Crossref MCP Server — 150M+ Scholarly Works

Crossref MCP Server — 150M+ Scholarly Works

MCP (Model Context Protocol) server providing AI agents with direct access to the Crossref scholarly metadata database — over 150 million works with DOIs from 20,000+ publishers.

Connect any MCP-compatible client (Claude Desktop, Cursor, Cline, etc.) to search articles, look up DOIs, find journals, and discover funders.

Tools

Connect via MCP

Claude Desktop / Cursor

Add to your MCP settings:

{
  "mcpServers": {
    "crossref": {
      "url": "https://aligned_kettledrum--crossref-mcp-server.apify.actor/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_APIFY_TOKEN"
      }
    }
  }
}

Using mcp-remote (npx)

npx mcp-remote https://aligned_kettledrum--crossref-mcp-server.apify.actor/mcp \
  --header "Authorization: Bearer YOUR_APIFY_TOKEN"

Example Queries

Once connected, ask your AI assistant:

• "Find recent papers about transformer architectures in NLP"
• "Look up DOI 10.1038/nature12373 and summarize the paper"
• "What are the top journals for machine learning research?"
• "Find NIH funding information"

Tool Details

search_works

Search scholarly works with filters.

Parameters:

• query (required): Search keywords
• max_results: 1-100 (default: 20)
• sort: relevance, published, deposited, indexed, is-referenced-by-count
• work_type: journal-article, book-chapter, proceedings-article, dataset
• from_date: Filter by publication date (YYYY-MM-DD or YYYY)
• until_date: Filter by publication date (YYYY-MM-DD or YYYY)

Returns: DOI, title, authors, abstract, citation count, journal, year, URL

lookup_doi

Look up a single work by DOI.

Parameters:

• doi (required): The DOI (e.g., "10.1038/nature12373")

Returns: Full metadata including references (up to 50)

search_journals

Search journals by name.

Parameters:

• query (required): Journal name
• max_results: 1-50 (default: 10)

Returns: Title, publisher, ISSN, total DOIs, subjects

search_funders

Search research funders by name.

Parameters:

• query (required): Funder name
• max_results: 1-50 (default: 10)

Returns: Name, location, alternative names, URI

Data Source

All data comes from the Crossref REST API. Crossref is the official DOI registration agency for scholarly content, maintained by the not-for-profit Publishers International Linking Association (PILA).

FAQ

Q: Do I need a Crossref API key?A: No. The Crossref REST API is free and open. This MCP server handles rate limiting and authentication automatically.

Q: How current is the data?A: Crossref metadata is updated in near real-time as publishers register new DOIs. Most new papers appear within hours of publication.

Q: What types of works can I search?A: Journal articles, book chapters, conference proceedings, datasets, preprints, reports, standards, and more — any content type with a DOI registered through Crossref.

Q: Is there a rate limit?A: The MCP server handles Crossref's rate limiting internally (polite pool at ~10 requests/second). You don't need to worry about rate limits from the client side.

Q: Can I use this with any MCP client?A: Yes — any client supporting the MCP Streamable HTTP transport can connect. This includes Claude Desktop, Cursor, Cline, and other MCP-compatible tools.

Batch Processing

For large-scale data extraction (thousands of works), use the Crossref Scholarly Metadata Actor which supports bulk downloads with deep pagination.