arXiv Search MCP Server

A Model Context Protocol (MCP) server that enables natural language searching of arXiv academic papers. Built with Node.js and TypeScript for local execution.

Features

- Natural Language Queries: Search for papers using conversational language (English and Japanese)
- Comprehensive Search: Search by keywords, authors, categories, and date ranges
- Structured Results: Get detailed paper information including titles, authors, abstracts, and links
- MCP Protocol Compliance: Works with Claude Desktop and other MCP-compatible clients
- Error Handling: Robust error handling and retry logic for API calls
- Real-time Data: Fetches live data from arXiv API

Quick Start

1. Clone and Setup

   git clone https://github.com/sodeyama/arxiv-mcp-server.git

   cd arxiv-mcp-server

   npm install

   npm run build

   

2. Configure Claude Desktop

Add to your Claude Desktop config file:

   {

     "mcpServers": {

       "arxiv-search": {

         "command": "node",

         "args": ["/absolute/path/to/arxiv-mcp-server/dist/index.js"],

         "env": {}

       }

     }

   }

   

3. Usage Examples
- "最新の深層学習研究を5件見つけて"
- "Geoffrey Hintonの機械学習論文"
- "量子コンピューティングに関する2023年以降の研究"
- "RAG retrieval augmented generation"

Configuration Files

- macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
- Windows: %APPDATA%\Claude\claude_desktop_config.json
- Linux: ~/.config/Claude/claude_desktop_config.json

Available Tools

search_arxiv_papers

Parameters:
- query (string, required): Natural language search query
- max_results (number, optional): Maximum papers to return (1-50, default: 10)
- sort_by (string, optional): Sort by relevance, lastUpdatedDate, or submittedDate
- sort_order (string, optional): ascending or descending

get_arxiv_paper

Parameters:
- arxiv_id (string, required): The arXiv ID (e.g., "2212.13345")

Installation

1. Clone this repository
2. Install dependencies:

   npm install

   

3. Build the project:
```bash
npm run build