The API
One JSON door. Reading needs no key; writing needs the registry key or a signed-in registrant.
Read (no key)
| Call | Gives |
|---|---|
GET https://registry.smartscholars.in/api.php?action=get&doi=… | the record: state, address, version, holder, the record itself, its sha256 |
GET https://registry.smartscholars.in/api.php?action=history&doi=… | the ledger of the DOI, oldest first, each entry with who, what changed, and the full record after it |
GET https://registry.smartscholars.in/api.php?action=resolve&doi=… | where it goes, without following — and, since 9 Oct 2026, fallback: null, or (when the address is known to be dead and an archived copy is held) the copy the resolver offers instead, with its archive, date and SHA-256 |
GET https://registry.smartscholars.in/api.php?action=list&prefix=&state=&q=&limit=&offset= | the names held, page by page (limit ≤ 500) |
GET https://registry.smartscholars.in/api.php?action=prefixes · ?action=counts | the prefixes and their holders · the figures on the front page |
GET https://registry.smartscholars.in/api.php?action=search&q=&prefix=&year=&type=&state=&sort=&page=&per_page= | search across every record: a DOI name finds that name, other words must all occur in the record (titles, people, journal, ISSN, volume/issue/pages, year, type); sort newest · oldest · title · year; per_page ≤ 100; the answer carries the total, the page count and year/type facets |
… &cursor=*&rows=100 on search and list | deep paging, as Crossref's REST API does it: cursor=* starts, each answer carries next-cursor for the next page (rows ≤ 1000), and no cursor at the end; a cursor is tied to its query and sort and lasts 24 hours — how it works |
… &from-pub-date=2025-01&until-pub-date=2025-12, or &filter=from-update-date:2026-10-01 | date ranges, with Crossref's filter names: from-pub-date / until-pub-date (the record's publication date), from-update-date / until-update-date (the last change here), from-created-date / until-created-date (the registration here); YYYY, YYYY-MM or YYYY-MM-DD, inclusive, UTC |
GET https://registry.smartscholars.in/api.php?action=get&doi=…&as=turtle|rdfxml|jsonld|unixref|unixsd|datacite | the record in one of the DOI content-negotiation service's formats — RDF (Turtle, RDF/XML), schema.org JSON-LD, Crossref UNIXREF and UNIXSD XML, DataCite XML 4.7 — or as=xml|system|csl|bibtex|ris|apa; a format whose schema requires what the record does not hold is answered 406 with the reason and the formats that can be made (the formats) |
GET https://registry.smartscholars.in/api.php?action=cite&doi=…&as=csl|bibtex|ris|apa | one citation as text (with its MIME type), or — without as — all four in one JSON answer |
GET https://registry.smartscholars.in/api.php?action=submissions®istrant=N&kind=&from=&to=&doi=&page=&per_page= | the submission log — a registrant's ledger rows, newest first, each with who, what changed and the checksum after (6 Oct 2026); for the registrant's own members (signed in) or the registry key (X-Registry-Token); the same on a page: submissions.php?registrant=N |
The open doors beside the API: feed.php — Atom (or ?as=rss) of every change, ?prefix= for one prefix, ?kind=registered for new registrations only · bulk.php — the whole registry or one prefix as JSON Lines (?as=jsonl) or CSV (?as=csv), &since= for a mirror that keeps itself current · cite.php — the four citation forms on one page.
Harvesting: oai.php — OAI-PMH 2.0, for repositories and aggregators: every registered and withdrawn DOI as Dublin Core (metadataPrefix=oai_dc) or as its Kernel Metadata Declaration (doi_kernel), sets by prefix and by registrant, from/until in UTC, 100 a page with a resumption token; a withdrawn DOI is a deleted record. Identifiers are oai:registry.smartscholars.in: and the DOI.
The resolver does the same by content negotiation on https://registry.smartscholars.in/resolve.php?doi=…: Accept: application/json gives the record, application/xml the Kernel Metadata Declaration, application/vnd.citationstyles.csl+json CSL-JSON, application/x-bibtex BibTeX, application/x-research-info-systems RIS, text/x-bibliography an APA reference — as doi.org does for every DOI; and, since 6 Oct 2026, text/turtle, application/rdf+xml, application/ld+json (schema.org), application/vnd.crossref.unixref+xml, application/vnd.crossref.unixsd+xml and application/vnd.datacite.datacite+xml. The XML bibliographic query (Crossref's XML API form) is query.php.
