semantic-scholar-mcp

MCP server for academic research powered by Semantic Scholar — access 200M+ papers, citations, authors, and recommendations.

• 📄 Paper lookup — find papers by arXiv ID, DOI, title, or Semantic Scholar ID
• 🔗 Citation tracing — explore reference chains and citation networks with context and intent
• 👤 Author profiles — get h-index, publication counts, affiliations, and paper lists
• 🎯 Smart recommendations — discover related papers using positive/negative seed papers
• 🔍 Advanced search — filter by year, field of study, citation count, and open access status
• 📦 Batch operations — fetch up to 500 papers in a single call

What is this?

This is a Model Context Protocol (MCP) server that connects AI assistants to the Semantic Scholar Academic Graph API. It exposes 8 tools that cover the full spectrum of academic literature operations — from single paper lookups to batch fetching, citation graph traversal, author profiling, and paper recommendations.

Designed for AI assistants like Claude to help with literature review, citation analysis, and research discovery. Works with any MCP-compatible client (Claude Desktop, Claude Code, or custom integrations).

Installation

npx @yogsoth-ai/semantic-scholar-mcp

No installation needed. The npx command downloads and runs the latest version automatically.

Or install globally:

npm install -g @yogsoth-ai/semantic-scholar-mcp

Quick Start

1. Add to Your MCP Client

Edit your MCP configuration file:

Claude Desktop — ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows)

Claude Code — .mcp.json in your project root

{
  "mcpServers": {
    "semantic-scholar": {
      "command": "npx",
      "args": ["@yogsoth-ai/semantic-scholar-mcp"],
      "env": {
        "SS_API_KEY": "optional-but-recommended"
      }
    }
  }
}

2. Get an API Key (Optional)

Without a key: 1 request/second
With a key: 100 requests/second

Get your free API key at Semantic Scholar API.

3. Restart Your MCP Client

The server will start automatically when the client needs it.

Available Tools

Tool	Description
paper	Get full metadata for a single paper (supports ARXIV, DOI, S2 ID, PMID, URL formats)
paperBatch	Batch fetch up to 500 papers in one call
references	Get papers cited by a paper (outgoing citations with context and intent)
citations	Get papers citing a paper (incoming citations with context and intent)
recommendations	Get paper recommendations based on positive/negative seed papers
relevanceSearch	Search papers by keyword with filters (year, field, citations, open access)
author	Get author profile (name, affiliations, h-index, paper/citation counts)
authorPapers	Get all papers by an author with full metadata and pagination

Supported ID Formats

The paper and paperBatch tools accept multiple identifier formats:

Format	Example	Notes
arXiv (prefixed)	ARXIV:1706.03762	Explicit arXiv prefix
arXiv (bare)	2005.14165	Auto-prefixed to ARXIV:
DOI (prefixed)	DOI:10.1038/s41586-021-03819-2	Explicit DOI prefix
DOI (bare)	10.1038/s41586-021-03819-2	Auto-prefixed to DOI:
S2 ID	204e3073870fae3d05bcbc2f6a8e263d9b72e776	Semantic Scholar hex ID
PubMed	PMID:34265844	PubMed identifier
URL	URL:https://arxiv.org/abs/1706.03762	Paper URL

Example Queries

Ask Claude things like:

• "Find the Attention Is All You Need paper and show me its top 5 most cited references"
• "Who are the authors of the GPT-3 paper and what's their h-index?"
• "Search for recent papers on diffusion models published after 2023 with at least 100 citations"
• "Recommend papers similar to BERT but not related to computer vision"
• "Show me all papers by Yann LeCun from the last 3 years"
• "Batch fetch these 10 papers and compare their citation counts"
• "Trace the citation graph from LoRA — what adaptations have been proposed?"

Search Filters

The relevanceSearch tool supports these filters:

Filter	Example	Description
year	"2023-2024", "2020-", "-2023"	Publication year range
fields_of_study	"Computer Science"	Academic field filter
min_citation_count	100	Minimum citation threshold
open_access_only	true	Only papers with free PDF

For Developers

Local Development

git clone https://github.com/yogsoth-ai/semantic-scholar-mcp.git
cd semantic-scholar-mcp
npm install
npm run build
npm run mcp  # Start the server locally

Running Tests

npm test

Use in Other MCP Clients

Any MCP-compatible client can use this server. The configuration is the same JSON block shown in Quick Start.

Links

• 📦 npm package
• 🐙 GitHub repository
• 📚 Semantic Scholar API Docs
• 🔧 Model Context Protocol

📄 License

Apache-2.0