Log inSubmit server
Scholar Feed MCP logo

Scholar Feed MCP

by YGao2005

Research over 600k+ CS/AI papers with semantic search, citation graphs, full text, author discovery, and library tools.

InstallRepositoryWebsite
Full README Compare alternatives Embed widget Claim this listing

Install Scholar Feed MCP

Pick your client, then add it in one click or paste the config.

STDIONode.js
~/Library/Application Support/Claude/claude_desktop_config.json
claude_desktop_config.json
{
  "mcpServers": {
    "ygao2005-scholar-feed-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "scholar-feed-mcp"
      ],
      "env": {
        "SF_API_KEY": "YOUR_VALUE_HERE"
      }
    }
  }
}

Settings → Developer → Edit Config, merge this into mcpServers, then fully quit and reopen Claude Desktop.

Values are written into the snippet above so you can paste it as is. They never leave your browser.

Install not yet sandbox-tested— the server itself is confirmed live (we called its tools/list endpoint). The stdio install command is a separate, ongoing check.

Overview

YGao2005/scholar-feed-mcp MCP server connects MCP clients to Scholar Feed's daily-indexed arXiv corpus of more than 600,000 CS, AI, and ML papers. It provides semantic and keyword search, paper metadata and BibTeX, citation traversal, section-level full-text extraction, author and co-author discovery, and research-library management. Anonymous access is available with a monthly call limit; setting SF_API_KEY enables an authenticated library and higher quota, while some features require Pro. Reach for it when an agent needs to find, compare, trace, annotate, or monitor research papers without leaving an MCP client.

Use cases

  • Search recent papers on a technical topic
  • Trace foundational work and citation relationships
  • Export BibTeX for a literature review
  • Monitor new papers with standing watches
  • Organize and annotate a research reading library

Key features

  • Semantic and filtered paper search
  • Citation graph and foundational lineage traversal
  • Section-level arXiv full-text extraction
  • Author and co-author discovery
  • BibTeX export
  • Authenticated libraries, collections, and watches

Tools27

What Scholar Feed MCP lets your agent do. Expand a tool for its parameters.

Verified live~12.4k tokens of context

Security & health

What it can reach and how reliably it runs — from our own checks, not the maintainer’s claims.

ActiveLast checked Today

Recent health check succeeded.

19/20 checks healthy over the last 80d

Supply-chain signal

No high-severity advisories surfaced by our automated scan.

Critical 0High 0Medium 0Low 0

Scanned 12d ago via OSV.dev · scholar-feed-mcp (npm)

Adoption & maintenance

Factual signals from GitHub, npm, and our automated checks — not a rating.

GitHub stars13Stargazers on the source repository.
npm downloads456Package downloads in the last 30 days.
Last commit9d agoMost recent push to the default branch.
Tools exposed27Callable tools this server registers over MCP.
Directory activity4 viewsConfig copies, upvotes, and views on AllMCPs.

How Scholar Feed MCP works

What YGao2005/scholar-feed-mcp does

YGao2005/scholar-feed-mcp MCP server exposes Scholar Feed's research index through MCP. The service indexes arXiv papers daily and covers more than 600,000 papers in CS, AI, and ML. Results can include generated summaries, novelty scores, citation counts, code links, venues, structured extraction, and library state for authenticated users.

The main workflow starts with search_papers, which supports semantic search by default as well as keyword and filtered discovery. get_paper retrieves one paper or batches up to 50 arXiv IDs, and can return selected fields, verbose structured data, full text, or a BibTeX entry for one paper. get_citations follows references or incoming citations, while fetch_fulltext reads selected sections from arXiv HTML, LaTeX, or PDF sources.

How it works

The server runs as a standard stdio MCP process. Agents call tools using arXiv IDs, search terms, author IDs, collections, or watch definitions. Search ranking can reflect relevance, citations, forecast impact, community adoption, or other tool-specific criteria. Citation and lineage tools help compensate for semantic retrieval's tendency to miss older canonical papers.

Author discovery has separate query and profile modes. find_author searches by topic or name, or retrieves a known author ID; co_author_graph expands that seed into a collaboration neighborhood. embed_text returns a 768-dimensional Gemini Flash embedding for custom retrieval workflows such as HyDE, but it is restricted to Pro accounts.

Setup and configuration

Install with npx scholar-feed-mcp@latest init. The interactive setup can ask for an API key, detect Claude Code, Cursor, or Claude Desktop, write the client configuration, and verify the connection. Manual configurations launch npx -y scholar-feed-mcp@latest as a stdio server.

