mcp-server-papers

Thank you to arXiv for use of its open access interoperability.

✅ What You CAN Do

• Retrieve and store descriptive metadata (titles, abstracts, authors, identifiers, classifications)

• Search and discover papers through API

• Retrieve paper content for research purposes

📋 Rate Limits

• Maximum 1 request every 3 seconds for legacy APIs (including the main ArXiv API)

• Single connection at a time

✨ Features

🔍 Search & Discovery

• Advanced Search: Query arXiv papers with field-specific searches (ti:, au:, abs:, etc.)
• Boolean Logic: Support for AND, OR, ANDNOT operators
• Flexible Sorting: Sort by relevance, submission date, or last updated
• ID-based Lookup: Direct paper retrieval by arXiv ID

📥 Paper Download

• PDF Download: Download papers directly by arXiv ID
• Flexible Paths: Save to custom locations or use auto-generated filenames
• Progress Tracking: File size reporting and download status
• Error Handling: Robust validation and error reporting

🌐 Online Paper Reading

• HTML Access: Fetch arXiv papers in HTML format for analysis
• Direct URL Support: Access papers via arXiv HTML interface
• Full Content: Get complete paper content including text and structure
• Research Integration: Perfect for AI agents and research tools

🖼️ Image Download & Analysis

• Universal Image Support: Download images from any URL (PNG, JPEG, GIF, WebP)
• AI Agent Compatible: Returns file paths for vision model analysis
• Smart Storage: Automatic file naming with conflict resolution
• Format Detection: Automatic image format detection from URL/headers
• Research Figures: Perfect for downloading and analyzing paper figures

🛡️ Validation Categories:

1. Query Parameter Validation

• Required parameters: Ensure either search_query or id_list is provided
• Parameter combinations: Validate logic between search_query and id_list
• Unknown parameters: Reject invalid parameter names

2. Field Prefix Validation

• Valid prefixes: ti, au, abs, co, jr, cat, rn, id, all
• Boolean operators: AND, OR, ANDNOT
• Syntax checking: Proper field:value format

3. Parameter Type Validation

• start: Non-negative integer
• max_results: Positive integer (1-2000 limit)
• sortBy: Must be "relevance", "lastUpdatedDate", or "submittedDate"
• sortOrder: Must be "ascending" or "descending"

4. ID List Validation

• Old format: math.GT/0309136v1 (subject-class/YYMMnnn)
• New format: 2301.00001v1 (YYMM.NNNN)
• Version numbers: Optional vN suffix
• Comma separation: Multiple IDs properly formatted

5. Date Range Validation (from your API spec)

• submittedDate format: [YYYYMMDDTTTT+TO+YYYYMMDDTTTT]
• Date logic: Start date before end date
• GMT time format: TTTT in 24-hour format

6. URL Encoding Validation

• Special characters: Proper encoding of spaces (+), quotes (%22), parentheses (%28, %29) in search_query
• Reserved characters: Ensure proper escaping in search_query

🚀 How to Use

Integrate with AI Agents

{
  "mcpServers": {
    "papers": {
      "command": "uv",
      "args": [
        "--directory",
        "/home/locch/Works/mcp-server-papers",
        "run",
        "mcp_server_papers"
      ]
    }
  }
}

Or:

{
  "mcpServers": {
    "papers": {
      "serverUrl": "http://localhost:8000/mcp"
    }
  }
}

📚 Reference

• arXiv API

• arXiv API Terms of Use

• arXiv API Basics

• arXiv API User Manual

• ArXiv MCP Server