Alterlab Preprint Deposition

Preprint Deposition — Pick a Server, Prepare the Submission, Link the Published Version

The write/deposit side of preprinting. Given a finished manuscript, this skill turns "I want to post a preprint" into a concrete, server-specific plan: which server fits the field, what metadata and category to enter, which license to choose (and what it commits you to), how versioning and the preprint DOI work, whether the target journal even allows a preprint, and how to link the preprint to the article once it is published.

It is deliberately the deposit-side complement to the read-side connectors alterlab-arxiv and alterlab-biorxiv (which search and fetch preprints) and to alterlab-open-science (which chooses data repositories and writes DMPs). This skill does not reimplement their search/metadata code — it calls them.

Quick Start

I finished a CS paper — how do I post it to arXiv? Which category and license?
Should this biology manuscript go on bioRxiv or medRxiv?
Does Elsevier's journal X allow me to post a preprint before submission?
My preprint just got accepted — how do I link the published DOI to the arXiv version?
What license should I pick on arXiv if I might publish in a closed journal later?

→ Identify the field → route to a server (see the matrix) → assemble metadata → choose a license → run the journal-policy check before posting → after acceptance, link the published DOI back to the preprint.

When to Use This Skill

Use this skill when the user wants to act on a preprint — post it, choose where, prepare its metadata/license, or reconcile it with a journal or a published version. Core jobs:

1. Server selection — match field + manuscript type to arXiv, bioRxiv, medRxiv, SSRN, or OSF Preprints. See references/server_selection.md.
2. Submission metadata — title, authors + ORCID, abstract, **arXiv primary
   • cross-list categories** or bioRxiv/medRxiv subject collection, funding, declarations. See references/submission_metadata.md.
3. License choice — pick from each server's actual license set and explain the downstream commitment (e.g. CC BY is irrevocable; a later journal may object to a permissive preprint license). See references/licensing.md.
4. Versioning & DOI — arXiv versions (v1, v2, …) are immutable and permanent; a withdrawal is a new version with a tombstone, never a deletion; bioRxiv/medRxiv assign a DOI on posting and accept revisions.
5. Journal preprint policy — query the Sherpa Romeo v2 API for the target journal's prearchiving (preprint) policy before posting. See references/journal_policy.md.
6. Post-publication linking — connect the preprint to the published article (publisher field on the server; the keyless api.biorxiv.org /pubs/ endpoint surfaces bioRxiv→journal links).

Does NOT Trigger — route adjacent requests to the right sibling

This skill answers "how and where do I deposit this manuscript as a preprint, and is that compatible with my journal plans?" It makes no claim about manuscript quality, novelty, or whether the work should be published.

Server Matrix (summary — full detail in references/server_selection.md)

Category/license rows above name the real option sets each server presents at submission. License implications (irrevocability, journal friction) are in references/licensing.md; never assert a "best" license without stating the trade-off.

Workflow

1. Determine the field and route to a server

Read the manuscript's domain. STEM-formal (physics/math/CS/stat/eess/econ/q-bio/ q-fin) → arXiv. Biology → bioRxiv. Clinical/health → medRxiv. Social science/law/economics → SSRN (or arXiv econ). Cross-disciplinary or a field-specific community server → OSF Preprints. Edge cases and the full decision tree live in references/server_selection.md.

2. Run the journal-policy check FIRST (if a target journal is known)

Before posting, confirm the intended journal permits preprints. Use scripts/journal_policy.py to query the Sherpa Romeo v2 API (https://v2.sherpa.ac.uk/cgi/retrieve, item-type=publication, requires a free api-key). Report the journal's prearchiving (preprint) permission, any conditions (embargo, version allowed, required statement), and link the source. If no key is available, fall back to WebFetch on the publisher's policy page and say so. Details: references/journal_policy.md.

Most major publishers permit preprints, but conditions vary (some bar posting the accepted version, some require a DOI link or a specific notice). Never assert a policy from memory — verify it per journal.

3. Assemble submission metadata

Build the metadata block the server needs: title, all authors with ORCID and affiliations, abstract, arXiv primary category + optional cross-lists (or bioRxiv/medRxiv subject collection), declarations (competing interests, funding, data/code availability, ethics/IRB for medRxiv), and the manuscript PDF. The canonical field-by-field checklist is in references/submission_metadata.md. To look up an existing arXiv/bioRxiv record's metadata for reuse, defer to alterlab-arxiv / alterlab-biorxiv rather than re-querying here.

4. Choose the license deliberately

Present the actual license set for the chosen server (see the matrix), then explain the commitment:

• CC BY 4.0 — maximum reuse; irrevocable; some closed-access journals dislike a permissive preprint and may ask you to change it (you cannot revoke CC BY on already-posted versions).
• arXiv non-exclusive license 1.0 — arXiv-specific; you keep copyright, grant arXiv a distribution license; the least journal-friction option on arXiv.
• CC BY-NC-*/-ND — narrower reuse; check it against any funder open-access mandate (e.g. cOAlition S Plan S generally requires CC BY).
• CC0 — public-domain dedication; broadest, also irrevocable.

Confirm whether a funder mandate forces a specific license before recommending. Full table + funder-mandate notes: references/licensing.md.

5. Post, then manage versions

After posting: record the arXiv ID / preprint DOI and the version. Revisions are replacements (arXiv) or new versions (bioRxiv/medRxiv) — the old version stays public and immutable. Do not advise "deleting" an announced arXiv paper; that is impossible — only a withdrawal-version with a tombstone.

6. Link the published article after acceptance

When the paper is published, link the DOI back to the preprint (the server's "published in" / publisher field; bioRxiv/medRxiv auto-detect many links and expose them via api.biorxiv.org /pubs/{server}/...). This makes the version of record discoverable from the preprint and vice versa.

Scripts

• scripts/journal_policy.py — query the Sherpa Romeo v2 API for a journal's preprint/self-archiving policy by ISSN or title (needs a free --api-key; prints a structured summary; degrades to a manual-check instruction offline).
• scripts/server_recommender.py — given a few flags (field, has-clinical-data, target-journal-known, needs-DOI), prints a recommended server + license shortlist with the trade-offs, from the rules in references/server_selection.md.
• scripts/preprint_link_check.py — query the keyless api.biorxiv.org /details/ and /pubs/ endpoints to confirm a bioRxiv/medRxiv DOI exists and surface any detected preprint→published-article link.

All scripts are stdlib-first (use requests if present, else urllib), run in a bare uv run python, and never require a key except the Sherpa Romeo lookup.

Self-Check Before Reporting

• Did I verify the journal's preprint policy (Sherpa Romeo or the live publisher page), or did I assert it from memory? Only the former is allowed.
• Did I name the server's real license options and state the irrevocability / journal-friction trade-off, not a bare "use CC BY"?
• Did I route a search/fetch request to alterlab-arxiv/alterlab-biorxiv, and a data-repository/DMP request to alterlab-open-science, and a TÜBİTAK request to alterlab-aperta, instead of handling it here?
• Did I avoid telling the user to "delete" an already-announced arXiv preprint?

References

• references/server_selection.md — full server decision tree, field-by-field.
• references/submission_metadata.md — per-server metadata field checklist.
• references/licensing.md — license option sets, irrevocability, funder mandates.
• references/journal_policy.md — Sherpa Romeo v2 API usage and policy reading.

Part of the AlterLab Academic Skills suite.