Hi folks,
A couple weeks ago, I found the Open Library's MCP server
<https://smithery.ai/server/@8enSmith/mcp-open-library> and hooked it up to
Claude to make it stop hallucinating book titles when I use it for readers'
advisory.
Well, then I went looking for an MCP server that could do the same for
articles and, finding none that really suited my use case (some providers
have them for their own content), decided to make my own
<https://smithery.ai/server/@tfscharff/doi-mcp>.
The obvious drawback is that it's DOI-based, so it won't return articles
without a DOI while using the MCP. I would expect that to be a bigger
drawback in humanities than in other disciplines.
You can always ask it not to use the MCP. You'll get hallucinations again,
but you'll also get real results that don't have a DOI. Maybe at that
point, you reach out to your librarian. 😁
Let me know if you have any suggestions for how I can improve it! I'm
totally new to this. I just learned about MCP servers two weeks ago!
Thanks,
*Thomas Scharff*
<https://library.wheatoncollege.edu/prf.php?account_id=205579> (he/him/his)
*Director of Discovery & Access*
Madeleine Clark Wallace Library <https://library.wheatoncollege.edu>, G22
Wheaton College <https://wheatoncollege.edu>
26 E. Main Street, Norton, MA 02766 <https://goo.gl/maps/d5JvoKZUEXy>
(508) 286-5601 <+15082865601>
Received on Mon Oct 27 2025 - 11:19:50 EDT