Citations, relations, data citations, funders (no key)
Read from the records held in this registry — and only from them: cited-by counts only records held here; a citation from a work held elsewhere is not seen. Each answer is paged (page, per_page ≤ 100) and carries total and pages.
| Call | Gives |
|---|---|
GET https://registry.smartscholars.in/api.php?action=cited_by&doi=… | the registered records held here whose references cite the DOI — the DOI itself may be held here or anywhere else |
GET https://registry.smartscholars.in/api.php?action=references&doi=… | a held record's references that carry a DOI, in the record's order, each with the record held here under that DOI (if any) and how many references the record lists in all |
GET https://registry.smartscholars.in/api.php?action=relations&doi=… | both directions: what the DOI's own record states ("direction": "from") and what other records held here state about it ("to") — Crossref's name for the relation, its Kernel role, the other item's identifier and type |
GET https://registry.smartscholars.in/api.php?action=data_citations&doi=&subject=&object= | links from records held here to datasets: a reference or relation to a DOI held here as a dataset, or a relation that says data by its own name (Crossref's basedOnData; isDataBasisFor read from the other end) — each pair with every statement behind it |
GET https://registry.smartscholars.in/api.php?action=funders&name=&ror=&funder_id= | funders with the number of registered works naming each — by ROR ID, by Funder Registry ID, or by name alone; filter by a piece of the name, a ROR ID or a Funder Registry ID |
GET https://registry.smartscholars.in/api.php?action=search&funder=… | the search, narrowed to works whose funders carry that ROR ID or Funder Registry ID (with state=registered it counts what the funder's page counts) |
The same on pages: every record page shows its cited-by, references, relations and funding; funder.php lists the funders and each one's works. How the indexes are made, and rebuilt from the bulk file: the documentation.
Keys and limits (reading)
Reading needs no key: without one, the reads are counted per internet address — 300 a minute and 10,000 a day. A free key (600 a minute and 100,000 a day) or a Plus key (1,200 a minute and 1,000,000 a day, and the monthly snapshot) goes in a header: X-Api-Key: ssr_… or Authorization: Bearer ssr_…. Every counted answer carries X-Api-Level and X-RateLimit-Limit / -Remaining / -Reset; over a limit the answer is 429 with Retry-After. The details · the two levels and how to get a key.
Write (the key, or a signed-in registrant)
POST a JSON body with action in the address, the header X-Registry-Token carrying the key; a signed-in member of a registrant needs no key for its own prefixes.
| action | body | does |
|---|---|---|
| reserve | {"doi": "10.80000/my-name"} | takes a name the registrant chose (checked against the naming policy); no record yet |
| mint | {"prefix": "10.46243", "parts": {"year": 2026, "vol": 11, "issue": 1, "first": 1, "last": 9}} | a name from the prefix's suffix rule (its pattern, or opaque), reserved |
| register | {"doi": …, "url": …, "record": {…}} | the record (the format) and the address — refused in words when any rule is not met |
| update | {"doi": …, "record": {…}} | a new version; an identical record changes nothing |
| url | {"doi": …, "url": …} | the address alone |
| withdraw · restore | {"doi": …, "reason": …} · {"doi": …} | the tombstone, and the way back |
| import · import_prefix | {"doi": …} · {"prefix": …, "rows": 100} | read from Crossref into the registry (a prefix, page by page — the key or an admin) |
| locations | {"doi": …, "locations": [{"url": "https://…", "label": "Hindi edition", "country": "IN", "language": "hi", "weight": 0}]} | the DOI's other addresses beside its main one — at most 10, the whole set replaced, [] removes them; country ISO 3166-1, language BCP 47, weight 0–1 (how the resolver chooses); get and resolve give them as locations |
| archive_copies | {"doi": …, "archive_copies": [{"archive": "Internet Archive", "url": "https://web.archive.org/web/…", "sha256": "…64 hex…", "timestamped_at": "2026-10-02T10:15:00Z", "verified_at": "2026-10-08"}]} | the copies of the work an archive keeps, each with the SHA-256 of the file as archived — the whole set replaced, [] removes them; a new version of the record (format 1.1), one row in its history. When the DOI's address is known to be dead (two checks in a row), the resolver offers the newest checked copy and resolve says so as fallback (format 1.1) |
Answers: {"ok": true, …}, or {"ok": false, "error": "…", "errors": [ … ]} with 400 (not sound), 401 (no key), 403 (not your prefix), 404 (not held), 409 (already registered / taken).



