It's the only option with cloud rendering, compliance standards, and a full PDF toolkit in a single MCP server. The AI agent generates HTML (which LLMs are excellent at), and WeasyPrint produces professional output.

For quick Markdown-to-PDF: Markdown2PDF. Simple, works well for documentation and notes.

For browser page capture: Playwright MCP with the PDF capability. Best when PDF generation is a side feature of a broader automation workflow.

For PDF analysis and extraction: PDF Reader MCP (Sylphx). Production-grade and well-tested.

For most developers: Start with PdfBroker.io for generation and PDF Reader for analysis. That covers the full lifecycle — create PDFs and read existing ones.

Try PdfBroker.io free

Generate your first PDF in minutes — no credit card required.

Get Started Free