💛 First, a big thank-you!
Two community contributions helped get this release ready. Let’s give their authors some love:
- @cenktekin, thank you for fixing
year_rangeJSON Schema compatibility and catching invalid start/end years in #2. Those date filters now have the validation they need! - @itsalljustaride, thank you for fixing HTTP response handling across JavaScript realms in #4. A subtle compatibility bug, caught and fixed.
We’re glad to have your fixes in ScholarMCP. Thanks for taking the time to contribute! 🙌
🎉 ScholarMCP 1.2.0 is here!
It’s been a while since our last release, and we’re happy to ship this one. 1.2.0 is available on npm and GitHub Packages, with fixes for paper matching, PDF ingestion, and publishing, plus a refreshed documentation site.
Here’s what you’ll find when you update.
🔎 More dependable research results
- The right paper for your DOI. DOI lookups now require an exact match, so an unrelated search result won’t slip in when the paper can’t be found.
- Research merging preserves distinct DOIs, authors, and Unicode titles. Author-year citation checks now match the year as well as the author.
- OpenAlex API keys travel in authentication headers instead of appearing in result provenance URLs. DOI lookups use the same authentication.
- Date filters use
year_range: { "start": 2020, "end": 2024 }. Arrays are not accepted, and the start year cannot exceed the end year.
🛠️ Fewer PDF and connection headaches
PDF ingestion skips the landing-page request when a PDF URL is already known. Downloads now have timeouts and size limits, check the PDF signature, and release parser resources after failures.
HTTP transport includes the community response-handling fix and correctly handles IPv6 loopback origins. Both publishing workflows now verify the registry version and run the installed CLI. We also refreshed dependencies and included the MIT license in published packages.
📚 Fresh docs, easier to explore
The docs website now runs on Astro 7.3.3, with a mobile navigation menu and search that finds API parameters and configuration names inside pages. Try searching for year_range!
Under the hood, the site uses native Markdown, bundled TypeScript controls, and CSS sticky navigation. The API reference now reflects object schemas and optional inputs with defaults. Configuration and ingestion guides explain parser fallbacks, job storage, size limits, and reserved settings. Section breadcrumbs link to existing pages.
📦 Ready to update?
npm install -g scholar-mcp@1.2.0
scholar-mcp --versionUsing GitHub Packages? Install @lstudlo/scholar-mcp@1.2.0 with the
registry authentication instructions.
Use Node.js 22.14 or newer; Node.js 24 LTS is recommended. The package also supports Node.js 20.19 and later in the Node.js 20 series. Restart your MCP client after updating.
✅ Checked before release
At release preparation, all 68 MCP tests and four docs tests passed. All 16 documentation pages built, with 669 internal links checked. The dependency audit reported zero known advisories.
A few things to know
Citation checks remain heuristic. OCR and cloud semantic extraction are not implemented. HTTP clients share the process’s document store, so use trusted clients and appropriate file/network access controls.
Thanks for using ScholarMCP, testing it, and helping it improve. Found a bug or have an idea for your research workflow? Open an issue. We’d love to hear it.
Happy researching! 🔬