Doc Reading and Converter
A server for reading and converting documents between PDF, DOCX, and Markdown formats using marker-pdf and pandoc.
Document Reading and Converter Tool
The goal is to have a tool that can read and edit files, using Cursor or Claude Desktop.
The filesystem MCP already allows reading, writing and editing documents, but is limited in that it cannot:
- Read pdfs or docx files
- Reads the full document, which can cause context overflow
To address these issues, we create:
- A fresh
doc-reading-mcpmcp service allowing for document conversions between pdf, docx and markdown.
Features
- PDF to Markdown conversion using marker-pdf
- DOCX to Markdown conversion using pandoc
- Markdown to DOCX conversion using pandoc
- Markdown to PDF conversion using pandoc
Prerequisites
- Python 3.10 or higher
- [pandoc] (https://pandoc.org/installing.html) installed on your system
- [uv] (https://docs.astral.sh/uv/) for Python package management
Installation
uv add doc-reading-mcp
Or install directly from the source:
uv add git+https://github.com/mffrydman/doc-reading-mcp.git
Usage
Run as a standalone server
uv run mcp install -m doc_reading_mcp
Or run directly:
uv run -m doc_reading_mcp
Use with MCP Inspector
npx @modelcontextprotocol/inspector uvx run -m doc_reading_mcp
Configure in Claude Desktop, Cursor or Windsurf
Add this to your MCP Configuration:
{
"mcpServers": {
"doc-reading-mcp": {
"command": "uvx",
"args": [
"doc-reading-mcp",
]
}
}
}
Run in Cursor/Windsurf/Claude using the following configuration:
"doc-reading-mcp": {
"command": "uv",
"args": [
"--directory",
"/absolute/path/to/mffrydman/doc-reading-mcp",
"run",
"-m",
"doc_reading_mcp"
]
}
Replace
/absolute/path/to/with the actual path on your system.
Related Servers
Qonto
Interact with the Qonto Business API to manage finances, transactions, and account information using API credentials.
PDF.co MCP
Perform various PDF processing tasks using the PDF.co API.
MCP Hub
A hub server that connects to and manages other MCP servers.
Cursor Admin
Monitor team productivity, usage, and spending for AI assistants using the Cursor Admin API.
MCP Task Orchestrator
A Kotlin MCP server for comprehensive task management, allowing AI assistants to interact with project data.
Penpot MCP Server
Integrates AI language models with the Penpot design platform to automate design workflows.
Jira MCP Server by CData
A read-only MCP server for Jira, enabling LLMs to query live Jira data using the CData JDBC Driver.
Folderr MCP Server
Interact with the Folderr API to manage and communicate with Folderr Assistants.
Fider
Interact with Fider, an open-source customer feedback tool, to manage user suggestions and feedback.
ClaudeKeep
Save and share AI conversations from Claude Desktop.