Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Access Latin American and developing world research via SciELO API
.claude/skills/brycewang-stanford-scielo-api/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 38% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -25% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 106% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -2% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 50% | 0% |
SciELO (Scientific Electronic Library Online) is the primary open access platform for scholarly journals in Latin America, the Caribbean, Spain, Portugal, and South Africa. It indexes 1,800+ peer-reviewed journals and 900K+ articles, many not indexed elsewhere. All content is open access. The API provides article search, journal metadata, and bibliometric indicators. Free, no authentication required.
Search and retrieve articles:
bash# Search articles by keyword curl "https://articlemeta.scielo.org/api/v1/article/?collection=scl&q=machine+learning" # Get article by PID (SciELO identifier) curl "https://articlemeta.scielo.org/api/v1/article/?code=S0100-204X2024000100001" # Filter by collection (country) curl "https://articlemeta.scielo.org/api/v1/article/?collection=esp&q=climate+change" # Filter by journal ISSN curl "https://articlemeta.scielo.org/api/v1/article/?issn=0100-204X&from_date=2024-01-01"
| Code | Country/Region | |------|---------------| | scl | Brazil | | esp | Spain | | mex | Mexico | | col | Colombia | | chl | Chile | | arg | Argentina | | cub | Cuba | | ven | Venezuela | | prt | Portugal | | zaf | South Africa |
bash# List journals in a collection curl "https://articlemeta.scielo.org/api/v1/journal/?collection=scl" # Get journal by ISSN curl "https://articlemeta.scielo.org/api/v1/journal/?issn=0100-204X" # Journal indicators curl "https://analytics.scielo.org/api/v1/journal/?issn=0100-204X"
Full-text search with facets:
bash# Full-text search curl "https://search.scielo.org/?q=biodiversity+conservation&format=json&count=20" # Filter by subject area curl "https://search.scielo.org/?q=neural+networks&filter[subject_area]=Computer+Science&format=json" # Filter by year range curl "https://search.scielo.org/?q=CRISPR&filter[year_cluster]=2023-2026&format=json" # Filter by language curl "https://search.scielo.org/?q=epidemiology&filter[la]=en&format=json"
| Parameter | Description | Example | |-----------|-------------|---------| | q | Free-text query | q=tropical+ecology | | collection | Country code | collection=scl | | issn | Journal ISSN | issn=0100-204X | | from_date | Articles from date | from_date=2024-01-01 | | until_date | Articles until date | until_date=2026-12-31 | | format | Response format | json, xml | | count | Results per page | count=50 | | offset | Pagination offset | offset=20 |
pythonimport requests ARTICLE_API = "https://articlemeta.scielo.org/api/v1" SEARCH_API = "https://search.scielo.org" def search_scielo(query: str, collection: str = None, count: int = 20) -> list: """Search SciELO articles.""" params = {"q": query, "format": "json", "count": count} if collection: params["collection"] = collection resp = requests.get(f"{SEARCH_API}/", params=params) resp.raise_for_status() data = resp.json() results = [] for doc in data.get("docs", data.get("results", [])): results.append({ "title": doc.get("title", {}).get("en", doc.get("title", "")), "authors": doc.get("authors", []), "journal": doc.get("journal_title", ""), "year": doc.get("publication_year", ""), "doi": doc.get("doi", ""), "pid": doc.get("pid", ""), "language": doc.get("la", []), "url": f"https://scielo.org/article/{doc.get('pid', '')}", }) return results def get_article(pid: str) -> dict: """Get full article metadata by SciELO PID.""" resp = requests.get( f"{ARTICLE_API}/article/", params={"code": pid, "format": "json"}, ) resp.raise_for_status() return resp.json() def list_journals(collection: str = "scl") -> list: """List journals in a SciELO collection.""" resp = requests.get( f"{ARTICLE_API}/journal/", params={"collection": collection, "format": "json"}, ) resp.raise_for_status() return resp.json() # Example: find Brazilian ecology research papers = search_scielo("Amazon deforestation biodiversity", collection="scl") for p in papers: print(f"[{p['year']}] {p['title']} — {p['journal']}") # Example: find Spanish medical research papers = search_scielo("diabetes treatment", collection="esp") for p in papers: print(f"{p['title']} (DOI: {p['doi']})")
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 11,102 | 30,574 | +175% | 1 | 1 | 0% | 2,127 | 2,936 | +38% | 0 | 0 | — |
case-02 | fail→pass | 33,976 | 15,098 | -56% | 1 | 1 | 0% | 5,559 | 4,144 | -25% | 0 | 0 | — |
case-03 | fail→pass | 21,948 | 6,421 | -71% | 1 | 1 | 0% | 1,450 | 2,980 | +106% | 0 | 0 | — |
case-04 | pass→pass | 5,614 | 3,726 | -34% | 1 | 1 | 0% | 894 | 2,350 | +163% | 0 | 0 | — |
case-05 | pass→pass | 21,892 | 20,162 | -8% | 1 | 1 | 0% | 4,628 | 5,735 | +24% | 0 | 0 | — |
case-06 | pass→pass | 9,401 | 4,167 | -56% | 1 | 1 | 0% | 1,827 | 2,286 | +25% | 0 | 0 | — |
case-07 | fail→pass | 27,003 | 3,133 | -88% | 1 | 1 | 0% | 2,271 | 2,235 | -2% | 0 | 0 | — |
case-08 | pass→pass | 12,550 | 4,553 | -64% | 1 | 1 | 0% | 2,442 | 2,425 | -1% | 0 | 0 | — |
case-09 | fail→pass | 9,298 | 6,353 | -32% | 1 | 1 | 0% | 1,747 | 2,613 | +50% | 0 | 0 | — |
case-10 | fail→pass | 40,832 | 12,220 | -70% | 1 | 1 | 0% | 8,086 | 3,951 | -51% | 0 | 0 | — |
case-11 | fail→pass | 13,743 | 3,646 | -73% | 1 | 1 | 0% | 2,343 | 2,347 | +0% | 0 | 0 | — |
case-12 | fail→pass | 10,785 | 5,504 | -49% | 1 | 1 | 0% | 1,860 | 2,654 | +43% | 0 | 0 | — |
case-13 | fail→pass | 16,003 | 4,088 | -74% | 1 | 1 | 0% | 2,410 | 2,358 | -2% | 0 | 0 | — |
case-14 | pass→pass | 12,953 | 3,704 | -71% | 1 | 1 | 0% | 2,175 | 2,147 | -1% | 0 | 0 | — |
case-15 | fail→pass | 9,955 | 4,427 | -56% | 1 | 1 | 0% | 1,490 | 2,286 | +53% | 0 | 0 | — |
case-16 | fail→pass | 9,387 | 2,849 | -70% | 1 | 1 | 0% | 1,499 | 2,207 | +47% | 0 | 0 | — |
case-17 | fail→pass | 10,312 | 2,708 | -74% | 1 | 1 | 0% | 1,937 | 1,983 | +2% | 0 | 0 | — |
case-18 | pass→pass | 10,279 | 2,798 | -73% | 1 | 1 | 0% | 1,498 | 1,962 | +31% | 0 | 0 | — |
case-19 | fail→pass | 11,956 | 2,738 | -77% | 1 | 1 | 0% | 1,719 | 2,056 | +20% | 0 | 0 | — |
case-20 | fail→pass | 17,370 | 3,860 | -78% | 1 | 1 | 0% | 2,967 | 2,253 | -24% | 0 | 0 | — |
case-21 | fail→pass | 14,124 | 3,115 | -78% | 1 | 1 | 0% | 2,054 | 2,130 | +4% | 0 | 0 | — |
case-22 | fail→pass | 8,117 | 4,099 | -50% | 1 | 1 | 0% | 1,587 | 2,440 | +54% | 0 | 0 | — |
case-23 | fail→pass | 11,642 | 6,571 | -44% | 1 | 1 | 0% | 1,843 | 2,634 | +43% | 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 +74 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.