PRIMA Scholar Search MCP Server

An MCP (Model Context Protocol) server for academic literature search across 10 databases. Designed to give researchers, students, and professionals broad access to scholarly work across every discipline.

PRIMA Scholar Search provides a search wizard that analyses your query, detects your discipline, and recommends the best databases to search. It prioritises open access papers, clearly labels gated content, and can retrieve full text where available.

What It Does

• Searches 10 academic databases simultaneously with automatic deduplication
• Detects your discipline from your query and routes to the strongest databases for that field
• Prioritises open access papers in results, with clear OA/gated labelling on every result
• Guides users through search refinement via a structured wizard before searching
• Retrieves full text for open access papers where available
• Tracks citations forward (who cited this paper?) and backward (what does this paper cite?)
• Formats citations in 6 styles: APA7, Harvard, Chicago, Vancouver, IEEE, MLA

Databases

| Database | Coverage | Auth Required | Open Access | |----------|----------|---------------|-------------| | OpenAlex | 250M+ works, all disciplines | No (polite pool with email) | OA status per paper | | PubMed | Biomedical, life sciences | Optional API key | Varies | | arXiv | Physics, maths, CS, biology, finance, stats | No | All OA (preprints) | | Semantic Scholar | All disciplines + citation graph | Optional API key | OA status per paper | | CrossRef | DOI metadata, all disciplines | No (polite pool with email) | OA via licence metadata | | CORE | 300M+ open access papers | Free API key required | All OA | | Europe PMC | Biomedical + European research council | No | OA status per paper | | ERIC | Education research | No | Full text for many | | bioRxiv/medRxiv | Biology and medicine preprints | No | All OA (preprints) | | DBLP | Computer science bibliography | No | Metadata only |

Installation

cd prima-scholar-search-mcp
npm install
npm run build

Configuration

Claude Desktop

Add to ~/Library/Application Support/Claude/claude_desktop_config.json:

{
  "mcpServers": {
    "prima-scholar-search": {
      "command": "node",
      "args": ["/path/to/prima-scholar-search-mcp/build/index.js"],
      "env": {
        "OPENALEX_MAILTO": "[email protected]",
        "SEMANTIC_SCHOLAR_KEY": "your-key-here",
        "PUBMED_API_KEY": "your-key-here",
        "CROSSREF_MAILTO": "[email protected]",
        "CORE_API_KEY": "your-key-here"
      }
    }
  }
}

Claude Code

Add to ~/.claude.json or project .mcp.json:

{
  "mcpServers": {
    "prima-scholar-search": {
      "command": "node",
      "args": ["/path/to/prima-scholar-search-mcp/build/index.js"],
      "env": {
        "OPENALEX_MAILTO": "[email protected]",
        "SEMANTIC_SCHOLAR_KEY": "your-key-here",
        "PUBMED_API_KEY": "your-key-here",
        "CROSSREF_MAILTO": "[email protected]",
        "CORE_API_KEY": "your-key-here"
      }
    }
  }
}

Environment Variables

| Variable | Required | Purpose | How to Get | |----------|----------|---------|------------| | OPENALEX_MAILTO | No | Polite pool for OpenAlex (higher rate limits) | Use your email address | | CROSSREF_MAILTO | No | Polite pool for CrossRef (higher rate limits) | Use your email address | | PUBMED_API_KEY | No | Higher rate limits (10 req/sec vs 3) | https://www.ncbi.nlm.nih.gov/account/ | | SEMANTIC_SCHOLAR_KEY | No | Higher rate limits | https://www.semanticscholar.org/product/api | | CORE_API_KEY | Yes (for CORE) | Required to use CORE database | https://core.ac.uk/services/api (free) |

The server works without any keys. 9 of the 10 databases require no authentication. Only CORE requires a free API key. Without keys, rate limits are lower but all other databases function normally.

Tools

scholar_wizard

Analyse a research query before searching. Returns:

• Detected disciplines from your query (e.g. psychology, neuroscience)
• Suggested databases optimised for those disciplines
• Structured questions to refine the search (discipline focus, OA preference, date range, preprint inclusion)
• A suggested search configuration with sensible defaults

Call this before scholar_search to guide users through search refinement.

scholar_search

Search across all 10 databases simultaneously. Features:

• Automatic deduplication by DOI (keeps the result with the richest metadata)
• Results sorted: open access first, then by citation count
• Configurable source selection, date range, and OA filtering
• Each result includes openAccess (boolean), openAccessUrl, and fullTextAvailable fields

Parameters:

• query (required) -- search terms
• sources -- array of database names to search (default: all available)
• open_access_only -- filter to OA papers only (default: false)
• year_from / year_to -- publication date range
• max_results -- per source (default: 10)
• citation_style -- return citations in one format only (default: all 6)

scholar_get_paper

Get full metadata for a single paper. Accepts multiple ID formats and routes automatically:

• DOI: 10.1037/0003-066X.55.1.68
• PubMed: PMID:38123456
• arXiv: ARXIV:2301.12345
• OpenAlex: W1234567890
• ERIC: EJ1234567 or ED123456
• Europe PMC: PMC1234567
• Semantic Scholar ID
• DBLP key

scholar_citations

Forward and backward citation tracking.

