
arXiv Search MCP Server
Provides search functionality for arXiv.org papers through the official arXiv API, allowing users to search papers by keywords, filter by subject categories and date ranges, and receive comprehensive metadata including PDF links.
README
arXiv Search MCP Server
An MCP (Model Context Protocol) server that provides search functionality for arXiv.org papers using the official arXiv API.
Features
- Search by terms: Search papers by keywords in titles, abstracts, and content
- Subject filtering: Filter results by arXiv subject categories (physics, cs, math, etc.)
- Date range filtering: Filter papers by submission date range
- Configurable results: Control the number of results returned (max 2000)
- PDF URLs included: Direct links to paper PDFs for easy access
- Comprehensive metadata: Author names, abstracts, categories, publication dates, and more
Installation
From PyPI (Recommended)
pip install arxiv-search-mcp-server
From Source
- Clone the repository:
git clone https://github.com/gavinHuang/arxiv-search-mcp.git
cd arxiv-search-mcp
- Install the package:
pip install -e .
Usage
Running the Server
As a CLI command (after pip install)
arxiv-search-mcp
As a Python module
python -m arxiv_search_mcp.server
From source
python src/arxiv_search_mcp/server.py
Available Tools
1. search_arxiv_papers
Search for papers on arXiv with various filtering options.
Parameters:
terms
(required): Search terms to look for in paper titles, abstracts, and contentsubject
(optional): Subject category to filter by (see available categories below)start_date
(optional): Start date for filtering papers in YYYY-MM-DD formatend_date
(optional): End date for filtering papers in YYYY-MM-DD formatmax_results
(optional): Maximum number of results to return (default: 10, max: 2000)
Example:
search_arxiv_papers(
terms="quantum computing",
subject="quant-ph",
start_date="2023-01-01",
end_date="2023-12-31",
max_results=20
)
2. get_subject_categories
Get a list of available subject categories for filtering.
Returns: Dictionary of category codes and their descriptions.
Available Subject Categories
physics
- Physics (general)astro-ph
- Astrophysicscond-mat
- Condensed Matter Physicsgr-qc
- General Relativity and Quantum Cosmologyhep-ex
- High Energy Physics - Experimenthep-lat
- High Energy Physics - Latticehep-ph
- High Energy Physics - Phenomenologyhep-th
- High Energy Physics - Theorymath-ph
- Mathematical Physicsnlin
- Nonlinear Sciencesnucl-ex
- Nuclear Experimentnucl-th
- Nuclear Theoryquant-ph
- Quantum Physicsmath
- Mathematicscs
- Computer Scienceecon
- Economicseess
- Electrical Engineering and Systems Sciencestat
- Statisticsq-bio
- Quantitative Biologyq-fin
- Quantitative Finance
Response Format
Search results include the following information for each paper:
{
"arxiv_id": "2301.12345",
"title": "Paper Title",
"authors": ["Author 1", "Author 2"],
"abstract": "Paper abstract...",
"categories": ["cs.LG", "cs.AI"],
"published_date": "2023-01-15",
"pdf_url": "http://arxiv.org/pdf/2301.12345v1",
"arxiv_url": "http://arxiv.org/abs/2301.12345",
"comment": "Author comments",
"journal_ref": "Journal reference if available",
"doi": "DOI if available"
}
API Limits
- Maximum 2000 results per request (arXiv API limitation)
- Rate limiting: Please be respectful with API calls (arXiv recommends 3-second delays between requests for bulk operations)
- Results are returned by relevance by default
Error Handling
The server handles various error conditions:
- Invalid search parameters
- Network connectivity issues
- arXiv API errors
- Malformed responses
Errors are returned in a structured format with descriptive messages.
Dependencies
fastmcp
- MCP server frameworkrequests
- HTTP client for arXiv API callsfeedparser
- Atom/RSS feed parsingurllib3
- URL handling utilities
License
This project follows the arXiv API Terms of Use. Please respect arXiv's resources and use the API responsibly.
推荐服务器

Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。

VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。

e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。