Eprints don't live on Chive's servers. They live in your Personal Data Server (PDS): the same infrastructure that powers @bsky.app. If Chive shuts down, your papers are still there, so you can switch to a different indexer or host your own. Chive is designed to be replaceable.

Some servers have opaque moderation. Chive takes a simpler approach: you submit; it's live. No institutional email required. No account endorsement system. The community handles quality through open review and non-blocking eprint endorsement.

PDFs aren't the only format researchers work in. Chive accepts LaTeX, Markdown, HTML, Jupyter notebooks, EPUB, and more. Submit your computational notebook directly. Keep your equations in LaTeX. The content is what matters, not the container.

Reviews are tied to your Bluesky identity and are public. Reviewers build reputations over time. You can leave inline annotations on specific passages or general comments. There's also an endorsement system for formally endorsing specific aspects of an eprint, like methodology.

When annotating a paper, you can link text spans to Wikidata. For instance, you could select a term like "lambda calculus" and connect it to its Wikidata Q-node. Over time this builds a semantic layer across all papers. Search for a concept and find every paper where someone linked that term.

Already published on arXiv or have an ORCID? Claim those papers and link them to your Chive identity. Chive supports ORCID OAuth, Semantic Scholar linking, and co-author network matching.

Chive has a Wikipedia-style taxonomy for classifying research. Anyone can propose a new field, the community discusses it, and trusted editors approve or reject. It's backed by Neo4j and integrated with Wikidata, so fields are semantic entities connected to a broader knowledge graph.

Alongside the formal taxonomy, anyone can tag papers with their own terms. Think of it as a folksonomy layer on top of the knowledge graph. Trending tags surface organically, so you can search by tags to find related work that the formal classification might miss.

The discovery system recommends papers based on your claimed papers, research interests, and citation networks. Chive pulls enrichment data from Semantic Scholar and OpenAlex. Each recommendation shows why it was suggested so you can tune what signals matter to you.

Rather than just listing authors in order, Chive tracks who did what using the CRediT taxonomy: Conceptualization, Methodology, Software, Validation, etc. The community can also propose new contribution types for things CRediT doesn't cover, like replication studies or reproducibility verification.

Chive supports detailed metadata for funding sources (with CrossRef and ROR identifiers), conference presentations, and pre-registrations from OSF or ClinicalTrials.gov. Supplementary materials like datasets, code, and protocols can also be properly linked to your eprint.

Chive tracks citation relationships across the network. You can see who's citing your work and discover citation chains through an interactive network visualization. Trending includes velocity indicators showing whether interest is accelerating or cooling off, not just raw counts.

Chive integrates with @semble.app collections, @leaflet.pub reading lists, and @whtwnd.com blogs. When someone adds your paper to a Semble collection or cites it in a WhiteWind blog post, Chive automatically creates a backlink. All of this happens over the AT Protocol firehose.

You use the same identity across Chive, @bsky.app, @semble.app, and @leaflet.pub. No new account needed. You can also share eprints directly to Bluesky with rich previews.

Everything is public at github.com/chive-pub/chive The stack is TypeScript, Hono, PostgreSQL, Elasticsearch, Neo4j, and Next.js.