• direction: "citations" -- papers that cite this one (forward)
• direction: "references" -- papers this one cites (backward)

Uses the Semantic Scholar citation graph.

scholar_full_text

Retrieve full text for open access papers. Tries:

1. CORE (full-text API)
2. Europe PMC (PMC articles)
3. arXiv (PDF link)
4. bioRxiv/medRxiv (preprint PDF)

Returns the text content or a direct download URL.

Discipline Detection

The wizard automatically detects disciplines from query keywords and maps to the strongest databases:

| Discipline | Primary Databases | |------------|-------------------| | Psychology | OpenAlex, Semantic Scholar, PubMed | | Education | ERIC, OpenAlex, Semantic Scholar | | Neuroscience | PubMed, Europe PMC, bioRxiv | | Business and Management | OpenAlex, Semantic Scholar, CrossRef | | Computer Science and AI | Semantic Scholar, arXiv, DBLP | | Philosophy and Humanities | OpenAlex, CrossRef | | Biomedical and Life Sciences | PubMed, Europe PMC, bioRxiv/medRxiv | | Engineering | OpenAlex, Semantic Scholar, arXiv | | Social Sciences | OpenAlex, Semantic Scholar, CrossRef | | Mathematics and Physics | arXiv, Semantic Scholar, OpenAlex | | Economics | OpenAlex, CrossRef, Semantic Scholar |

Open Access Handling

Every paper in search results includes three fields:

| Field | Type | Meaning | |-------|------|---------| | openAccess | boolean | Whether this paper is openly accessible | | openAccessUrl | string or null | Direct URL to the OA version | | fullTextAvailable | boolean | Whether scholar_full_text can retrieve the content |

When open_access_only is false (default), results are partitioned:

• OA papers first, sorted by citation count
• Gated papers second, sorted by citation count

The response includes openAccessCount and gatedCount for transparency.

Example Usage

Guided search (recommended)

1. Call scholar_wizard with query "self-determination theory in workplace motivation"
2. Present the wizard's questions to the user
3. User answers: psychology + education, open access only, last 10 years, include preprints
4. Call scholar_search with the refined parameters

Direct search

scholar_search with:
  query: "transformer architecture attention mechanism"
  sources: ["arxiv", "semantic_scholar", "dblp"]
  open_access_only: true
  year_from: 2020

Citation tracking

scholar_citations with:
  paper_id: "DOI:10.1037/0003-066X.55.1.68"
  direction: "citations"
  max_results: 20

Get full text

scholar_full_text with id: "PMC1234567"

Output Format

Every paper result includes:

• title -- paper title
• authors -- list of author names with optional affiliations
• abstract -- full abstract text
• year -- publication year
• journal -- journal or venue name
• doi -- DOI if available
• url -- direct link to the paper
• source -- which database returned this result
• sourceId -- source-specific identifier
• citationCount -- citation count (where available)
• keywords -- subject keywords
• openAccess -- whether this is an OA paper
• openAccessUrl -- direct OA link
• fullTextAvailable -- whether full text can be retrieved
• citations -- pre-formatted citation strings in up to 6 styles

Rate Limits

The server implements per-source rate limiting to respect API terms:

| Source | Limit (no key) | Limit (with key) | |--------|---------------|-------------------| | PubMed | 3 req/sec | 10 req/sec | | arXiv | 1 req/3 sec | N/A | | Semantic Scholar | 100 req/5 min | Higher | | CrossRef | 50 req/sec | N/A (polite pool) | | OpenAlex | 10 req/sec | N/A (polite pool) | | CORE | 10 req/sec | N/A | | Europe PMC | 10 req/sec | N/A | | ERIC | 10 req/sec | N/A | | bioRxiv/medRxiv | 5 req/sec | N/A | | DBLP | 5 req/sec | N/A |

Migration from v1

v2.0.0 replaces the 11-tool surface with 5 tools. If you were calling per-source tools directly:

| v1 Tool | v2 Equivalent | |---------|---------------| | pubmed_search | scholar_search with sources: ["pubmed"] | | arxiv_search | scholar_search with sources: ["arxiv"] | | semantic_search | scholar_search with sources: ["semantic_scholar"] | | crossref_search | scholar_search with sources: ["crossref"] | | pubmed_get_paper | scholar_get_paper with id: "PMID:xxx" | | arxiv_get_paper | scholar_get_paper with id: "ARXIV:xxx" | | semantic_get_paper | scholar_get_paper with id: "xxx" | | crossref_resolve_doi | scholar_get_paper with id: "10.xxx/xxx" | | semantic_citations | scholar_citations with direction: "citations" | | semantic_references | scholar_citations with direction: "references" |

Licence

MIT