Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Resolve dataset DOIs and query research data metadata via DataCite
.claude/skills/brycewang-stanford-datacite-api/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | 87% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 54% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 14% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 114% | 0% |
| case-15 | ✗→✓ | ▲ Improved | 50% | 0% |
DataCite is a leading global DOI registration agency focused on research data. While CrossRef primarily handles DOIs for publications, DataCite specializes in assigning persistent identifiers to datasets, software, samples, instruments, and other research outputs. DataCite has registered over 50 million DOIs from thousands of data repositories worldwide.
The DataCite REST API provides access to metadata for all DataCite DOIs. It is essential for researchers and developers working with research data discovery, data citation, FAIR (Findable, Accessible, Interoperable, Reusable) data practices, and repository integration. The metadata follows the DataCite Metadata Schema, which is designed specifically for describing research data and includes fields for resource types, funding references, geolocation, and related identifiers.
The API is free, open, and requires no authentication. It returns JSON responses following the JSON:API specification, with robust filtering, faceting, and pagination support.
No authentication required. The DataCite API is fully open and free to use. No API key, registration, or email is needed. For write operations (DOI registration and metadata updates), authentication via DataCite member credentials is required, but read-only access is completely open.
GET https://api.datacite.org/dois| Param | Type | Required | Description | |-------|------|----------|-------------| | query | string | No | Full-text search query | | resource-type-id | string | No | Filter by resource type (dataset, software, text, etc.) | | affiliation-id | string | No | Filter by creator affiliation ROR ID | | registered | string | No | Filter by registration year (e.g., 2024) | | pagesize] | integer | No | Results per page (default: 25, max: 1000) | | pagenumber] | integer | No | Page number for pagination | | sort | string | No | Sort field: relevance, created, -created, updated |
bash curl "https://api.datacite.org/dois?query=climate+change+dataset&resource-type-id=dataset&page[size]=10&sort=-created"
data array containing DOI records. Each record has attributes with doi, titles, creators, publisher, publicationYear, resourceType, descriptions, subjects, dates, relatedIdentifiers, fundingReferences, and geoLocations.GET https://api.datacite.org/dois/{doi}| Param | Type | Required | Description | |-------|------|----------|-------------| | doi | string | Yes | The full DOI (e.g., 10.5281/zenodo.1234567) |
bash curl "https://api.datacite.org/dois/10.5281/zenodo.3678171"
GET https://api.datacite.org/providers| Param | Type | Required | Description | |-------|------|----------|-------------| | query | string | No | Search provider name or description | | region | string | No | Filter by region (e.g., EMEA, Americas, Asia Pacific) | | pagesize] | integer | No | Results per page |
bash curl "https://api.datacite.org/providers?query=CERN&page[size]=5"
name, displayName, region, memberType, website, and associated repositories and DOI prefixes.GET https://api.datacite.org/clients| Param | Type | Required | Description | |-------|------|----------|-------------| | query | string | No | Search repository name | | provider-id | string | No | Filter by provider | | software | string | No | Filter by repository software (e.g., dspace, dataverse) | | pagesize] | integer | No | Results per page |
bash curl "https://api.datacite.org/clients?query=zenodo&page[size]=5"
No published rate limits. DataCite does not enforce strict API quotas for read access. However, the service is operated by a nonprofit organization, so users should implement reasonable request pacing. For large-scale data mining, use the DataCite OAI-PMH endpoint or the public data file available at https://datafiles.datacite.org. Sustained high-volume requests may be throttled without notice.
