pdf-mcp
Active·★ 155·MIT·Updated 2026-10-13
pdf-mcp is a Model Context Protocol (MCP) server that enables AI agents to read, search, and extract content from PDF files. It uses PyMuPDF for PDF parsing, SQLite for persistent caching, and supports hybrid search combining BM25 keyword and semantic embeddings, OCR for scanned documents, and structured extraction of tables and images.
#agentic-ai#ai#claude#codex-cli#copilot#document-processing#llm#mcp
01
Features
01Hybrid search (BM25 keyword + semantic embeddings) with Reciprocal Rank Fusion
02Paginated reading to avoid context overflow
03OCR support for scanned and image-based PDFs via Tesseract
04Structured extraction of tables, images, and table of contents
05Persistent SQLite cache with automatic invalidation
02
Compatibility
Claude Code
Claude Code
Verified via docs
Claude Desktop
Claude Desktop
Verified via docs
Visual Studio Code
VS Code
Verified via docs
Codex CLI
Codex CLI
Verified via docs
Kiro
Kiro
Verified via docs
03
Quick start
1
$ pip install pdf-mcp
04
Use cases
↳Efficiently read and analyze large PDF documents without exceeding context limits
↳Search for specific content or concepts within PDFs using natural language
↳Extract structured data such as tables and images from PDFs
05
Alternatives
openrouter-mcp-multimodal★ 99
MCP server for OpenRouter providing text chat and image analysis tools
word-mcp-live★ 239
The only MCP server that edits Word documents while they're open — 114 tools, live editing, tracked changes, per-action undo
rails-ai-context★ 165
Auto-expose your Rails app to AI assistants via MCP — zero config, 28 introspectors, 11 read-only tools, 37% fewer tokens. Works with Claude Code, Cursor, Windsurf & Copilot.
gnosis-mcp★ 29
Zero-config MCP server for searchable documentation (SQLite default, PostgreSQL optional)
pixelpanda-mcp★ 21
MCP server for PixelPanda — 31 AI image processing tools for Claude Desktop, Cursor & other MCP clients
Comments
Log in to leave a comment
No comments yet. Be the first!