dora_mcp

MCP (Model Context Protocol) server for accessing DORA (Digital Object Repository for Academia) publications from the lib4ri Empa repository.

Overview

This MCP server provides access to scientific publications in the DORA database at https://www.dora.lib4ri.ch/empa. It allows you to search for publications by author name, title, abstract, and other metadata fields.

Features

• Search DORA publications database
• Query across multiple fields (title, abstract, authors, contributors)
• Returns JSON-formatted publication data
• Runnable in Docker container

Installation

Prerequisites

• Python 3.10+
• uv package manager (recommended, somehow pip causes errors)

Install uv:

# On macOS/Linux
curl -LsSf https://astral.sh/uv/install.sh | sh

# On Windows
powershell -c "irm https://astral.sh/uv/install.ps1 | iex"

Local Installation

# Install dependencies using uv
uv sync

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

Docker Installation

Build and start with Docker Compose:

# Build and start
./docker.sh build
./docker.sh start

# Or manually
docker-compose up -d

Usage

Running Locally (stdio mode)

For use with Claude Desktop or other MCP clients:

# Using uv
uv run python -m dora_mcp

# Or with activated venv
python -m dora_mcp

Running with Docker (HTTP mode)

The Docker deployment runs in HTTP/SSE mode for web access:

# Start the server
./docker.sh start

# Access at http://localhost:8000
# SSE endpoint: http://localhost:8000/sse
# Messages endpoint: http://localhost:8000/messages

# View logs
./docker.sh logs

# Run tests
./docker.sh test

# Stop the server
./docker.sh stop

See DOCKER.md for detailed Docker deployment guide.

MCP Tools

The server provides the following tool:

search_publications

Search the DORA database for scientific publications.

Parameters:

• search_string (required): The search term to query. Can be an author name, title keyword, or any search term.

Example:

{
  "search_string": "manfred heuberger"
}

API Details

The server queries the DORA API with the following endpoint pattern:

https://www.dora.lib4ri.ch/empa/islandora/search/json_cit_a/[query]?search_string=[term]&extension=false

The search query includes weighted searches across multiple fields:

• Title (weight: 5)
• Abstract (weight: 2)
• Creator (weight: 2)
• Original Author List (weight: 2)
• Contributor (weight: 1)
• Type (weight: 1)
• Catch-all MODS fields (weight: 1)

Development

Requirements

• Python 3.10+
• uv package manager
• Dependencies listed in pyproject.toml

Testing

# Install dev dependencies
uv sync --all-extras

# Run all tests
uv run python -m pytest tests/ -v

# Run specific test suites
uv run python -m pytest tests/test_server.py -v          # Unit tests
uv run python -m pytest tests/test_with_metrics.py -v   # Metrics tests
uv run python -m pytest tests/test_docker.py -v         # Docker tests

# Run manual tests
uv run python examples/test_search.py

See TESTING.md for comprehensive testing documentation.

Transport Modes

The server supports two transport modes:

• stdio (default): For local MCP clients like Claude Desktop
• http: For web deployment with SSE (Server-Sent Events)

Set mode via environment variable:

# stdio mode (default)
uv run python -m dora_mcp

# HTTP mode
MCP_TRANSPORT=http MCP_PORT=8000 uv run python -m dora_mcp

License

See LICENSE file for details.