Skip to content
MCP server

Paper Librarian

Local searchable library of research papers with cited answers for Claude

Runs locally
Install

Paper Librarian is an open-source MCP server for health and science. It runs locally on your machine and is started by your AI client.

Pricing
Open source
Platforms
CLI, macOS, Linux
Runs
Locally
Version
0.1.0
Listing updated

Similar MCP servers

Install

Run in a terminal

claude mcp add paper-librarian -- uvx paper-librarian

Add to claude_desktop_config.json

{
  "mcpServers": {
    "paper-librarian": {
      "command": "uvx",
      "args": [
        "paper-librarian"
      ]
    }
  }
}
Install in CursorOpens Cursor, which asks you to confirm. Or add it by hand:

Add to ~/.cursor/mcp.json

{
  "mcpServers": {
    "paper-librarian": {
      "command": "uvx",
      "args": [
        "paper-librarian"
      ]
    }
  }
}
Install in VS CodeOpens VS Code, which asks you to confirm. Or add it by hand:

Add to .vscode/mcp.json

{
  "servers": {
    "paper-librarian": {
      "type": "stdio",
      "command": "uvx",
      "args": [
        "paper-librarian"
      ]
    }
  }
}

Add to ~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "paper-librarian": {
      "command": "uvx",
      "args": [
        "paper-librarian"
      ]
    }
  }
}

Add to your client's MCP settings (most use this shape)

{
  "mcpServers": {
    "paper-librarian": {
      "command": "uvx",
      "args": [
        "paper-librarian"
      ]
    }
  }
}

Using several servers? Build one config file with the MCP config generator.

Configuration

  • LIBRARIAN_HOMEFolder for the library database and downloaded PDFs (default: ~/.librarian)

About Paper Librarian

Paper Librarian keeps your research PDFs and arXiv papers in a single local database that an MCP client can query. Papers are added by arXiv id, URL or file path, then split into sections, pages and searchable passages, so answers can point back to the section and page they came from.

Search combines meaning-based vector matching with keyword ranking. Reference lists are parsed and matched to other papers in the library, so you can see which papers cite each other in both directions. Notes can be attached to any paper, and a section reader and a comparison helper pull relevant passages for side-by-side tables.

It suits students and researchers who work from a stack of papers and want an assistant to answer questions from the actual text. Papers, the database and the embedding model stay on the local machine, apart from arXiv downloads and a one-time model download.

Tools this server provides (8)

Taken from the project documentation; the running server may differ.

add_paper
Adds a paper by arXiv id or PDF path.
search
Searches the library in hybrid, semantic or keyword mode with year and paper filters.
get_paper
Returns paper metadata, notes, or one section's full text.
get_citations
Lists a paper's references and the papers that cite it.
compare_papers
Gathers passages from several papers for each comparison aspect.
add_note
Attaches a note to a paper.
list_papers
Lists papers in the library.
remove_paper
Removes a paper from the library, keeping its PDF file.

What it does

Cited passage search
Hybrid vector and keyword search returns passages labelled with paper, section and page.
Two-way citation links
Reference lists are matched to library papers and linked forwards and backwards as papers are added.
arXiv and PDF ingestion
Adds papers from an arXiv id, arXiv URL or a local PDF, skipping duplicates by id or file hash.
Section reader
Retrieves the full text of a named section from a stored paper.
Paper comparison
Gathers the most relevant passages per paper for each aspect you want compared.
Attached notes
Stores free-text notes against a paper and returns them when the paper is looked up.
Local storage
Keeps the database, PDFs and embedding model on your own machine in one folder.

Good for

  • Ask questions across a set of papers and get answers that cite section and page
  • Map which papers in a reading list cite each other
  • Build a comparison table of methods or results across several papers
  • Draft a literature review from papers already in the library
  • Keep private research PDFs searchable without uploading them to a service

Pricing

Open source

Paper Librarian FAQ

Does Paper Librarian need an external service?

Only arXiv for downloads and a one-time download of the embedding model; the database and PDFs stay local.

Can Paper Librarian read scanned PDFs?

No. Scanned PDFs without a text layer are rejected, and there is no OCR.

Which clients can use Paper Librarian?

It runs over stdio and is set up for Claude Code and Claude Desktop, and works with any MCP client.

Where is the library stored?

In the folder set by LIBRARIAN_HOME, defaulting to ~/.librarian, which holds library.db and a pdfs folder.

Is Paper Librarian free?

Yes. Paper Librarian is open source, so you can use and self-host it for free.

How do I add Paper Librarian to Claude Code?

Run this command in your terminal: claude mcp add paper-librarian -- uvx paper-librarian

What are the best alternatives to Paper Librarian?

The closest alternatives to Paper Librarian are Arxiv MCP Server, Peer Research Papers MCP, Paper Finder, Scholar Feed and Endnote MCP.

Where this comes from

  • Description: written from the project's README and its MCP Registry entry.
  • Install steps: from the official MCP Registry entry.

How AIPoweredBot checks servers