M
MCP Directory CURATED DIRECTORY
911 MCP servers indexed
中文
← Back to directory
P

Paper Search MCP

Community
MCP server for searching and downloading academic papers from multiple sources
How to install ↓ Compare GitHub source repository ↗
Category
Other #57 of 230
Stars
★ 2.8k Very popular
Transport
stdio (local process)
Runtime
Python 3.10+ · Node.js · Docker
Credentials
Optional API key
License
MIT
Last commit
Tools
2
62FMRS · C

This project is feature-rich, supporting multiple academic sources with a free-first approach and various installation methods. However, it has some upstream dependencies and optional key configuration complexity, suitable for users with moderate technical background.

Strongest · Documentation 14/20 Weakest · Reliability 9/20

Reliability
9/20
Security and permissions
12/20
Maintenance
14/20
Documentation
14/20
Setup experience
13/20
Why each score
Reliability 9/20
Source code not available; cannot verify actual runtime behavior or tool list consistency. README claims extensive functionality and tests, but lacks executable evidence. Conservative score given; deducted for lack of source verification and potential instability of external dependencies (e.g., Google Scholar, SSRN bot detection).
Security and permissions 12/20
Source code unavailable for detailed review of credential handling and least privilege. README shows API keys managed via environment variables, optional and not forced, which seems reasonable. However, there is legal risk from Sci-Hub integration, and no code-level security audit evidence. Not a red-line issue, but transparency gap limits score.
Maintenance 14/20
Repository is active with 2363 stars, 48 open issues, and README shows frequent updates. MIT license. However, no concrete evidence of commit history, release cadence, or issue response times. Dependency updates and governance unclear, so slightly below full.
Documentation 14/20
README is extremely detailed, covering install methods, environment variables, platform capability matrix, known limitations, and troubleshooting. Excellent. However, some content may be based on unverified claims, and source code not available to cross-check tool parameters. Deducted for inability to verify doc-impl consistency.
Setup experience 13/20
Multiple installation paths (pip, uvx, uv, npx, Docker, source) with concrete client config examples. Steps are clear. But reliance on external tools (uv) and potential platform issues (e.g., macOS realpath), and inability to verify installation process statically cap score.

Static review · not runListed 2026-08-07

Read the FMRS scoring method →

Fit and risk

What it can accessUses the network

Best for

  • Researchers needing to search and download papers from multiple open-access sources
  • Developers wanting to integrate paper retrieval into MCP clients like Claude Desktop
  • Cost-conscious users who prefer free public data sources

Not for

  • Users needing full features from commercial databases (e.g., IEEE, ACM)
  • Users requiring complete full-text downloads without configuring optional API keys or proxies (some sources are limited)
  • Users running in environments without network access or strict firewalls

Required permissions

  • Requires network access to call multiple academic APIs
  • May need to write to the local filesystem to download PDFs
  • Optional environment variables for API keys, stored in the user config directory

Risks and side effects

  • Dependency on external API stability; some sources may fail due to upstream limitations
  • Google Scholar may trigger bot detection and require proxy settings
  • Use of Sci-Hub carries legal and policy risks; users are responsible

Setup

Before you start

Runtime：Python 3.10+ · Node.js · Docker

PAPER_SEARCH_MCP_UNPAYWALL_EMAIL required Required for Unpaywall (source disabled without it); any valid email, register at unpaywall.org
PAPER_SEARCH_MCP_CORE_API_KEY optionalsecret Recommended for CORE, free at core.ac.uk/services/api; reduces rate-limit errors
PAPER_SEARCH_MCP_SEMANTIC_SCHOLAR_API_KEY optionalsecret Optional Semantic Scholar key, free at semanticscholar.org; improves rate limits
PAPER_SEARCH_MCP_ZENODO_ACCESS_TOKEN optionalsecret Optional Zenodo access token, generate at zenodo.org account settings; needed for private records
PAPER_SEARCH_MCP_IEEE_API_KEY optionalsecret Activation key for the IEEE Xplore skeleton connector, free at developer.ieee.org
PAPER_SEARCH_MCP_ACM_API_KEY optionalsecret Activation key for the ACM DL skeleton connector, see libraries.acm.org/digital-library/acm-open
PAPER_SEARCH_MCP_DOAJ_API_KEY optionalsecret Optional DOAJ key, free at doaj.org; raises hourly rate limit
Other optional settings (2)
PAPER_SEARCH_MCP_GOOGLE_SCHOLAR_PROXY_URL optional Optional HTTP/HTTPS proxy URL to bypass Google Scholar bot detection
PAPER_SEARCH_MCP_ENV_FILE optional Optional; absolute path to a custom .env config file

Recommended installation via uvx (requires uv). Add the configuration to your Claude Desktop config (see install_config), or use Smithery one-liner: npx -y @smithery/cli install @openags/paper-search-mcp --client claude. Alternatively, install via pip, Docker, or run from source. API keys can be stored in ~/.config/paper-search-mcp/.env.

