What is Research Papers Mcp

The research_papers_mcp is an MCP server designed to facilitate interaction with research papers stored in a GitHub repository. It enables users to list, search, and read PDF papers using an MCP client like Claude Desktop.

Use cases

Use cases include researchers looking for specific papers, students needing to access academic resources, and developers wanting to integrate paper retrieval functionalities into their applications.

How to use

To use research_papers_mcp, clone the repository from GitHub, ensure you have Python 3.11 installed, and set up a GitHub Personal Access Token for accessing private repositories. Then, run the server and connect using an MCP client such as Claude Desktop.

Key features

Key features include listing available PDF papers, searching for papers by keywords, reading specific papers, and a prompt support for initiating searches.

Where to use

research_papers_mcp can be used in academic research, educational institutions, and any environment where access to a collection of research papers is needed.

Research Papers MCP Server

📚 Overview

The research_papers_mcp project is an MCP (Message Control Protocol) server that provides tools to interact with research papers stored in a GitHub repository (Santhosraj/research_papers). It allows users to list, search, and read PDF papers using an MCP client like Claude Desktop. The server is built using the FastMCP framework and runs locally with stdio transport.

✨ Features

• List Papers: Retrieve a list of PDF files in the papers/ folder of the repository.
• Search Papers: Search for papers by keyword in filenames or content.
• Read Papers: Extract and read the content of a specific PDF file.
• Prompt Support: Includes a prompt (search_papers_prompt) to initiate searches.

📁 Project Structure

research_papers_mcp/
├── src/
│ └── server.py # Main server script
├── tools/
│ ├── list_papers.py # Tool to list PDF files
│ ├── search_papers.py # Tool to search papers by keyword
│ └── read_paper.py # Tool to read a specific PDF
├── prompts/
│ └── search_prompts.py # Prompt definition for searching papers
├── config/
│ └── config.json # Claude Desktop configuration
├── requirements.txt # Python dependencies
├── README.md # Project documentation
├── LICENSE # License file (add your preferred license)
└── .gitignore # Git ignore file

✅ Prerequisites

• Python 3.11: Make sure Python 3.11 is installed.
• GitHub Personal Access Token (PAT): Required to access private repos. Generate one with repo scope.
• Claude Desktop: An MCP client to interact with this server. Install here.

⚙️ Setup Instructions

Setup Instructions

## 1. Clone the Repository

Clone the project to your local machine:
```bash
git clone https://github.com/Santhosraj/research_papers_mcp.git
cd research_papers_mcp

3. Set Up a Virtual Environment

Create and activate a virtual environment:

python -m venv .venv
.venv\Scripts\activate  # On Windows
# source .venv/bin/activate  # On Unix/Linux/Mac

3. Install Dependencies

Install the required Python packages:

uv pip install -r requirements.txt

The requirements.txt includes dependencies like mcp, PyGitHub, PyPDF2, uvicorn, and fastapi.

4. Configure Claude Desktop

Create a config.json file in the project root to configure Claude Desktop for local use:

{
  "mcpServers": {
    "research-papers": {
      "url": "stdio://local",
      "command": "research_papers_mcp\\.venv\\Scripts\\python.exe",
      "args": [
        "research_papers_mcp\\src\\server.py"
      ],
      "env": {
        "GITHUB_PERSONAL_ACCESS_TOKEN": "your_github_pat_here"
      }
    }
  }
}

Replace “your_github_pat_here” with your GitHub PAT.
Adjust the command and args paths if your project directory differs.

5. Run the Server

Start the MCP server locally:

python src/server.py

You should see:

bash Server initialized

6. Interact with the Server Using Claude Desktop

Launch Claude Desktop:claude-desktop --config research_papers_mcp\config.json

List Tools and Prompts:/list-tools

Expected Output:Available tools:

- list_papers
- search_papers
- read_paper

Available prompts:

- search_papers_prompt

Search for Papers:/search_papers {“keyword”: “cnn”}

Expected Output (example):{
  "matches": [
    {"file": "papers/some_paper_with_cnn.pdf", "match_type": "filename"},
    {"file": "papers/another_paper.pdf", "match_type": "content"}
  ]
}

Usage Examples

List All Papers:/list_papers

Search for Papers with Keyword “transformer”:/search_papers {“keyword”: “transformer”}

Read a Specific Paper:/read_paper {“paper_name”: “some_paper.pdf”}

Troubleshooting

Server Fails to Start:
1)Ensure all dependencies are installed: uv pip install -r requirements.txt.
2)Verify your GitHub PAT has the repo scope.

Tool Errors:

Check the server logs for GitHub API errors (e.g., rate limits, invalid PAT).
Ensure the papers folder exists in Santhosraj/research_papers.

Claude Desktop Not Connecting:

Confirm the server is running before starting Claude Desktop.
Verify the config.json paths and url are correct.

Contributing 🚀

Feel free to fork the repository, make changes, and submit pull requests. For major changes, please open an issue first to discuss your ideas.

License

This project is licensed under the MIT License. See the LICENSE file for details.

Contact

For questions or support, reach out to Santhosraj at [[email protected]].