IEEE Xplore, in your AI assistant.

How it works

1. InstallOne npm command installs the server from the latest GitHub release. Adding it to your client takes one more line, or a short JSON block.
2. Add a free OpenAlex keySearch, metadata and citations come from OpenAlex, restricted to IEEE publications. A free key keeps searches from being throttled. An IEEE Xplore API key is optional and adds IEEE's own search filters.
3. Sign in to your library onceOnly needed for paywalled full text. Point the server at your library's proxied Xplore address and run ieee-xplore-mcp login. Your university's own sign-in page opens in a browser window.
4. AskOpen your assistant and ask about the literature. It searches, follows citations, reads papers and writes the references.

Everything runs on your machine. The session, the paper cache and the browser profile live in ~/.ieee-mcp. There is no hosted backend.

Things people ask it

• Find the most cited IEEE papers on chopper-stabilised amplifiers since 2018.
• What has Kofi Makinwa published in JSSC in the last three years?
• Read the second result and summarise its architecture section.
• Which recent papers cite this one, and what did they improve?
• Give me BibTeX for these five DOIs.
• Find papers on spotting wake words with a microwatt power budget.
• Survey the IEEE literature on time-to-digital converters since 2020.
• Save the PDF of this paper to my downloads folder.

What it covers

Search

Keywords or boolean queries with AND, OR, NOT and quoted phrases, narrowed by title, author, affiliation, venue, year, content type and open access. Semantic search finds papers from a plain-language description. Sort by relevance, citations or date.

Paper details

Authors with affiliations, venue, date, DOI and IEEE article number, citation counts, open-access status, keywords and the abstract. A DOI, article number, Xplore URL or OpenAlex id all work as input.

Citation graph

Papers that cite a paper, its reference list and closely related work, each sortable and paged, so the assistant can walk forwards and backwards from a starting point.

Full text

read_paper returns the text in chunks from an open-access copy or through your institution's proxy. download_pdf saves the file. Open-access PDFs are checked against the paper's title before use.

References

BibTeX, RIS and CSL-JSON, or formatted IEEE, APA, Chicago, Harvard and MLA references, for up to 50 papers in one call.

No key required

Without an IEEE API key everything runs on OpenAlex and doi.org. With one, search uses IEEE's own filters, and when IEEE rejects the key or its quota runs out, the server switches to OpenAlex and says so.

Tools

There is also a literature_review prompt that walks the assistant through a structured survey of a topic. Every input field is listed in docs/tools.md.

Connect a client

The server speaks MCP over stdio. Both keys below are optional: leave out what you do not use.

claude mcp add ieee-xplore -s user \
  -e OPENALEX_API_KEY=your-key \
  -e IEEE_PROXY_URL=https://ieeexplore-ieee-org.tudelft.idm.oclc.org \
  -- ieee-xplore-mcp

{
  "mcpServers": {
    "ieee-xplore": {
      "command": "ieee-xplore-mcp",
      "env": {
        "OPENALEX_API_KEY": "your-key",
        "IEEE_PROXY_URL": "https://ieeexplore-ieee-org.tudelft.idm.oclc.org"
      }
    }
  }
}

{
  "servers": {
    "ieee-xplore": {
      "type": "stdio",
      "command": "ieee-xplore-mcp",
      "env": {
        "OPENALEX_API_KEY": "your-key",
        "IEEE_PROXY_URL": "https://ieeexplore-ieee-org.tudelft.idm.oclc.org"
      }
    }
  }
}

[mcp_servers.ieee-xplore]
command = "ieee-xplore-mcp"
env = { OPENALEX_API_KEY = "your-key", IEEE_PROXY_URL = "https://ieeexplore-ieee-org.tudelft.idm.oclc.org" }

Get the OpenAlex key at openalex.org/settings/api. If a desktop app cannot find ieee-xplore-mcp (common with nvm), use the absolute paths of node and the package's dist/cli.js. All settings and troubleshooting are in docs/clients.md.

Institutional access

If your university subscribes to IEEE Xplore through EZproxy or OCLC, the server reads paywalled papers the same way you would in a browser.

1. Find your proxy addressOpen IEEE Xplore from your library's website and copy the address. It looks like https://ieeexplore-ieee-org.<institution>.idm.oclc.org. Set it as IEEE_PROXY_URL.
2. Sign in onceRun ieee-xplore-mcp login. A browser window opens on your library's sign-in page. Sign in as usual, including MFA; the window closes once Xplore loads.
3. Readread_paper and download_pdf now fetch PDFs through the proxy. An expired session renews silently from the saved profile. If your university asks for the password again, the tools report AUTH_REQUIRED and you sign in once more.

Library licences allow personal use and forbid systematic downloading, and heavy downloading can get a whole institution blocked. So the server fetches one paper per request, waits at least 10 seconds between proxy downloads, stops at 40 a day by default and caches every paper, so rereading costs nothing. It has no bulk download tool.

Where the data comes from

Metadata comes only from documented public APIs; the server does not scrape IEEE Xplore pages. Every search result names the source that answered it.

Privacy

• Runs locally. There is no hosted service and no account with anyone but your library.
• Proxy cookies and the browser profile live under ~/.ieee-mcp with owner-only permissions and never reach the model.
• Your password and MFA are typed only into your university's own sign-in page.
• download_pdf writes only inside your home directory, never into hidden folders, and never overwrites a file.
• ieee-xplore-mcp logout removes the session and profile; ieee-xplore-mcp cache clear removes cached papers.

Questions