LighterAPI reference

LighterAPI answers the read requests of api.crossref.org, with the same paths, parameters, and responses, over Crossref's March 2026 public data file.

Base URL

https://demo-crossref-lighterapi.knowledgefutures.org

Replace https://api.crossref.org with this URL in an existing request. No key or account. HTTPS only.

Quick start

# one work
curl -s https://demo-crossref-lighterapi.knowledgefutures.org/works/10.1038/nature12373

# a member's journal articles from 2024, counted by year
curl -s 'https://demo-crossref-lighterapi.knowledgefutures.org/works?filter=member:78,type:journal-article,from-pub-date:2024&facet=published:*&rows=0'

# every work indexed on one day, 1,000 at a time
curl -s 'https://demo-crossref-lighterapi.knowledgefutures.org/works?filter=from-index-date:2026-03-01,until-index-date:2026-03-01&cursor=*&rows=1000'

The console builds and sends requests from the browser.

Limits and access

Limit	Value
Requests per visitor	50 a second, bursts to 400. Over that: `429` with `Retry-After: 1`.
`rows`	0 to 1,000. Default 20.
`offset`	0 to 10,000. Use a cursor beyond.
`sample`	Up to 100.
Facet values	Up to 1,000 per facet (`*` means 1,000). Default 20.
Methods	`GET` and `HEAD`.
Compression	gzip or zstd when the request sends `Accept-Encoding`.

No mailto or polite pool: all visitors get the same limit. The parameter is accepted and ignored.

Routes

Route	Returns
`/works/{doi}`	One work. DOI case-insensitive; `https://doi.org/` prefix accepted. `select` applies.	Served
`/works/{doi}/agency`	The registration agency: always Crossref.	Served
`/works`	A list of works, with every list parameter below.	Served
`/members/{id}/works`	`/works` with `filter=member:{id}`.	Served
`/prefixes/{prefix}/works`	`/works` with `filter=prefix:{prefix}`.	Served
`/funders/{id}/works`	`/works` with `filter=funder:{id}`.	Served
`/journals/{issn}/works`	`/works` with `filter=issn:{issn}`.	Served
`/types/{id}/works`	`/works` with `filter=type:{id}`.	Served
`/members`, `/journals`, `/funders`, `/prefixes/{p}`, `/types`, `/licenses`	Metadata about members, journals, and funders.	Not yet

Parameters

Parameter	Meaning
`filter`	`name:value` pairs, comma-separated. Repeated names are ORed; different names are ANDed. A value may contain commas when the next piece has no colon.	Served
`facet`	`name:count`, comma-separated. Counts are exact.	Served
`sort`, `order`	Sort field; `asc` or `desc` (default `desc`). Works without the field sort last either way.	Served
`rows`, `offset`	Page size and start.	Served
`cursor`	`*` for the first page, then each response's `next-cursor`.	Served
`sample`	Random works from the result.	Served
`select`	Fields to return, comma-separated.	Served
`query`, `query.*`	Text search. Returns `501`.	Not yet
`mailto`	Accepted, ignored.

Filters

Value formats follow api.crossref.org. Identifiers are normalised the way Crossref normalises them.

Filter	Value
`member`	Member ID: `78`, or `http://id.crossref.org/member/78`
`prefix`	DOI prefix: `10.1016`
`type`, `type-name`	Type ID (`journal-article`) or label (`Journal Article`)
`issn`, `isbn`	With or without hyphens
`orcid`	Bare iD or `https://orcid.org/…`
`funder`, `award.funder`	Funder Registry ID (`100000001`) or DOI (`10.13039/100000001`)
`award.number`	Award number, case-insensitive
`doi`	DOI
`container-title`	Exact title
`license.url`, `license.version`	License URL; `vor`, `am`, `tdm`, `stm-asf`, …
`full-text.type`, `.application`, `.version`	Link content type, intended application, version
`ror-id`	ROR ID or URL, from affiliations and institutions
`updates`, `update-type`	DOI updated by the work; update type (`retraction`, `correction`, …)
`relation.type`, `relation.object`	Relation type (`is-preprint-of`, …); related identifier
`archive`, `article-number`, `group-title`, `alternative-id`, `assertion`, `assertion-group`, `clinical-trial-number`, `content-domain`	Exact values
`category-name`	Accepted; matches nothing, as on api.crossref.org

Not yet: license.delay, has-alias, has-prime-doi, from-/until-awarded-date, funder-doi-asserted-by, relation.object-type, relation.claimed-by, lte-/gte-award-amount, lte-/gte-citation, citation-id. They return 400 filter-not-available.

