scholar_mcp_server

This MCP server provides a Go-based aggregation service for scholarly literature. It queries multiple major academic databases in parallel, merges results, removes duplicates intelligently (based on DOI and title), and presents a unified metadata format with relevance-aware ranking. The server exposes MCP-compatible tools that let you perform a multi-source search and fetch individual paper details through a single API surface. You can leverage the built-in tools to search across sources such as arXiv, Semantic Scholar, Crossref, Scopus, ADSABS, and Sci-Hub, with optional API keys for sources that require authentication.

Key tools available:

• searchScholarPapers: Runs a multi-source search, merges results, removes duplicates, and sorts by relevance, citation count, or published date. Supports advanced filters like author, title, journal, year range, min_citations, open_access_only, and source selection via enabled_sources. Example inputs show both basic and advanced queries, including enabling specific sources.
• getScholarPaper: Retrieves full paper details by a given identifier (DOI or arXiv ID). The system selects the best data source automatically to fetch complete metadata.

To use these tools, send MCP-formatted requests to the server’s /mcp endpoint, specifying the tool name (e.g., searchScholarPapers or getScholarPaper) and the required arguments. The server returns standardized paper metadata suitable for downstream processing or presentation in client applications like Claude Desktop, which can read the MCP configuration and route to the server.

Prerequisites:

• Go 1.24+ installed on your system
• Network access to the external scholarly databases (and any required API keys for restricted sources)

Installation steps:

1. Clone the repository

git clone https://github.com/Seelly/scholar_mcp_server.git
cd scholar_mcp_server

2. Resolve dependencies

go mod tidy

3. Run the server (development mode, using free data sources)

go run main.go logging.go

4. (Optional) Configure API keys for full functionality

# Example for sources requiring keys
export SCOPUS_API_KEY="your_scopus_api_key"
export ADSABS_API_KEY="your_adsabs_api_key"

# Run the server again
go run main.go logging.go

5. Build a production binary (optional)

# Build the binary
go build -o scholar-server main.go logging.go

# Run in background (example)
ohup ./scholar-server > server.log 2>&1 &

The server will typically start and listen on http://127.0.0.1:8080/ unless configured otherwise.

Notes and tips:

• The MCP interface mirrors the standard tools: searchScholarPapers for multi-source search and getScholarPaper for detailed item fetch. Use enabled_sources to limit data sources when needed.
• For Scopus and ADSABS, ensure you provide the necessary API keys; without them, those sources may be unavailable.
• The default port is 8080; if you need to operate behind a reverse proxy or in a container, adjust the binding as required and ensure /mcp is reachable.
• When running in production, consider enabling logging and monitoring to observe source performance and latencies, since the service performs asynchronous parallel calls to multiple data sources.