Papersearch Mcp

papersearch-mcp 📄🔍

An Model Context Protocol (MCP) server for searching, analyzing, and extracting text from academic research papers. Built for Claude Code and Claude Desktop, it integrates arXiv and Semantic Scholar directly into your AI assistant's toolkit.

[](https://opensource.org/licenses/MIT) [](https://www.python.org/downloads/) [](https://modelcontextprotocol.io)

---

Architecture

sequenceDiagram
    participant Claude
    participant FastMCP
    participant ArXiv
    participant SemanticScholar
    Claude->>FastMCP: Call search_arxiv_papers
    FastMCP->>ArXiv: Query API
    ArXiv-->>FastMCP: Returns XML Metadata
    FastMCP-->>Claude: Returns JSON
    Claude->>FastMCP: Call extract_pdf_content
    FastMCP->>ArXiv: Download PDF
    FastMCP->>FastMCP: PyMuPDF Extract Text
    FastMCP-->>Claude: Returns Extracted Text

Features

• arXiv Integration:
• search_arxiv_papers: Query arXiv with support for filters like primary category (e.g., cs.CL for NLP, cs.LG for Machine Learning) and sort order.
• get_arxiv_paper_details: Retrieve full paper metadata, primary and secondary categories, author list, abstract, published/updated dates, and PDF/DOI URLs.
• Semantic Scholar Integration:
• search_semantic_scholar_papers: Broader academic graph search across multiple publishers. Includes citation and reference counts, open-access status, and journal/venue metadata.
• get_citation_graph: Query citing or cited papers for any paper ID (supporting Semantic Scholar ID, DOI, or arXiv ID).
• Document Analysis:
• extract_pdf_content: Downloads open-access paper PDFs and extracts the text content page-by-page using PyMuPDF (fitz). Truncation limits prevent overwhelming AI context windows.

---

Installation

Prerequisites

• Python >= 3.10
• pip (or uv / poetry)

Install via PyPI

pip install papersearch-mcp

Install from Source (Development)

git clone https://github.com/saitejabandaru-in/papersearch-mcp.git
cd papersearch-mcp
pip install -e .

---

Configuration

1. Claude Code

To add this server to your Claude Code CLI instance, run:

claude mcp add papersearch-mcp -- python -m papersearch_mcp.server

Or manually add it to your Claude Code settings file (usually located at ~/.claude/settings.json):

{
  "mcpServers": {
    "papersearch": {
      "command": "python",
      "args": ["-m", "papersearch_mcp.server"]
    }
  }
}

2. Claude Desktop

Add the following configuration to your Claude Desktop config (usually located at ~/Library/Application Support/Claude/claude_desktop_config.json on macOS):

{
  "mcpServers": {
    "papersearch": {
      "command": "python",
      "args": ["-m", "papersearch_mcp.server"]
    }
  }
}

---

Available Tools

1. search_arxiv_papers

Search the arXiv database for preprints and published papers.

• Arguments:
• query (string, required): Search query, e.g. "transformers attention" or "ti:\"Attention Is All You Need\"".
• max_results (integer, optional): Maximum papers to return (default 10, max 100).
• category (string, optional): Specific category filter, e.g. "cs.CL".
• sort_by (string, optional): Criteria to sort by: "relevance", "lastUpdatedDate", "submittedDate".
• sort_order (string, optional): "descending" (default) or "ascending".

2. get_arxiv_paper_details

Fetch full metadata for a specific arXiv ID.

• Arguments:
• arxiv_id (string, required): The unique paper ID, e.g., "2301.00001" or "cs/0601001".

3. search_semantic_scholar_papers

Search the wider Semantic Scholar database for peer-reviewed papers and citations.

• Arguments:
• query (string, required): Search query.
• max_results (integer, optional): Maximum papers to return (default 10).
• year (string, optional): Filter by year or range, e.g. "2023", "2020-2023", "2020-".
• fields_of_study (string, optional): Comma-separated fields, e.g. "Computer Science, Art".
• open_access_only (boolean, optional): Only return papers with open-access PDFs.

4. get_citation_graph

Retrieve the bibliography (references) or citations for a given academic paper.

• Arguments:
• paper_id (string, required): Paper identifier (supporting Semantic Scholar ID, "ARXIV:2301.00001", or "DOI:10.1234/example").
• direction (string, optional): "citations" (default) or "references".
• max_results (integer, optional): Limit results (default 20).

5. extract_pdf_content

Download a paper PDF and extract its text layout for study.

• Arguments:
• url (string, required): A direct link to the PDF file (e.g. "https://arxiv.org/pdf/2301.00001").
• max_pages (integer, optional): Maximum pages to extract (to manage context limits).

---

Development

Run tests using pytest:

pip install -e ".[dev]"
pytest

---

Contributing

Contributions are welcome! Please follow these steps:

1. Fork the repository.
2. Create a feature branch (git checkout -b feature/amazing-feature).
3. Run tests before committing.
4. Push to the branch (git push origin feature/amazing-feature).
5. Open a Pull Request.

License

This project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details.

Source & license

This open-source MCP server is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: saitejabandaru-in
• Source: saitejabandaru-in/papersearch-mcp
• License: MIT

Install and usage instructions live in the source repository linked above.