Paper Search · adamamer20 — Claude Code MCP server: search papers, extract PDF text
Paper Search — a remote MCP server that searches arXiv, PubMed, bioRxiv, medRxiv, Google Scholar, Semantic Scholar, IACR and CrossRef, downloads PDFs and pulls out text for literature reviews.
no license 59,106 installs updated 2025-09-18
claude mcp add --transport http paper-search-mcp-openai https://paper-search-mcp-openai--adamamer20.run.toolsWhat Paper Search does
The server runs remotely at https://paper-search-mcp-openai--adamamer20.run.tools. Each source gets its own search tool (search_arxiv, search_pubmed, search_semantic and the rest). On top of those sit search, which merges hits from several databases, and fetch, which pulls the whole document for a hit. Metadata comes back in one shape across sources.
arXiv, bioRxiv, medRxiv and IACR each have a download_* and a read_* tool. The first saves the PDF to ./downloads, the second returns the extracted text, and save_path changes the folder. Semantic Scholar has the same pair, and search_semantic can narrow results by year.
PubMed and CrossRef are different. download_pubmed and download_crossref only reply that PDF download is unsupported, and the matching read_* tools behave the same way. Use those two for metadata, including get_crossref_paper_by_doi for a lookup by DOI. Google Scholar is search only.
What it helps with
- find preprints on a topic across arXiv and bioRxiv
- download an arXiv paper by ID into the project folder
- extract text from an IACR ePrint paper to review its protocol
- limit Semantic Scholar results to a range of years
- pull citation metadata for a DOI from CrossRef
Example prompts for the agent
- Use Paper Search to find IACR ePrint papers on secret sharing and download the best match to
./downloads - Run
search_semanticfor machine learning papers from 2016-2020 and list their metadata - Read arXiv paper
2106.12345with Paper Search and summarize the method it proposes
How to install Paper Search
Platform CLI: remote server
claude mcp add --transport http paper-search-mcp-openai https://paper-search-mcp-openai--adamamer20.run.toolsPlatform CLI: remote server
codex mcp add paper-search --url https://paper-search-mcp-openai--adamamer20.run.toolsConfig file
{
"mcpServers": {
"paper-search": {
"url": "https://paper-search-mcp-openai--adamamer20.run.tools"
}
}
}Config file
{
"servers": {
"paper-search": {
"type": "http",
"url": "https://paper-search-mcp-openai--adamamer20.run.tools"
}
}
}Platform CLI: remote server
gemini mcp add -t http paper-search https://paper-search-mcp-openai--adamamer20.run.toolsStep by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about Paper Search
How do I install Paper Search in Claude Code?
Run claude mcp add --transport http paper-search-mcp-openai https://paper-search-mcp-openai--adamamer20.run.tools, then claude mcp add --transport http paper-search https://paper-search-mcp-openai--adamamer20.run.tools.
What does Paper Search do?
It searches arXiv, PubMed, bioRxiv, medRxiv, Google Scholar, Semantic Scholar, IACR and CrossRef, downloads PDFs where the source allows it, and hands the agent the paper text to read.
Can Paper Search download PDFs from PubMed?
No. download_pubmed and read_pubmed_paper return a message that direct download is unsupported. PDFs work for arXiv, bioRxiv, medRxiv, IACR and Semantic Scholar.
Similar MCP servers
All MCP servers for Science and research →Paper Search by other authors comes first.