1
Add it to Claude Code
claude mcp add pdf-mcp -- uvx --from git+https://github.com/I-CAN-hack/pdf-mcp.git pdf-mcp2
Make your agent remember this setup
pdf-mcp's config, env vars, and the gotchas you hit — recalled in every future Claude Code, Cursor, and Codex session.
npx conare@latestFree · one command · indexes the sessions already on disk. Set up in the browser instead →
What it does
- Extract text in multiple formats including JSON, text, markdown, and HTML
- Render PDF pages as PNG images with configurable DPI
- Perform case-insensitive full-text search across PDF documents
- Retrieve document metadata and table of contents
- Optionally exclude headers and footers during text extraction
Tools 5
get_pdf_infoGet metadata about a PDF (page count, author, title, etc.)get_table_of_contentsGet the outline/bookmarks with page numbers for each sectionget_page_textExtract text from a page range in json, text, markdown, or html format.get_page_imageRender a single page as a PNG image, returned as base64 or written to a temp file.search_textCase-insensitive text search across the entire PDF, returning page numbers and surrounding contextTry it
→Extract the text from pages 5 to 10 of the document 'manual.pdf' in markdown format.
→Search for the term 'error code 404' in 'datasheet.pdf' and tell me which page it appears on.
→Get the table of contents for 'report.pdf' to understand the document structure.
→Render page 1 of 'diagram.pdf' as an image so I can analyze the technical drawing.
→What is the author and total page count of 'research_paper.pdf'?
Original README from I-CAN-hack/pdf-mcp
pdf-mcp
An MCP server for reading, rendering, and searching PDF files. Built with PyMuPDF and PyMuPDF4LLM.
Designed for use with LLMs that need to read datasheets and other PDFs containing diagrams, tables, and technical content.
Tools
| Tool | Description |
|---|---|
get_pdf_info |
Get metadata about a PDF (page count, author, title, etc.) |
get_table_of_contents |
Get the outline/bookmarks with page numbers for each section |
get_page_text |
Extract text from a page range in json (default), text, markdown, or html format. Optionally exclude headers/footers |
get_page_image |
Render a single page as a PNG image, returned as base64 or written to a temp file. Configurable DPI (default 150) |
search_text |
Case-insensitive text search across the entire PDF, returning page numbers and surrounding context |
All requests are stateless and take the PDF filename as a parameter.
Setup
Add the following to your .mcp.json:
{
"mcpServers": {
"pdf-mcp": {
"command": "uvx",
"args": ["--from", "git+https://github.com/I-CAN-hack/pdf-mcp.git", "pdf-mcp"]
}
}
}
Or for codex run codex mcp add pdf-mcp -- uvx --from git+https://github.com/I-CAN-hack/pdf-mcp.git pdf-mcp.
This will automatically install and run the server using uvx.
Development
# Install dependencies
uv sync
# Generate test PDFs
uv run python assets/generate.py
# Run tests
uv run pytest tests/ -v
# Run the server locally
uv run pdf-mcp