Openalex Research Mcp

OpenAlex MCP Server

A Model Context Protocol (MCP) server that provides access to OpenAlex, a comprehensive open catalog of scholarly papers, authors, institutions, and more. Designed to empower AI assistants to conduct literature reviews, analyze research trends, and map the scholarly landscape.

Quick Install

npx openalex-research-mcp setup

This auto-detects your Claude Desktop config, prompts for your email and optional API key, writes the config, and verifies connectivity — all in one step. Restart Claude Desktop when done.

Flags:

• --print — print the config JSON without writing anything
• --config-path — print the detected config file path and exit
• --email you@example.com --api-key YOUR_KEY — non-interactive / scripted mode

Features:

• ⚡️ In-memory caching with TTL for fast repeated requests
• 🔄 Retry logic with exponential backoff for resilient API calls
• ✅ Input validation with Zod schemas
• 🏥 Health check tool for monitoring
• 📊 31 specialized tools for research
• 🎓 Curated journal presets — UTD24, FT50, AJG/ABS tiers, top AI conferences, and more
• 🏛️ Institution group presets — Ivy League, Top US, INSEAD+London, and more

Use as a Claude Skill (lightweight alternative)

The same OpenAlex access is also packaged as a Claude Skill under [skill/](skill/) — a token-frugal alternative to the always-on MCP server. A skill loads into the agent's context only when invoked and shells out to a small zero-dependency CLI, so it costs nothing while idle. Use the MCP server in MCP clients (Claude Desktop, TypingMind); use the skill in coding agents (Claude Code, etc.) that already have shell access.

chmod +x skill/bin/openalex && export PATH="$PWD/skill/bin:$PATH"
export OPENALEX_EMAIL="you@example.com"     # optional: faster polite pool
openalex works "supply chain network" -n 5

See [skill/README.md](skill/README.md) for install and the MCP-vs-skill trade-off. (The industry is steadily shifting routine API access from always-on MCP servers toward on-demand skills — this repo ships both.)

Features

Access 240+ million scholarly works through 31 specialized tools:

Literature Search & Discovery

• search_works: Advanced search with Boolean operators, venue/journal filters, institution filters, citation thresholds, and sorting
• get_work: Get complete metadata for a specific work (all authors, full abstract, references)
• getrelatedworks: Find similar papers based on citations and topics
• searchbytopic: Explore literature in specific research domains
• autocomplete_search: Fast typeahead search for all entity types

Credibility-Gated Search (Journal & Conference Presets)

• listjournalpresets: List all available named journal/conference and institution group presets
• searchinjournal_list: Search within a named preset list (UTD24, FT50, AJG 4*/4/3, top AI conferences, etc.) with optional institution filtering
• searchworksin_venue: Search within a specific venue by name, ISSN, or OpenAlex ID
• gettopvenuesforfield: Discover top journals/conferences in a field ranked by h-index
• checkvenuequality: Inspect h-index, impact, and indexing status of any venue

Citation Analysis

• getworkcitations: Forward citation analysis (who cites this work)
• getworkreferences: Backward citation analysis (what this work cites)
• getcitationnetwork: Build complete citation networks for visualization
• gettopcited_works: Find the most influential papers in a field

Author & Institution Analysis

• search_authors: Find researchers with h-index, citation metrics, and affiliations
• searchauthorsby_expertise: Find leading experts in a topic ranked by h-index
• getauthorprofile: Full research profile: h-index, i10-index, top works, recent works
• getauthorworks: Analyze an author's publication history
• getauthorcollaborators: Map co-authorship networks
• search_institutions: Find leading academic institutions

High-Value Citation Finding

• findreviewarticles: Find review papers and meta-analyses (high-value context citations)
• findseminalpapers: Find foundational "must-cite" papers (high citation count, published 5+ years ago)
• findopenaccess_version: Find freely available versions of papers with PDF links
• batchresolvereferences: Validate up to 20 DOIs/IDs at once

Research Landscape & Trends

• analyzetopictrends: Track research evolution over time
• compareresearchareas: Compare activity across different fields
• gettrendingtopics: Discover emerging research areas
• analyzegeographicdistribution: Map global research activity

Entity Lookup

• get_entity: Get detailed information for any OpenAlex entity
• search_sources: Find journals, conferences, and publication venues (sorted by h-index)

---

Journal & Conference Presets

Presets let you restrict searches to credible, high-impact venues and institution groups without manually specifying ISSNs or names. Call list_journal_presets to see all available options at any time.

