pdf-mcp
An agentic RAG MCP server that enables AI agents to search and read large PDFs or folders using hybrid semantic and keyword search. It supports selective page reads, table extraction, OCR for scanned documents, and handles multi-column and CJK layouts with optional CUDA acceleration.
Package and access
Local package with source-attributed installation metadata. Koskamo does not execute packages during review.
What you can do with it
- Search large PDFs or folders using hybrid semantic and keyword search
- Extract structured tables and chart data from documents
- Perform OCR on scanned PDF pages with parallel processing
- Read specific pages from multi-column or Japanese layout PDFs
Source and maintenance
- Maintenance checked
Packages and installation
Source-provided package data. Koskamo has not executed or live-probed this package.
pdf-mcp 3.6.0
{
"version": "3.6.0",
"transport": "stdio",
"identifier": "pdf-mcp",
"registryType": "pypi",
"registryBaseUrl": "https://pypi.org",
"environmentVariables": [
{
"name": "PDF_MCP_CACHE_DIR",
"format": "string",
"description": "Directory for storing PDF cache (default: ~/.cache/pdf-mcp)"
},
{
"name": "PDF_MCP_CACHE_TTL",
"format": "number",
"description": "Cache time-to-live in hours (default: 24)"
}
]
}Tools
Reported by the attributed source; not independently observed by Koskamo.
Connect from your client
Each snippet configures the same source-provided package; replace placeholders before use.
Cursor
{
"mcpServers": {
"pdf-mcp-mcp": {
"command": "uvx",
"args": [
"pdf-mcp==3.6.0"
],
"env": {
"PDF_MCP_CACHE_DIR": "<YOUR_PDF_MCP_CACHE_DIR>",
"PDF_MCP_CACHE_TTL": "<YOUR_PDF_MCP_CACHE_TTL>"
}
}
}
}Cline
{
"mcpServers": {
"pdf-mcp-mcp": {
"command": "uvx",
"args": [
"pdf-mcp==3.6.0"
],
"env": {
"PDF_MCP_CACHE_DIR": "<YOUR_PDF_MCP_CACHE_DIR>",
"PDF_MCP_CACHE_TTL": "<YOUR_PDF_MCP_CACHE_TTL>"
}
}
}
}VS Code
{
"servers": {
"pdf-mcp-mcp": {
"type": "stdio",
"command": "uvx",
"args": [
"pdf-mcp==3.6.0"
],
"env": {
"PDF_MCP_CACHE_DIR": "<YOUR_PDF_MCP_CACHE_DIR>",
"PDF_MCP_CACHE_TTL": "<YOUR_PDF_MCP_CACHE_TTL>"
}
}
}
}Claude Desktop
{
"mcpServers": {
"pdf-mcp-mcp": {
"command": "uvx",
"args": [
"pdf-mcp==3.6.0"
],
"env": {
"PDF_MCP_CACHE_DIR": "<YOUR_PDF_MCP_CACHE_DIR>",
"PDF_MCP_CACHE_TTL": "<YOUR_PDF_MCP_CACHE_TTL>"
}
}
}
}Let your agent find its tools
Find servers like pdf-mcp from your agent. Add Koskamo once to search the index for endpoints, access requirements, and client setup. Public discovery needs no key.