Skip to main content
Glama
arxiv-academic-search-mcp

arxiv-academic-search-mcp

arXiv & OpenAlex Academic Paper Search + Clean Web Extract MCP Server

MCP Compatible Zero Dependencies Free Tier Unlimited Pass License: MIT

Search peer-reviewed scientific papers, citations, abstracts & PDFs across arXiv and OpenAlex, plus extract clean prompt-injection-scanned Markdown from any URL.

Looking for a fast, zero-dependency arXiv MCP or OpenAlex academic research MCP server for Claude Desktop, Cursor, or Codex? arxiv-academic-search-mcp searches millions of preprints and peer-reviewed papers with citation counts, DOIs, and direct PDF links — paired with a clean Markdown web extractor that scans pages for hidden prompt-injection attacks.


Why Use arxiv-academic-search-mcp?

  • Unified arXiv + OpenAlex Literature Search (academic_search) — Query computer science, AI/ML, physics, quant finance, biomedicine, and economics papers sorted by citations or recency, complete with abstracts, authors, year, and open-access PDF links.

  • Clean LLM Markdown Web Extractor (web_extract) — Converts documentation, HTML papers, blog posts, and research pages into clean Markdown while flagging hidden prompt-injection payloads.

  • Zero Python / Zero API Key Setup — Runs immediately on Node >=18 (npx or remote Streamable-HTTP) with 2 free calls/day out of the box.

  • 2 Free Calls / Day Out of the Box — Works immediately with zero configuration and no API key required.

  • Instant Unlimited Upgrade — Need high-volume production calls? Grab an instant API key at https://agent-kiosk.vercel.app/buy ($1 Starter 7-Day Pass or $5 Pro 30-Day Pass) or pay per request natively via x402 (USDC on Base).


Related MCP server: Research Paper Ingestion MCP Server

Option 1: Zero-Install Remote MCP URL (Fastest)

Connect directly via Streamable-HTTP MCP without installing anything locally:

  • Free Tier (2 Free Calls/Day, Zero Config):

    https://agent-kiosk.vercel.app/mcp
  • Paid Unlimited URL (With Prepaid Pass Key):

    https://agent-kiosk.vercel.app/mcp?api_key=ak_live_YOUR_KEY_HERE

Get an Instant API Key in 10 Seconds: https://agent-kiosk.vercel.app/buy

  • $1.00 USDC — 7-Day Starter Pass (Instant ak_live_... key)

  • $5.00 USDC — 30-Day Pro Pass (Full unmasked datasets across all endpoints)


Option 2: Local npx / Stdio Setup (Claude Desktop, Cursor, Codex)

This package is pure Node.js >= 18 with zero external npm dependencies, exposing only the 2 focused tools for this workflow so your LLM context window stays lean.

Claude Desktop (claude_desktop_config.json)

{
  "mcpServers": {
    "arxiv-academic-search-mcp": {
      "command": "npx",
      "args": ["-y", "github:ralf9090/arxiv-academic-search-mcp"],
      "env": {
        "AGENT_KIOSK_API_KEY": ""
      }
    }
  }
}

(Leave AGENT_KIOSK_API_KEY empty for 2 free calls/day, or paste your ak_live_... key from agent-kiosk.vercel.app/buy for unlimited access.)

Cursor (.cursor/mcp.json)

{
  "mcpServers": {
    "arxiv-academic-search-mcp": {
      "command": "npx",
      "args": ["-y", "github:ralf9090/arxiv-academic-search-mcp"],
      "env": {
        "AGENT_KIOSK_API_KEY": "ak_live_OPTIONAL_PASS_KEY"
      }
    }
  }
}

OpenAI Codex CLI (~/.codex/config.toml)

[mcp_servers.arxiv_academic_search_mcp]
command = "npx"
args = ["-y", "github:ralf9090/arxiv-academic-search-mcp"]
env = { AGENT_KIOSK_API_KEY = "" }

Run Directly via git clone + node index.js

git clone https://github.com/ralf9090/arxiv-academic-search-mcp.git
cd arxiv-academic-search-mcp
node index.js

Included MCP Tools (2)

Deep scientific & academic research tool. Queries OpenAlex and arXiv for peer-reviewed papers and preprints, returning citation counts, reconstructed abstracts, authors, and direct Open-Access PDF links.

Parameter

Type

Required

Description

query

string

Yes

Research topic or scientific query (e.g., 'autonomous AI agents', 'zero knowledge proofs')

limit

number

No

Number of papers to return (1-15, default: 5)

x_api_key

string

No

Optional ak_live_... 30-day Prepaid API Pass key to unlock the full unmasked dataset

web_extract

Extract clean, LLM-ready Markdown from any web URL while automatically scanning the page for hidden Prompt Injection traps targeting AI agents.

Parameter

Type

Required

Description

url

string

Yes

Full URL to extract (must start with http:// or https://)

max_length

number

No

Maximum markdown character length to return (500-25000, default: 8000)

x_api_key

string

No

Optional ak_live_... 30-day Prepaid API Pass key to unlock the full unmasked dataset


Example Prompts

  • "Search arXiv and OpenAlex for 2025-2026 papers on test-time compute scaling in reasoning LLMs."

  • "Find the highest-cited papers on high-frequency market making and order book imbalance."

  • "Extract clean Markdown from https://arxiv.org/abs/2305.18290 and summarize the core objective function."


Pricing & Rate Limits

Tier

Cost

Quota

Setup

Free Trial

$0

2 Free Calls / Day

Zero config — works out of the box

Starter Pass

$1 USDC

7 Days Unlimited

Get ak_live_... at agent-kiosk.vercel.app/buy

Pro Pass

$5 USDC

30 Days Unlimited

Get ak_live_... at agent-kiosk.vercel.app/buy

x402 Pay-Per-Call

$0.008–$0.10

Unlimited autonomous

Pass payment_signature (Base USDC EIP-3009)


License

MIT © ralf9090 — Powered by Agent Kiosk (Upgrade / Buy API Key).

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      B
      quality
      D
      maintenance
      Enables searching and retrieving academic papers from arXiv by various criteria including title, author, and category, with support for extracting full text content from PDFs.
      4
      MIT
    • F
      license
      Not graded
      quality
      D
      maintenance
      Enables searching, downloading, and analyzing academic papers from arXiv and Semantic Scholar to extract key insights and citation metrics. It facilitates autonomous knowledge acquisition by processing research findings and integrating them into persistent AI memory systems.
      -
    • F
      license
      Not graded
      quality
      C
      maintenance
      Enables AI agents to search and retrieve academic papers from arXiv and PubMed with citation-grade extraction, providing clean citations and summaries.
      -