MCPIndependentjztan
pdf-mcp
Searches a PDF or a whole folder and cites source pages; reads tables and scans.
The server lets the agent search one PDF or a whole folder and read only the pages it needs, with every excerpt returned with its page number. It extracts tables, images and scanned text (OCR) and handles multi-column and Japanese layouts. Hidden text such as injected instructions is flagged instead of passed on. It requires Python 3.11–3.14.
Installation
Manual install
$uv tool install pdf-mcp && claude mcp add pdf-mcp -- pdf-mcpDetails
- Works in
- Claude Code, Claude, VS Code, Cursor, Codex
- Transport
- local
- License
- MIT
- Tasks
- Docs, Research
- Rating
- Source
- github.com/jztan/pdf-mcp (opens in a new tab)
Tags
- rag
- search
- ocr
- tables
- citations
A pack is your own set of skills and MCP servers. sciver loads it in your AI with one command.