Available Journal/Conference Presets

| Key | Name | Description | |---|---|---| | utd24 | UT Dallas 24 | Official UTD journal list for business school rankings (34 journals) | | ft50 | FT50 Journals | Financial Times 50 journals for MBA/business school rankings | | abs4star | AJG/ABS 4\* | World elite journals — the most prestigious tier in the ABS Guide | | abs4 | AJG/ABS 4 | Top international journals — excellent quality | | abs3 | AJG/ABS 3 | Internationally recognised journals — solid quality | | ms_misq_ops | MS + IS + Operations | Management Science, M&SOM, MIS Quarterly, ISR, JMIS, OR, POM | | top_ai_conferences | Top AI Conferences | NeurIPS, ICML, ICLR, AAAI, CVPR, ICCV, ACL, EMNLP, KDD, IJCAI | | top_cs_conferences | Top CS Conferences | SOSP, OSDI, SIGCOMM, CHI, VLDB, SIGMOD, PLDI | | nature_science | Nature & Science Family | Nature, Science, and branded sub-journals |

Available Institution Group Presets

| Key | Name | Institutions | |---|---|---| | harvard_stanford_mit | Harvard / Stanford / MIT | Harvard, Stanford, MIT | | ivy_league | Ivy League | All 8 Ivy League universities | | top_us | Top US Research Universities | Harvard, Stanford, MIT, Berkeley, Caltech, Chicago, Princeton, Yale, Columbia, Penn | | top_us_business | Top US Business Schools | Harvard, Stanford, Wharton, Booth, Kellogg, Sloan, Columbia, Stern, Darden, Tuck | | insead_london | INSEAD + London Schools | INSEAD, LBS, Imperial, LSE, Oxford, Cambridge | | top_global_business | Top Global Business Schools | Best of top_us_business + INSEAD, LBS, Oxford, Cambridge | | top_china | Top Chinese Universities | Peking, Tsinghua, Fudan, SJTU, ZJU, CUHK, HKU |

Example Preset Queries

# AI papers in UTD24 journals
search_in_journal_list(query="artificial intelligence", journal_list="utd24")

# AI papers in Management Science + M&SOM
search_in_journal_list(query="artificial intelligence", journal_list="ms_misq_ops")

# AI papers in FT50 journals since 2020
search_in_journal_list(query="artificial intelligence", journal_list="ft50", from_year=2020)

# AI papers in top AI conferences
search_in_journal_list(query="artificial intelligence", journal_list="top_ai_conferences")

# AI papers in AJG 4* journals
search_in_journal_list(query="artificial intelligence", journal_list="abs4star")

# AI papers in UTD24 journals by Harvard/Stanford/MIT authors
search_in_journal_list(query="artificial intelligence", journal_list="utd24", institution_group="harvard_stanford_mit")

# AI papers by professors at INSEAD
search_works(query="artificial intelligence", author_institution="INSEAD")

# AI papers by anyone from Harvard, Stanford, or MIT
search_works(query="artificial intelligence", institution_group="harvard_stanford_mit")

Requesting Additional Presets

> 📬 Want a new journal group added? > The preset lists (UTD24, FT50, AJG tiers, etc.) are curated in the source code. If your field uses a different ranking system — ABDC, VHB-JQ, CNRS, Norwegian list, discipline-specific lists, or any custom journal group — open a GitHub issue and I will add it. Include the list name, a short description, and the ISSNs or venue names. Community contributions via pull requests are also very welcome.

Installation

Option 1: Install from npm (Recommended)

# Install globally
npm install -g openalex-research-mcp

# Or use directly with npx (no installation needed)
npx openalex-research-mcp

Option 2: Install from source

# Clone the repository
git clone https://github.com/oksure/openalex-research-mcp.git
cd openalex-research-mcp

# Install dependencies
npm install

# Build the TypeScript code
npm run build

Configuration

Environment Variables (Optional but Recommended)

Set your email to join the "polite pool" for better rate limits:

export OPENALEX_EMAIL="your.email@example.com"

For premium users with an API key:

export OPENALEX_API_KEY="your-api-key"

Claude Desktop Configuration

Add to your Claude Desktop config file:

MacOS: ~/Library/Application Support/Claude/claude_desktop_config.json Windows: %LOCALAPPDATA%\Packages\Claude_\LocalCache\Roaming\Claude\claude_desktop_config.json

> Tip: In Claude Desktop, go to File → Settings → Developer → Edit Config to open the correct config file directly. The npx openalex-research-mcp setup command also auto-detects the path.

If you installed via npm/npx:

{
  "mcpServers": {
    "openalex": {
      "command": "npx",
      "args": ["-y", "openalex-research-mcp"],
      "env": {
        "OPENALEX_EMAIL": "your.email@example.com"
      }
    }
  }
}

If you installed from source:

{
  "mcpServers": {
    "openalex": {
      "command": "node",
      "args": ["/absolute/path/to/openalex-research-mcp/build/index.js"],
      "env": {
        "OPENALEX_EMAIL": "your.email@example.com"
      }
    }
  }
}

TypingMind and Other MCP Clients

The same configuration format works for TypingMind and other MCP-compatible clients.