Anonymous operation does not require credentials and provides 200 calls per month. Set SF_API_KEY to a free Scholar Feed key for authenticated access and a 500-call monthly quota. Authentication is required for saved papers, collections, watches, annotations, and other library operations. The setup material names Claude Desktop, Claude Code, Cursor, Windsurf, Cline, Roo Code, Gemini CLI, LM Studio, JetBrains, OpenAI Codex, VS Code, Zed, and Continue as supported configuration targets.

Tools and capabilities

  • Search papers semantically or with filters for category, novelty, task, dataset, contribution type, date, and citation scope.
  • Retrieve compact or verbose paper records and export individual BibTeX citations.
  • Traverse citation relationships and retrieve foundational lineage for a paper's niche.
  • Read abstracts, introductions, related work, methods, results, conclusions, or complete extracted papers.
  • Find researchers, inspect profiles, and map co-authorship relationships.
  • Orient a literature search with candidate foundational papers.
  • Save, like, annotate, organize, and query papers in authenticated libraries.
  • Create, preview, update, check, list, and delete standing watches for new papers.
  • Find unsaved foundational or frontier gaps and check whether methods have been superseded in supported families.

Limitations and notes

YGao2005/scholar-feed-mcp MCP server does not guarantee that semantic search returns a field's oldest or most cited anchor. For canonical work, inspect repeated baseline references or use get_foundational_lineage. Citation count, forecast impact percentile, and GitHub stars represent different kinds of impact and should not be treated as interchangeable.

Anonymous responses omit authenticated fields such as is_saved; missing values should not be interpreted as false. Full-text extraction can fail when all supported arXiv sources are unavailable, and section labels may be uncertain. fetch_fulltext can consume substantial context, especially with sections='all', so targeted sections are preferable.

Free or anonymous callers cannot use embed_text; it returns pro_required without a Pro account. find_gaps also requires an authenticated Pro account, while ask_library has a lower free-account allowance. The check_drift endpoint grounds receipts in arXiv benchmark evidence, but its attribution can still require reviewing the quoted source paper.

Read the full README View source on GitHub

Reviews

No reviews yet — be the first to share how it worked for you.

Frequently Asked Questions about Scholar Feed MCP

YGao2005/scholar-feed-mcp is an MCP server that connects AI clients to Scholar Feed's arXiv research index of more than 600,000 CS, AI, and ML papers. Its main tools search papers, retrieve metadata and BibTeX, traverse citations, extract full text, find authors, and manage authenticated research libraries.

Run `npx scholar-feed-mcp@latest init`. The setup wizard can detect Claude Code, Cursor, or Claude Desktop, create the configuration, and test the connection. For manual setup, configure the client to launch `npx -y scholar-feed-mcp@latest` as a standard stdio MCP server.

No, an API key is optional for anonymous access. Anonymous use provides 200 calls per month; setting `SF_API_KEY` with a free Scholar Feed key provides authenticated features and a 500-call monthly quota. Library operations such as saving, collections, watches, and annotations require the key.

Yes, it works with Cursor. Add the stdio server to `.cursor/mcp.json` for a project or `~/.cursor/mcp.json` globally, then restart Cursor. The initialization command can also detect Cursor and write its configuration.

Yes, `get_paper` can return a BibTeX entry by passing `format='bibtex'`. BibTeX export is single-paper only, so fetch multiple citations individually when a batch is needed.

Related MCP servers

All in Research All alternatives
Add a badge to your READMEShow it’s listed — live SVG, Markdown or HTML.

AllMCPs Directory Badge

Full Badge Customizer

Showcase your server listing on GitHub or your project documentation. Embed this dynamic SVG badge to highlight official listing status and live engagement.

Badge Style:
Live Dynamic SVG PreviewScholar Feed MCP AllMCPs Directory Badge
Markdown (GitHub README)
[![AllMCPs](https://allmcps.com/api/badge/ygao2005-scholar-feed-mcp?style=directory)](https://allmcps.com/mcp/ygao2005-scholar-feed-mcp)
HTML Embed
<a href="https://allmcps.com/mcp/ygao2005-scholar-feed-mcp"><img src="https://allmcps.com/api/badge/ygao2005-scholar-feed-mcp?style=directory" alt="Scholar Feed MCP on AllMCPs" /></a>