Date filters

from-X and until-X, with a value of YYYY, YYYY-MM, or YYYY-MM-DD. until includes the whole period named: until-pub-date:2020 runs to 31 December 2020.

X	Field
`pub-date`	Earliest of posted, print, online, and other publication dates
`print-pub-date`, `online-pub-date`	Print or online publication date
`issued-date`	Issued date
`created-date`	First deposit
`deposit-date`, `update-date`	Last deposit
`index-date`	Last indexed
`posted-date`, `accepted-date`, `approved-date`, `event-start-date`, `event-end-date`	Those dates

Existence filters

true keeps works that have the field, false keeps works without it (t, 1, f, 0 also accepted).

has-abstract
has-affiliation
has-affiliation-ror-id
has-archive
has-assertion
has-authenticated-orcid
has-award
has-clinical-trial-number
has-content-domain
has-description
has-domain-restriction
has-event
has-full-text
has-funder
has-funder-doi
has-funder-ror-id
has-license
has-orcid
has-references
has-relation
has-ror-id
has-update
has-update-policy
is-update

type-name
published
publisher-name
container-title
issn
funder-name
funder-doi
orcid
license
link-application
archive
update-type
relation-type
assertion
assertion-group
journal-volume
journal-issue
category-name
source
ror-id

Values sort by count, then key, both descending. value-count is the number of values returned. Not yet: affiliation.

Sorts

published
published-print
published-online
issued
created
deposited
updated
indexed
is-referenced-by-count
references-count
score
relevance

score and relevance are constant without a text query. Not yet: citation.

Cursors

# first page
GET /works?filter=member:78&cursor=*&rows=1000
# next page: pass message.next-cursor back
GET /works?filter=member:78&cursor=djF8ZnVsbHwyNTV8MHw0MTc3OTQ&rows=1000

Without a sort, cursors run in creation order, ascending, as on api.crossref.org. With a sort, they follow it. Stop when items is empty. Cursors don't expire, but belong to one index build: after a rebuild, start again from *.

Select

Any top-level field of a work: DOI, title, author, container-title, published, type, member, funder, license, reference, is-referenced-by-count, and the rest of api.crossref.org's list. Unknown names return 400 select-field-not-available.

Envelope

{
  "status": "ok",
  "message-type": "work-list",
  "message-version": "1.0.0",
  "message": {
    "facets": { "type-name": { "value-count": 2, "values": { "Journal Article": 812, "Chapter": 40 } } },
    "total-results": 852,
    "items": [ { "DOI": "10.1016/…", "title": ["…"], … } ],
    "items-per-page": 20,
    "query": { "start-index": 0, "search-terms": null },
    "next-cursor": "…"
  }
}

Single works: "message-type": "work", with the work as message.

Errors

Status	When	Body
`400`	Unknown filter, facet, sort, or select field; bad date; rows, offset, or sample out of range; bad cursor	`{"status":"failed","message-type":"validation-failure","message":[{"type","value","message"}]}`
`404`	Unknown DOI or route	`Resource not found.`
`429`	Over the per-visitor limit	JSON, with `Retry-After: 1`
`501`	Text queries	JSON, `"message-type":"not-implemented"`

Headers

`Server-Timing: app;dur=…`	Milliseconds spent in the server for this request.
`Content-Encoding`	`gzip` or `zstd` when requested and the body is over 1 KB.
`Access-Control-Allow-Origin: *`	Callable from any web page.

Differences from api.crossref.org

Counts	`total-results` and facet counts are exact.
Order	Without a sort or cursor, items come in creation order (unspecified on api.crossref.org).
Ties	Works created in the same second keep the builder's order, not DOI order.
Freshness	A snapshot: works registered or changed after March 2026 aren't here.
Text search	Not yet.

Data

Crossref public data file, March 2026 (10.13003/nggf-vt1j): 179,536,204 works. Records are served as the file has them, with the empty fields api.crossref.org adds to single works (title, container-title, subtitle, original-title, short-title, short-container-title, subject, relation).

Status endpoints

`/healthz`	`ok` when serving.
`/_info`	Index build: works, build time, source.
`/_stats`	Requests and latency by kind since the last reset; cache hit rates.

Built by Knowledge Futures. Not a Crossref service.