About semanticscholar
A Semantic Scholar MCP server for researchers and literature-review workflows. It covers paper and author records, citation networks, recommendations, and searches within paper text, with batch retrieval and multiple ID formats.
Key features
- 01Search and retrieve papers with filters and multiple ID formats
- 02Explore citations, references, and paper authors
- 03Search authors and retrieve profiles or papers
- 04Get paper recommendations
- 05Search passages within paper text
- 06Batch retrieval of papers and author profiles
Use cases
- 01Find papers by topic, year, venue, or field
- 02Trace citations and references around a paper
- 03Discover related papers from examples
- 04Search paper text for a specific phrase
MCP support
Exposes Semantic Scholar Academic Graph API tools to MCP clients for papers, authors, recommendations, citations, and text snippets.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
npx -y semanticscholar-mcp-serverSEMANTICSCHOLAR_API_KEYSecretSemantic Scholar API key from https://www.semanticscholar.org/product/api (optional but recommended for higher rate limits)Local servers need a runtime on your machine (e.g. Node.js 18+). Config file locations differ by client; see your client’s docs.
Tools
16Saved tool list · updated 2w ago
| Parameter | Type | Required | Description |
|---|---|---|---|
| year | string | No | Year filter. Single year '2023' or range '2020-2023'. Also supports open ranges: '2020-' or '-2023'. |
| limit | integer | No | Number of results to return (1-100, default: 10) |
| query | string | Yes | Search query string. Supports phrases with quotes and boolean operators. |
| venue | string | No | Filter by publication venue name, e.g. 'Nature', 'NeurIPS', 'ICLR'. Comma-separated for multiple. |
| fields | string | No | Comma-separated fields to return, overriding defaults. Paper fields: paperId, title, abstract, authors, year, citationCount, referenceCount, influentialCitationCount, isOpenAccess, openAccessPdf, fieldsOfStudy, externalIds, url, venue, publicationVenue, publicationTypes, publicationDate, journal, citations, references. Author fields: authorId, name, affiliations, homepage, paperCount, citationCount, hIndex. |
| offset | integer | No | Offset for pagination (default: 0) |
| fieldsOfStudy | string | No | Comma-separated fields of study, e.g. 'Computer Science,Medicine'. Valid values: Computer Science, Medicine, Physics, Mathematics, Biology, Chemistry, etc. |
| openAccessPdf | boolean | No | If true, only return papers with an open access PDF available. |
| response_format | enum | No | Output format: 'markdown' for human-readable text (default), 'json' for raw structured datamarkdown | json |
| minCitationCount | integer | No | Minimum number of citations a paper must have. |
| publicationTypes | string | No | Comma-separated publication types: JournalArticle, Conference, Review, Book, BookSection, Preprint, LettersAndComments, ClinicalTrial, CaseReport, Editorial, News. |
| publicationDateOrYear | string | No | Filter by publication date. Supports ranges: '2019-03-05:2020-06-15', '2019-03:', ':2020-06'. |
Details
From the project READMESemantic Scholar MCP Server
An MCP (Model Context Protocol) server for the Semantic Scholar Academic Graph API — 200M+ papers with citation networks, author profiles, and full-text snippet search.
Built with the MCP TypeScript SDK and Zod.
Tools (16 total)
Papers (9 tools)
| Tool | Description |
|---|---|
paper_search | Search papers by keyword with filtering (year, venue, field, open access, citations, type) |
paper_search_bulk | Bulk search with cursor pagination for large result sets (up to 10M) |
paper_match | Find the single best match for a known paper title |
paper_autocomplete | Autocomplete partial paper titles (up to 10 suggestions) |
paper_get | Get full details for a paper by ID (S2 ID, DOI, ArXiv, PMID, etc.) |
paper_batch | Retrieve up to 500 papers in a single request |
paper_citations | Get papers that cite a given paper (forward citations) |
paper_references | Get a paper's reference list (bibliography) |
paper_authors | Get author details for a specific paper |
Authors (4 tools)
| Tool | Description |
|---|---|
author_get | Get full author profile (h-index, affiliations, stats) |
author_batch | Retrieve up to 1000 author profiles in one request |
author_search | Search for authors by name |
author_papers | Get all papers by a specific author |
Recommendations (2 tools)
| Tool | Description |
|---|---|
recommendations_for_paper | Get ML-powered paper recommendations for a given paper |
recommendations_from_lists | Get recommendations from positive/negative example papers |
Snippets (1 tool)
| Tool | Description |
|---|---|
snippet_search | Search within paper bodies for specific text passages |
Setup
1. Get an API key (optional but recommended)
Get a free key at semanticscholar.org/product/api. Without a key, requests are heavily rate-limited.
2. Install and build
cd semanticscholar-mcp-server
npm install
npm run build
3. Add to Claude Desktop
Add this to your claude_desktop_config.json:
{
"mcpServers": {
"semantic-scholar": {
"command": "node",
"args": ["/path/to/semanticscholar-mcp-server/dist/index.js"],
"env": {
"SEMANTICSCHOLAR_API_KEY": "your-api-key"
}
}
}
}
Or if using Claude Code CLI:
claude mcp add semantic-scholar \
node \
/path/to/semanticscholar-mcp-server/dist/index.js \
-e SEMANTICSCHOLAR_API_KEY=your-api-key
Features
- Multiple ID formats: Accepts S2 Paper IDs, DOIs, ArXiv IDs, PMIDs, PMCIDs, CorpusIds
- Markdown or JSON output: Every tool supports
response_format: "json"for structured data - Custom field selection: Override default fields with the
fieldsparameter - Built-in rate limiting: Token bucket limiter prevents API throttling
- Batch operations: Retrieve up to 500 papers or 1000 authors in a single call
Usage examples
- "Search Semantic Scholar for papers on transformer architectures from 2023"
- "Get recommendations for papers similar to this one"
- "Find the citation network for DOI 10.1038/nature14539"
- "Search for text snippets mentioning 'chain of thought prompting'"
- "Look up author profile for researcher 1741101"
License
MIT
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.