claude_desktop_config.json 
{
  "mcpServers": {
    "paper-search-mcp": {
      "command": "uvx",
      "args": [
        "paper-search-mcp"
      ],
      "env": {
        "PAPER_SEARCH_MCP_UNPAYWALL_EMAIL": "[email protected]",
        "PAPER_SEARCH_MCP_CORE_API_KEY": "",
        "PAPER_SEARCH_MCP_SEMANTIC_SCHOLAR_API_KEY": "",
        "PAPER_SEARCH_MCP_ZENODO_ACCESS_TOKEN": "",
        "PAPER_SEARCH_MCP_GOOGLE_SCHOLAR_PROXY_URL": "",
        "PAPER_SEARCH_MCP_IEEE_API_KEY": "",
        "PAPER_SEARCH_MCP_ACM_API_KEY": ""
      }
    }
  }
}

Shown for Claude Desktop. Other clients may use a different file or key (VS Code uses "servers") — the configurator below converts it.

.vscode/mcp.json
{
  "servers": {
    "paper-search-mcp": {
      "command": "uvx",
      "args": [
        "paper-search-mcp"
      ],
      "env": {
        "PAPER_SEARCH_MCP_UNPAYWALL_EMAIL": "[email protected]",
        "PAPER_SEARCH_MCP_CORE_API_KEY": "",
        "PAPER_SEARCH_MCP_SEMANTIC_SCHOLAR_API_KEY": "",
        "PAPER_SEARCH_MCP_ZENODO_ACCESS_TOKEN": "",
        "PAPER_SEARCH_MCP_GOOGLE_SCHOLAR_PROXY_URL": "",
        "PAPER_SEARCH_MCP_IEEE_API_KEY": "",
        "PAPER_SEARCH_MCP_ACM_API_KEY": ""
      }
    }
  }
}

Goes in your project's .vscode/mcp.json (VS Code uses a "servers" key).

Terminal
claude mcp add paper-search-mcp -e [email protected] -e PAPER_SEARCH_MCP_CORE_API_KEY= -e PAPER_SEARCH_MCP_SEMANTIC_SCHOLAR_API_KEY= -e PAPER_SEARCH_MCP_ZENODO_ACCESS_TOKEN= -e PAPER_SEARCH_MCP_GOOGLE_SCHOLAR_PROXY_URL= -e PAPER_SEARCH_MCP_IEEE_API_KEY= -e PAPER_SEARCH_MCP_ACM_API_KEY= -- uvx paper-search-mcp

Run it in a terminal; replace any <…> placeholders with your own values first.

Generate and validate client config

Check that it works

After starting Claude Desktop, search_papers and download_with_fallback should appear in the client's tool list; you can also ask it to search arXiv for papers on diffusion models to confirm results are returned.

Troubleshooting

  1. Check environment variables, especially `PAPER_SEARCH_MCP_UNPAYWALL_EMAIL` (required for Unpaywall)
  2. If Google Scholar returns empty results, set a proxy URL
  3. If Semantic Scholar hits 429 rate limits, configure an API key
  4. If downloading PMC/Europe PMC PDFs fails with proxy errors, disable proxy or use `download_with_fallback`

Things to try

Once connected, you can ask your AI assistant things like:

  • Search arXiv and Semantic Scholar for papers on transformer attention mechanisms
  • Download the PDF for arXiv paper 2106.12345
  • Find me recent papers on CRISPR base editing
  • Look up the open-access version of DOI 10.1038/s41586-020-2649-2 via Unpaywall

Tools 2

search_papers read-only
Performs concurrent searches across multiple academic sources (e.g., arXiv, PubMed, bioRxiv) with deduplication, returning a standardized list of papers.
download_with_fallback read-only
Downloads PDFs of papers, prioritizing publisher open-access links with sequential fallbacks to other sources, optionally ending with Sci-Hub.

Use cases

Researchers quickly finding and downloading academic papers
Integrating paper search and full-text retrieval into AI workflows
Conducting literature reviews across multiple databases (e.g., arXiv, PubMed)

Supported clients

Claude Desktop

Listed from the project's documentation, not tested by this site.

Overview

Paper Search MCP is a Python-based tool for searching and downloading academic papers from various platforms like arXiv, PubMed, bioRxiv, and more. It provides tools for searching papers, downloading PDFs, and extracting text, making it ideal for researchers and AI-driven workflows. It can be used as an MCP server (for Claude Desktop and other MCP clients) or as a Claude Code skill with a CLI interface. The project follows a free-first strategy: prioritize open and public data sources, support optional API keys when they improve stability or coverage, and keep source-specific connectors extensible.

Similar servers

BioMCP 75 · B

One binary. One grammar. Evidence from the biomedical sources you already trust.

★ 654 · Tools 11 Compare with this →
Compare with other MCP servers

Source revision cab1b5cc276a Data synced 2026-10-12 Read the FMRS scoring method