ArXiv Docker Hub MCP Server

Projects | | Links: Docker Hub | GitHub | MCP Hub

ArXiv Docker Hub MCP Server

A Model Context Protocol (MCP) server that connects LLMs to arXiv papers through a clean API. Built with Docker-first architecture and published on the official Docker MCP Hub.

The ArXiv MCP Server bridges the gap between Large Language Models and academic research by providing structured access to arXiv papers through the Model Context Protocol. Built with a Docker-first approach, it’s designed for easy deployment and integration with AI assistants.

Key Features

🔍 Paper Discovery

  • Search arXiv by keywords, authors, or categories
  • Filter by publication date and relevance
  • Access to 2M+ research papers

📄 Content Access

  • Download and parse PDF content
  • Extract abstracts, titles, and metadata
  • Generate structured summaries for LLM consumption

🐳 Docker-First Architecture

  • One-command deployment with Docker
  • Persistent storage for downloaded papers
  • Environment-based configuration
  • Production-ready container setup

🔗 MCP Integration

  • Official MCP protocol compliance
  • Works with Claude, ChatGPT, and other MCP-compatible LLMs
  • Structured tool definitions for reliable AI interaction

Technical Stack

  • Language: Python 3.11
  • Framework: FastAPI for the MCP server
  • Storage: Local filesystem with Docker volumes
  • PDF Processing: PyPDF2 for content extraction
  • API: arXiv API for paper metadata and downloads
  • Containerization: Docker with multi-stage builds

Architecture

┌─────────────────┐    ┌─────────────────┐    ┌─────────────────┐
│   LLM Client    │───▶│  MCP Server     │───▶│   arXiv API     │
│  (Claude, etc.) │    │  (Docker)       │    │                 │
└─────────────────┘    └─────────────────┘    └─────────────────┘
                              │
                              ▼
                       ┌─────────────────┐
                       │  Local Storage  │
                       │  (Papers Cache) │
                       └─────────────────┘

Getting Started

Quick Start with Docker

# Pull and run the official image
docker run -v $(pwd)/papers:/mnt/papers \
  -e ARXIV_STORAGE_PATH=/mnt/papers \
  mcp/arxiv-mcp-server

Integration with Claude

{
  "mcpServers": {
    "arxiv": {
      "command": "docker",
      "args": ["run", "--rm", "-v", "./papers:/mnt/papers", 
               "-e", "ARXIV_STORAGE_PATH=/mnt/papers", 
               "mcp/arxiv-mcp-server"]
    }
  }
}

Use Cases

📚 Research Assistant

  • “Find recent papers on transformer architectures”
  • “Summarize the key findings from this arXiv paper”
  • “What are the latest developments in multi-agent systems?”

🔬 Literature Review

  • Systematic paper discovery and analysis
  • Trend identification across research areas
  • Citation and reference tracking

🤖 AI-Powered Research

  • LLM-assisted paper analysis
  • Automated research summaries
  • Intelligent paper recommendations

Development Journey

This project started as a fork of an existing arXiv MCP server that lacked Docker support. Through iterative development with AI assistance (primarily Cursor and Claude), I:

  1. Dockerized the application with proper volume mounts and environment configuration
  2. Simplified the deployment process to a single Docker command
  3. Added production-ready features like proper logging and error handling
  4. Submitted to Docker MCP Hub and achieved official publication status

The entire development process took about 10 days, demonstrating the power of AI-assisted development for rapid prototyping and deployment.

Impact

  • Published on Docker MCP Hub as an official, verified container
  • Zero-friction deployment for researchers and AI enthusiasts
  • Active community usage with positive feedback from early adopters
  • Template for future MCP projects showing Docker-first best practices

What I Learned

  • Docker containerization patterns for Python applications
  • MCP protocol implementation and tool definition standards
  • AI-assisted development workflows using Cursor and Claude
  • Open source publishing processes for Docker Hub and MCP ecosystem

Want to see how this was built? Check out the detailed build log for a complete walkthrough of the development process.