> ⚠️ TypingMind Users: If you encounter "tooluseid" errors, see [TYPINGMIND.md](TYPINGMIND.md) for troubleshooting steps and best practices. TL;DR: Start a new chat, request fewer results (5-10), and use specific queries with filters.

Usage Examples

Example 1: Literature Review for AI Safety

Find the most influential papers on AI safety published since 2020

The assistant will use get_top_cited_works with appropriate filters to find highly-cited papers in AI safety research. The tool automatically filters for papers with at least 50 citations by default, ensuring results focus on influential work. For the most impactful papers, you can specify a higher threshold like min_citations: 200.

Example 2: Citation Network Analysis

Get the citation network for the paper "Attention Is All You Need" (DOI: 10.48550/arXiv.1706.03762)

The assistant will use get_citation_network to build a network of citing and referenced papers, enabling visualization of research impact.

Example 3: Research Trend Analysis

Show me how quantum computing research has evolved over the past 10 years

The assistant will use analyze_topic_trends to group publications by year and show growth patterns.

Example 4: Finding Collaborators

Who are the main collaborators of Geoffrey Hinton?

The assistant will use get_author_collaborators to analyze co-authorship patterns.

Example 5: Comparative Research Analysis

Compare research activity in "deep learning", "reinforcement learning", and "federated learning" from 2018-2024

The assistant will use compare_research_areas to show relative publication volumes.

Example 6: Geographic Research Mapping

Which countries are leading research in climate change mitigation?

The assistant will use analyze_geographic_distribution to map research activity by country.

Example 7: Top-Journal Citation Search

Find influential papers on "large language models" published in UTD24 journals since 2020

The assistant will use search_in_journal_list with journal_list="utd24" and from_year=2020.

Example 8: Institution-Filtered Search

Find papers on supply chain resilience published by researchers at Harvard, Stanford, or MIT

The assistant will use search_works with institution_group="harvard_stanford_mit".

Example 9: Seminal Paper Discovery

What are the must-cite foundational papers in transformer models?

The assistant will use find_seminal_papers with min_citations=500 to find highly-cited, older foundational works.

Example 10: Expert Discovery

Who are the top researchers in reinforcement learning, and where are they based?

The assistant will use search_authors_by_expertise with topic="reinforcement learning", sorted by h-index.

Response Format

The MCP server uses a two-tier response system to balance performance and completeness:

Summarized Responses (Search Results)

For list operations (search_works, get_citations, get_author_works, etc.), responses include only essential information:

Included:

• Core identifiers (ID, DOI, title)
• Publication metadata (year, date, type)
• Citation metrics (citedbycount)
• First 5 authors (with authors_truncated flag if more exist)
• Primary topic classification
• Open access status and URLs
• Source/journal name
• Abstract preview (first 500 chars)

Excluded to reduce size:

• Full author lists beyond 5 authors
• All secondary topics/concepts
• Complete affiliation details
• Full reference lists
• Detailed bibliographic data

This optimization reduces response sizes by ~80-90% (from ~10 KB to ~1.7 KB per work), making the server compatible with all MCP clients including TypingMind and Claude Desktop.

Full Details (get_work tool)

When you need complete information about a specific paper, use the get_work tool with a work ID or DOI. This returns:

Complete Author Information:

• ALL authors (not just first 5)
• Position indicators (first, middle, last author)
• Institutions and affiliations
• ORCID IDs
• Corresponding author flags
• Country information

Complete Content:

• Full abstract (reconstructed from OpenAlex index)
• All topics (not just primary)
• Complete bibliographic data
• Funding and grant information
• Keywords
• Complete reference and citation lists

Use Cases:

• Identifying PIs (often last author in biomedical fields)
• Finding corresponding authors
• Getting complete author affiliations
• Accessing full abstracts
• Comprehensive paper analysis

Tool Reference

Search Parameters

Most search tools support these common parameters:

• fromyear / toyear: Filter by publication year range
• min_citations: Minimum citation count (e.g., 50 for solid papers, 200 for highly influential)
• citedbycount: Citation filter with operator (e.g., ">100") — prefer min_citations for simplicity
• sourcename / sourceissn / source_id: Filter by journal or conference
• author_institution: Filter by author institution name (pipe-separated for OR, e.g., "Harvard University|MIT")
• institution_group: Named institution group preset (e.g., harvard_stanford_mit)
• is_oa: Filter for open access works only
• sort: Sort results (relevance_score, cited_by_count:desc, publication_year:desc)
• page / per_page: Pagination (max 200 per page; default 10, use 20 for broader coverage)

Boolean Search

The search_works and related tools support Boolean operators:

"machine learning" AND (ethics OR fairness)
"climate change" NOT "climate denial"
(AI OR "artificial intelligence") AND safety

…

Source & license

This open-source MCP server is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: oksure
• Source: oksure/openalex-research-mcp
• License: MIT

Install and usage instructions live in the source repository linked above.