Find published datasets related to a research area:
bashcurl -s "https://api.datacite.org/dois?query=CRISPR+genome+editing&resource-type-id=dataset&page[size]=5&sort=-created" | jq '.data[] | {doi: .attributes.doi, title: .attributes.titles[0].title, year: .attributes.publicationYear, publisher: .attributes.publisher}'
Search for research software registered with DataCite:
bashcurl -s "https://api.datacite.org/dois?query=python+machine+learning&resource-type-id=software&page[size]=10" | jq '.data[] | {doi: .attributes.doi, title: .attributes.titles[0].title, year: .attributes.publicationYear}'
Use related identifiers to find papers associated with a dataset:
bashcurl -s "https://api.datacite.org/dois/10.5281/zenodo.3678171" | jq '.data.attributes.relatedIdentifiers[] | select(.relationType == "IsSupplementTo" or .relationType == "IsReferencedBy") | {type: .relationType, id: .relatedIdentifier}'
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 6,766 | 11,567 | +71% | 1 | 1 | 0% | 1,259 | 2,817 | +124% | 0 | 0 | — |
case-02 | fail→fail | 17,688 | 11,007 | -38% | 1 | 1 | 0% | 2,933 | 3,405 | +16% | 0 | 0 | — |
case-08 | pass→pass | 14,558 | 7,205 | -51% | 1 | 1 | 0% | 2,552 | 2,683 | +5% | 0 | 0 | — |
case-03 | fail→pass | 7,213 | 6,394 | -11% | 1 | 1 | 0% | 1,416 | 2,644 | +87% | 0 | 0 | — |
case-04 | fail→fail | 17,345 | 14,495 | -16% | 1 | 1 | 0% | 2,881 | 4,151 | +44% | 0 | 0 | — |
case-05 | fail→pass | 9,528 | 5,882 | -38% | 1 | 1 | 0% | 1,682 | 2,595 | +54% | 0 | 0 | — |
case-06 | fail→pass | 16,095 | 11,869 | -26% | 1 | 1 | 0% | 3,034 | 3,446 | +14% | 0 | 0 | — |
case-07 | pass→pass | 7,633 | 2,827 | -63% | 1 | 1 | 0% | 1,153 | 1,971 | +71% | 0 | 0 | — |
case-09 | pass→pass | 7,592 | 5,328 | -30% | 1 | 1 | 0% | 1,322 | 2,343 | +77% | 0 | 0 | — |
case-10 | pass→pass | 11,243 | 6,529 | -42% | 1 | 1 | 0% | 1,938 | 2,749 | +42% | 0 | 0 | — |
case-11 | pass→pass | 12,644 | 10,587 | -16% | 1 | 1 | 0% | 2,485 | 3,634 | +46% | 0 | 0 | — |
case-12 | pass→pass | 13,896 | 16,332 | +18% | 1 | 1 | 0% | 2,438 | 4,286 | +76% | 0 | 0 | — |
case-13 | fail→pass | 16,625 | 4,432 | -73% | 1 | 1 | 0% | 1,097 | 2,346 | +114% | 0 | 0 | — |
case-14 | pass→pass | 10,184 | 6,420 | -37% | 1 | 1 | 0% | 1,592 | 2,582 | +62% | 0 | 0 | — |
case-15 | fail→pass | 12,021 | 25,941 | +116% | 1 | 1 | 0% | 1,750 | 2,617 | +50% | 0 | 0 | — |
case-16 | pass→pass | 11,156 | 7,878 | -29% | 1 | 1 | 0% | 1,727 | 3,157 | +83% | 0 | 0 | — |
case-17 | pass→pass | 7,567 | 3,760 | -50% | 1 | 1 | 0% | 1,494 | 2,270 | +52% | 0 | 0 | — |
case-18 | pass→pass | 6,687 | 7,197 | +8% | 1 | 1 | 0% | 1,025 | 2,534 | +147% | 0 | 0 | — |
case-19 | fail→pass | 4,605 | 1,988 | -57% | 1 | 1 | 0% | 731 | 1,841 | +152% | 0 | 0 | — |
case-20 | pass→pass | 7,414 | 5,720 | -23% | 1 | 1 | 0% | 1,374 | 2,575 | +87% | 0 | 0 | — |
case-21 | pass→pass | 10,040 | 8,198 | -18% | 1 | 1 | 0% | 1,581 | 2,904 | +84% | 0 | 0 | — |
case-22 | pass→pass | 9,339 | 8,955 | -4% | 1 | 1 | 0% | 1,976 | 2,937 | +49% | 0 | 0 | — |
case-23 | pass→pass | 4,481 | 2,812 | -37% | 1 | 1 | 0% | 730 | 1,890 | +159% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 23 cases were attempted, and 22 counted toward the lift figure. The other 1 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +26 percentage points is the difference between those two pass rates over the 22 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.