Agent Test - MCP Research Server

MCP server for searching and managing arXiv research papers with intelligent paper information extraction and local storage.

📋 Prerequisites

1. Install Node.js 18 with NVM

Linux/macOS:

# Install NVM
# Refer to https://github.com/nvm-sh/nvm for latest version
curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.40.3/install.sh | bash

# Reload shell configuration
source ~/.bashrc  # or ~/.zshrc

# Install and use Node.js 18
nvm install 18
nvm use 18

# Verify installation
node --version  
npm --version   

Windows:

1. Download and install nvm-windows
2. Open new command prompt and run:

nvm install 18
nvm use 18
node --version
npm --version

2. Install UV Package Manager

# Install UV via pip
pip install uv

# Verify installation
uv --version

3. Configure UV Mirror Source (Optional)

For faster package downloads in China:

# Add packages with Tsinghua mirror
uv add --index https://mirrors.tuna.tsinghua.edu.cn/pypi/web/simple <package_name>

# Or set as default index in pyproject.toml (already configured in this project):
# [[tool.uv.index]]
# url = "https://pypi.tuna.tsinghua.edu.cn/simple"

🚀 Quick Start

# Clone the repository
git clone https://github.com/jieli-matrix/abacus-mcp.git
cd abacus-mcp

# For network issues (China/slow connections), set mirror first:
export UV_PYTHON_DOWNLOADS_URL="https://mirrors.tuna.tsinghua.edu.cn/python-release-for-build/"

# Create virtual environment with Python 3.11
uv venv --python 3.11

# Activate virtual environment
source .venv/bin/activate  # Linux/macOS
# .venv\Scripts\activate   # Windows

# Install dependencies
uv sync

# Start the server (opens port 50001)
uv run research_server.py

# In another terminal, start Inspector
npx @modelcontextprotocol/inspector node build/index.js

# Access Inspector at http://127.0.0.1:6274 and configure:
# - Transport Type: streamable-http
# - URL: http://127.0.0.1:50001/mcp

🛠️ Features

• 📚 Paper Search: Search arXiv papers by topic with relevance ranking
• 🔍 Information Extraction: Extract detailed paper metadata (title, authors, summary, etc.)
• 💾 Local Storage: Store paper information in organized JSON files
• 🤖 MCP Integration: Compatible with MCP-enabled AI assistants
• ⚡ Fast Performance: Built with UV package manager for speed

📦 Available Tools

search_papers

Search for papers on arXiv based on a topic and store their information locally.

Parameters:

• topic (str): The research topic to search for
• max_results (int, optional): Maximum number of results to retrieve (default: 5)

Returns:

• List of paper IDs found in the search

Example:

{
  "topic": "machine learning",
  "max_results": 3
}

extract_info

Retrieve stored information about a specific paper from local storage.

Parameters:

• paper_id (str): The arXiv paper ID to look for

Returns:

• JSON string with paper information if found, error message if not found

Example:

{
  "paper_id": "2301.12345"
}

🏗️ Project Structure

agent-test/
├── .venv/                 # Virtual environment 
├── research_server.py     # Main MCP server implementation
├── main.py               # Entry point
├── pyproject.toml        # UV project configuration
├── uv.lock              # Dependency lock file
├── papers/              # Generated paper storage directory
│   └── topic_name/
│       └── papers_info.json
├── README.md            
└── .gitignore          # Git ignore rules

🔧 Development

Prerequisites

• Python 3.11+
• Node.js 18+ (for MCP Inspector)
• UV package manager

Setup Development Environment

# Install UV (if not already installed)
pip install uv

# Install development dependencies
uv add mcp arxiv

🌐 Deployment Options

Option 1: Local Development

For local development:

# Start the server locally (opens port 50001)
uv run research_server.py

# In another terminal, start Inspector
npx @modelcontextprotocol/inspector node build/index.js

Access Inspector at http://127.0.0.1:6274
# Configure in Inspector UI:
- Transport Type: streamable-http
- URL: http://127.0.0.1:50001/mcp

Option 2: Remote Server Deployment

For remote server deployment using streamable-http transport:

Step 1: Start Server on Remote

# On remote server (opens port 50001)
uv run research_server.py

Step 2: Start Inspector Locally

# On local machine
npx @modelcontextprotocol/inspector node build/index.js

Step 3: Configure Inspector

Access http://127.0.0.1:6274 in your local browser and configure:

• Transport Type: streamable-http
• URL: http://YOUR_SERVER_IP:50001/mcp

📊 Usage Examples

Search for AI Papers

# Start the server
uv run research_server.py

# In another terminal, start Inspector
npx @modelcontextprotocol/inspector node build/index.js

# In Inspector UI (http://127.0.0.1:6274), configure:
# - Transport Type: streamable-http
# - URL: http://127.0.0.1:50001/mcp

# Then call search_papers:
{
  "topic": "artificial intelligence",
  "max_results": 5
}

Extract Paper Information

# After searching, extract specific paper info:
{
  "paper_id": "2301.07041"
}

🤝 Contributing

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add amazing feature')
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 file for details.

🔗 Resources

• MCP Documentation
• arXiv API Documentation
• FastMCP
• UV Package Manager

🐛 Troubleshooting

Common Issues

Issue	Solution
ModuleNotFoundError: mcp	Run uv sync to install dependencies
Inspector connection failed	Check Node.js version (node --version)