by jztan
pdf-mcp is an MCP server that lets AI agents search one PDF or a folder of PDFs by keyword or meaning and read only the pages they need. It also extracts tables, charts and scanned text.
PluginContact for pricing
Step 1: In Claude Desktop, download the pdf-mcp.mcpb file and drop it onto Settings > Extensions, then click Install.
Step 2: For Claude Code and other clients, install with uv tool install pdf-mcp or pipx install pdf-mcp (Python 3.11 or later).
Step 3: Register it, for example with claude mcp add pdf-mcp -- pdf-mcp. Other clients are covered in the project's docs/clients.md.
Step 4: Install Tesseract if you need OCR on scanned pages.
Step 5: Ask your agent to read a PDF by path or a folder, for example which pages discuss supply chain risks.
Yes. pdf_corpus_warm, pdf_corpus_overview and pdf_corpus_search work across a folder and return document and page provenance.
Yes, through OCR with Tesseract, which is parallel across pages. Tesseract must be installed separately for the pip route.
Yes. pdf-mcp-http serves the same tools over HTTP and requires an auth token or an allow list, otherwise it exits rather than serve an open endpoint.
Text from a PDF is treated as untrusted content. pdf_info with content_trust=True reports hidden text a reader cannot see, and the read tools flag it per page without stripping anything.
Requirements
Zotero MCPZotero MCP connects a Zotero research library to AI assistants so they can search, read, annotate and add to it. It also ships a chat panel that runs inside Zotero.Sage MCPSAGE is a governed memory layer for AI agents. Each memory is signed by an agent, admitted through a review step, limited by permissions and kept with an audit trail.
Stratify MCPStratify is an MCP-based backtester for NIFTY and SENSEX options. You describe a strategy in plain words to Claude, ChatGPT or Gemini and get a costed result and a shareable report.