OpenAlex MCP Server

A Model Context Protocol (MCP) server that provides access to the OpenAlex API - a fully open catalog of the global research system covering over 240 million scholarly works.

Features

This MCP server provides tools to search and retrieve:

- Works - Scholarly articles, preprints, datasets, books (240M+ items)
- Authors - Researchers and creators with ORCID integration
- Sources - Journals, conferences, repositories (~250K venues)
- Institutions - Universities, hospitals, labs with ROR matching
- Concepts - Hierarchical research topics (levels 0-5)
- Publishers - Publishing organizations
- Funders - Grant-making bodies
- Autocomplete - Type-ahead search across all entity types
- Text Classification - Concept prediction for arbitrary text

Installation

From npm (Recommended)

npm install -g openalex-mcp

From Source

git clone https://github.com/reetp14/openalex-mcp.git
cd openalex-mcp
npm install
npm run build

Usage

As an MCP Server

Add to your MCP client configuration:

{
  "mcpServers": {
    "openalex": {
      "command": "npx",
      "args": ["openalex-mcp"]
    }
  }
}
json

Or if installed locally:

{
  "mcpServers": {
    "openalex": {
      "command": "node",
      "args": ["./node_modules/openalex-mcp/build/index.js"]
    }
  }
}

Available Tools

Entity Search Tools

All search tools support the full OpenAlex query grammar:

- search_works - Search scholarly works
- search_authors - Search researchers and creators
- search_sources - Search journals, conferences, repositories
- search_institutions - Search universities, hospitals, labs
- search_concepts - Search research topics
- search_publishers - Search publishing organizations
- search_funders - Search grant-making bodies

Common Parameters:

- search - Full-text search query
- filter - Boolean filters (e.g., concept.id:C12345,from_publication_date:2022-01-01)
- sort - Sort field with optional :desc (e.g., cited_by_count:desc)
- page/per_page - Standard pagination (max 10,000 results total)
- cursor - Deep pagination (use for first call)
- group_by - Faceting/aggregation by field
- select - Comma-separated fields to return
- sample - Random sample size with optional seed
- mailto - Your email for higher rate limits

{
  "tool": "search_works",
  "arguments": {
    "search": "artificial intelligence",
    "filter": "from_publication_date:2023-01-01,to_publication_date:2023-12-31",
    "sort": "cited_by_count:desc",
    "per_page": 10,
    "mailto": "[email protected]"
  }
}

{
  "tool": "search_authors",
  "arguments": {
    "filter": "last_known_institution.id:I27837315",
    "sort": "works_count:desc",
    "select": "id,display_name,works_count,cited_by_count"
  }
}

{
  "tool": "search_works",
  "arguments": {
    "filter": "concepts.id:C154945302",
    "group_by": "publication_year"
  }
}

{
  "tool": "autocomplete",
  "arguments": {
    "search": "nature",
    "type": "sources",
    "per_page": 5
  }
}

{
  "tool": "classify_text",
  "arguments": {
    "title": "Deep Learning for Medical Image Analysis",
    "abstract": "We present a novel approach using convolutional neural networks..."
  }
}

- Standard: page=2&per_page=100 (max 10,000 results)
- Deep: cursor= (first call), then use returned next_cursor

Rate Limits

- Anonymous: 10 requests/second, 100,000/day
- With mailto: 100 requests/second, 1,000,000/day

API Response Format

All tools return the standard OpenAlex JSON envelope:

{
  "meta": {
    "count": 249256387,
    "db_response_time_ms": 12,
    "page": 1,
    "per_page": 25,
    "next_cursor": "ZjEwMD..."
  },
  "results": [
    {
      / entity object /
    }
  ]
}

Development

```bash