pdf-mcp

Full-stack PDF intelligence MCP server — extract, manipulate, annotate, convert, validate, and RAG-search PDFs through a unified tool surface and React workbench.

v0.1.0 · Ports 11130 (frontend) / 11131 (backend)

FastMCP 3.4+ · PyMuPDF + pypdf + pdfplumber · LanceDB RAG · Prefab UI · dual transport (stdio + HTTP)

Features

• pdf_extract — text, images, tables, metadata, fonts, links, outline

• pdf_manipulate — merge, split, rotate, reorder, delete pages, compress, encrypt/decrypt, optimize

• pdf_annotate — watermark, stamp, highlight, underline, header/footer, page numbers

• pdf_forms — list / fill / flatten / export form fields

• pdf_convert — PDF ↔ Markdown / images / HTML

• pdf_validate — PDF/A, structure, accessibility, integrity, compare

• pdf_rag — chunk, index (LanceDB), semantic search

Related MCP server: pdf-mcp

Quick start

git clone https://github.com/sandraschi/pdf-mcp
cd pdf-mcp
uv sync
Copy-Item .env.example .env
.\start.ps1

Dashboard: http://127.0.0.1:11130 · MCP/API: http://127.0.0.1:11131

Webapp

Architecture

• Backend: Python 3.12, FastMCP 3.4+, FastAPI, PyMuPDF, pypdf, pdfplumber, LanceDB

• Frontend: React 18, Vite, Tailwind, Zustand, Framer Motion, PDF.js

• Transport: stdio (Claude Desktop / Cursor) + HTTP/SSE

License

MIT