Academic Paper Search

Stop writing four separate integrations for OpenAlex, Crossref, arXiv and PubMed just to do a literature search. One query here searches all four and hands back one normalized table.

Get clean, structured paper records from OpenAlex, Crossref, arXiv and PubMed as JSON, CSV or Excel, or call it as a tool from Claude, Cursor, ChatGPT or any MCP client. Pay only for the results you receive.

What you get

Give it a search query and it returns matching papers from up to four open sources at once: title, authors, abstract (where the source provides one), publication year, DOI, the paper's own URL, an open-access link when one exists, and a citation count where the source tracks it. Results interleave across sources as they come in, so you are not stuck waiting on one slow source before seeing anything from the others.

Every result is a flat record with stable field names, so it drops straight into a spreadsheet, a database or an AI agent's context.

Why use this instead of the website

• One query covers OpenAlex, Crossref, arXiv and PubMed instead of four separate searches and four response formats to parse
• Abstracts reconstructed from OpenAlex's inverted-index format and cleaned of HTML from Crossref, not left for you to decode
• Open-access links and citation counts pulled out into their own fields where the source has them
• A source that is briefly down or rate-limited is skipped, not fatal; the run still returns results from the rest
• Results in JSON, CSV, Excel or via API, or piped into Zapier, Make, n8n and Google Sheets
• Works as an MCP tool, so AI agents can pull papers for a lit review on demand
• No browser, no proxies, no personal data: fast runs and a tiny cost per result

Input

Field	Type	Default	Meaning
query	string		Search terms
sources	array	all four	Which sources to search: openalex, crossref, arxiv, pubmed
maxResults	integer	100	Cap on results saved. You are charged per result, so this caps your cost.

Example input:

{
  "query": "machine learning interpretability",
  "sources": ["openalex", "crossref", "arxiv", "pubmed"],
  "maxResults": 100
}

Output

Sample rows from a live run for "machine learning interpretability", showing all four sources:

source	title	year	venue	citationCount
openalex	Explainable AI: A Review of Machine Learning Interpretability Methods	2020	Entropy	2933
crossref	Bias, noise, and interpretability in machine learning	2020	Machine Learning	4
arxiv	Changing Data Sources in the Age of Machine Learning for Official Statistics	2023	arXiv	null
pubmed	Transforming Nanomaterials Development with Artificial Intelligence Techniques	2026	Nanotechnology, Science and Applications	null

Full example result (OpenAlex):

{
  "source": "openalex",
  "id": "https://openalex.org/W2896008109",
  "title": "Peeking Inside the Black-Box: A Survey on Explainable Artificial Intelligence (XAI)",
  "authors": ["Amina Adadi", "Mohammed Berrada"],
  "abstract": "At the dawn of the fourth industrial revolution, we are witnessing a fast and widespread adoption of Artificial Intelligence...",
  "year": 2018,
  "doi": "https://doi.org/10.1109/access.2018.2870052",
  "url": "https://openalex.org/W2896008109",
  "openAccessUrl": "https://ieeexplore.ieee.org/ielx7/6287639/8274985/08466590.pdf",
  "citationCount": 5049,
  "venue": "IEEE Access",
  "fetchedAt": "2026-09-24T18:12:03.441Z"
}

Field reference:

• source: which of the four sources this record came from (openalex, crossref, arxiv, pubmed)
• id: the source's own identifier for the paper (OpenAlex work ID, Crossref DOI, arXiv abstract URL, PubMed PMID)
• title, authors (list), abstract, year, doi, url, venue (journal, conference or "arXiv")
• openAccessUrl: a free full-text link where the source provides one, otherwise null
• citationCount: where the source tracks it (OpenAlex and Crossref), otherwise null
• fetchedAt: ISO timestamp of when this Actor fetched the record
• PubMed records have no abstract or doi: PubMed's summary endpoint does not return either reliably, so both are null for that source

Pricing

Pay per event. You are charged $1.50 per 1,000 results saved to the dataset, plus a fraction of a cent per run start. Nothing is charged for results you do not receive. Set "Max total charge per run" in the run options to cap spending on any run. When a run reaches your cap it stops cleanly and keeps everything it already saved.

Rough guide: 1,000 results cost $1.50 and take about 400 seconds, mostly PubMed's required pacing between requests.

Use it from an AI agent (MCP)

This Actor is available as an MCP tool through the Apify MCP server. Add it to your client, then ask the agent for the data in plain language.

Claude Desktop, Claude Code or Cursor (mcp.json / claude_desktop_config.json):

{
  "mcpServers": {
    "apify": {
      "url": "https://mcp.apify.com/?actors=brightpath-data/academic-paper-search",
      "headers": { "Authorization": "Bearer YOUR_APIFY_TOKEN" }
    }
  }
}

ChatGPT and other clients that support remote MCP servers: add https://mcp.apify.com/?actors=brightpath-data/academic-paper-search as a connector with your Apify token.

Example prompt once connected: "Find recent papers on machine learning interpretability with open-access PDFs and their citation counts."

Use it from code

curl -X POST "https://api.apify.com/v2/acts/brightpath-data~academic-paper-search/run-sync-get-dataset-items?token=YOUR_APIFY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"query":"machine learning interpretability","sources":["openalex","crossref","arxiv","pubmed"],"maxResults":100}'

Python:

from apify_client import ApifyClient
client = ApifyClient("YOUR_APIFY_TOKEN")
run = client.actor("brightpath-data/academic-paper-search").call(run_input={"query": "machine learning interpretability", "sources": ["openalex", "crossref", "arxiv", "pubmed"], "maxResults": 100})
items = client.dataset(run["defaultDatasetId"]).list_items().items

Limits and fair use

• Up to 10,000 results per run.
• PubMed is paced more slowly than the other three sources (NCBI rate-limits hard without an API key), so runs that include it take noticeably longer per 1,000 results; drop pubmed from sources for a faster run if you do not need it.
• OpenAlex, Crossref and arXiv are open, rate-limit-friendly APIs and are queried at a normal, polite pace.
• PubMed's summary endpoint does not reliably return an abstract or DOI, so those fields are null for PubMed results.
• A source that errors or is unreachable during a run is skipped for the rest of that run rather than failing it; the other sources still produce results.

Data source and legal

Data comes from four public, no-login metadata services: OpenAlex (CC0, an open catalog of scholarly works), Crossref (the DOI registration agency's public metadata API), arXiv (the open preprint repository's public API), and PubMed via NCBI's E-utilities (public biomedical literature metadata). This Actor collects public, non-personal data only (paper and author metadata, no reviewer or personal contact information) and does not bypass logins, paywalls or access controls. You are responsible for how you use the data.

Support

Found a problem or need a field added? Open an issue on the Actor's Issues tab. Fixes for broken runs are prioritized.