hermes-agent
356e476d1d
Tests / test (push) Successful in 40s
Render only a cropped region of a PDF page via PyMuPDF clip rect. Coordinates are normalized 0-1 relative to the page, matching the bbox returned by zotero_get_page_layout. Keeps rendered figure images small so they can be embedded inline as base64 data URIs.
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