What it does
Paper Search MCP provides programmatic access to academic papers across 20+ platforms including arXiv, PubMed, bioRxiv, medRxiv, Google Scholar, Semantic Scholar, Crossref, OpenAlex, and others. It exposes two primary tools: search_papers for concurrent multi-source search and deduplication, and download_with_fallback which chains retrieval from source-native repositories, OpenAIRE, CORE, Unpaywall DOI metadata, and optional Sci-Hub. Results are returned in a standardized Paper format compatible with downstream LLM workflows. Installation is available via pip, uvx, or Smithery.
Who it's for
Researchers building systematic literature reviews, AI engineers wiring Claude into academic databases for real-time retrieval, and teams developing LLM-driven research assistants or knowledge synthesis tools that need lawful, standardized access to open and publisher-provided papers.
Common use cases
- Search for papers on a topic across arXiv, PubMed, and bioRxiv simultaneously to get discipline-specific breadth.
- Download preprints and open-access PDFs with automatic fallback across multiple repositories when the primary source is unavailable.
- Extract text from downloaded papers and pass abstracts or full text to Claude for summarization or analysis.
- Backfill DOI metadata from Crossref or Unpaywall to discover open-access links not listed in the original search result.
Setup pitfalls
- Optional API keys improve quality and rate limits for Semantic Scholar and other sources, but the server works without them—plan accordingly for production scale.
- Some sources like Google Scholar employ bot detection; the server treats these as discovery-only, not full-text retrieval.
- Requires filesystem write access to download and cache PDFs; isolate or configure as needed in sandboxed environments.
- Sci-Hub is included as an optional fallback strategy but operates in a legal gray area; adjust your fallback chain or policies based on your jurisdiction.