Skip to content

ALMa

Another Library Manager — a personal, self-hosted academic literature monitor.

ALMa watches the publication record on your behalf, surfaces new work from the people and topics you care about, recommends adjacent papers you haven't seen, and maintains the curated library that anchors all of it. One user, one machine, one SQLite file.

  • Get started in 5 minutes

    Install, point ALMa at OpenAlex, and follow your first author.

    Installation

  • Understand the model

    Feed, Library, and Discovery are three views of the same paper lifecycle. Read this once, the rest of the UI clicks into place.

    Vision & philosophy

  • Browse concepts

    What each surface does, what it doesn't, and how they hand off work to each other.

    Concepts

  • Use the REST API

    Every endpoint ALMa exposes, grouped by domain, with live Swagger UI for try-it-yourself.

    REST API

What ALMa is

A personal research feed — designed for one user with their own literature, not a multi-tenant SaaS. ALMa runs entirely on your machine, stores everything in one SQLite file (data/scholar.db), and only talks to the outside world when it's fetching a paper or a profile from a public source.

It pulls primarily from OpenAlex and Semantic Scholar, with optional fall-throughs to Crossref, arXiv, bioRxiv, and Google Scholar (scholarly) for author resolution. Embedding-based discovery is opt-in and uses SPECTER2 either via Semantic Scholar's pre-computed vectors or a local fall-back.

What ALMa is not

  • Not a citation manager. It doesn't replace Zotero or Mendeley. It imports from them so the work you've already curated becomes the seed for discovery.
  • Not a multi-user platform. There is no auth, no permissions, no sharing — by design.
  • Not a SaaS. You host it yourself. The only network calls are to public scholarly APIs and (optionally) AI providers you configure.
  • Not a search engine. It's a recommendation engine over a curated personal corpus. Search is a side effect, not the point.

How the parts fit together

Monitors generate the Feed (chronological, deterministic). Saving from the Feed grows the Library (curated, organised). Discovery recommends papers adjacent to the Library (probabilistic, ranked). Every save, rating, removal, and tracked interaction strengthens the feedback loop that tunes Discovery over time. Insights projects the Library into charts and a clustered SPECTER2 graph. Alerts turns rules into Slack digests.

License

CC BY-NC 4.0 © Andrea Ivan Costantino. Personal and academic use are free; academic use requires a citation. Commercial use is not permitted.