PDFForge

Runs a local PDF toolkit so an assistant can read, edit, OCR, redact, fill forms, compare and convert PDFs without uploading files.

What you can do

PDFForge exposes a local PDF engine to MCP clients. Tools cover reading and searching documents, rendering pages as images, merging, splitting and reorganising pages, creating PDFs from Word, Excel or images, and exporting PDFs to .docx, .txt, .html, .odt or .xlsx. It can replace text using the document's embedded fonts, OCR scans with per-word confidence, find and permanently remove IBANs, card numbers, national IDs and emails, fill AcroForms, extract bank statement transactions into Excel, compare versions and detect tampering, and apply watermarks, AES-256 protection or compression.

When to use it

Worth adding when an assistant needs to work on PDFs on the same machine, especially for bulk redaction of sensitive data, turning scanned documents into searchable text, extracting statement lines into spreadsheets, or checking whether a contract changed. It suits privacy-sensitive work because processing stays local.

Requirements

Runs as a local process over stdio; the manifest declares no authentication, environment variables or headers. Install from PyPI as pdfforge-pro and launch with uvx or the pdfforge-mcp command. OCR additionally needs Tesseract installed on the system. Desktop only; no web executable.