Sugra Research: preprints, scholarly metadata, and long-run development data

Series: Sugra product directions (7 of 7)
Audience: developers integrating the Sugra API over HTTP
Live samples verified: 2026-07-24 against https://sugra.ai
Auth: x-api-key header on every data call

Standalone article. Sugra Research covers the scholarly record and long-run development statistics - from this morning’s preprints to multi-century growth panels.

Essence

Scholarly and development data - from this morning’s preprints to a century of growth statistics.

Core surfaces:

arXiv preprint search, paper fetch, categories
Crossref DOI works, journals, funders, references, citations
NBER working papers and (when available) recession dates
Maddison Project long-run GDP and population
Penn World Table growth and productivity panel
Adjacent development slices: WHO GHO, OWID composites, UN/ILO/FAO series (often under neighboring reference categories but one key)

All academic or intergovernmental sources, named in meta.source (for example arxiv, crossref, maddison, pwt, nber).

What this direction is not

Not this	Why
Full-text paper hosting	Metadata, links, and abstracts - not a pirate library
Citation manager product	Graph walks via API, you store the library
Official high-frequency macro prints	Sugra Macro (FRED, BIS, Treasury)
News about research	Sugra News
Investment advice from papers	Literature access only

Capability map

Category	Endpoints (approx.)	Coverage
Research core	34	arXiv (7), Crossref (10), NBER (2), Maddison (7), PWT (8) under `/api/v1/research/*`
Health	10	WHO Global Health Observatory
Adjacent development	varies	OWID, UN Population, ILOSTAT, SDG, FAOSTAT live in related categories with the same key

Why the response shape is useful

1. Fielded arXiv search with full abstract and URLs

Live search q=agentic+reasoning (limit 2): 680 total hits; papers include arxiv_id, title, full summary, authors, categories, published timestamps, urls.pdf / urls.abstract.

{
  "data": {
    "query": "all:\"agentic reasoning\"",
    "papers": [
      {
        "arxiv_id": "2601.12538",
        "title": "Agentic Reasoning for Large Language Models",
        "primary_category": "cs.AI",
        "published": "2026-01-18T18:58:23Z",
        "urls": {
          "abstract": "https://arxiv.org/abs/2601.12538v1",
          "pdf": "https://arxiv.org/pdf/2601.12538v1"
        }
      }
    ],
    "pagination": { "total_results": 680, "limit": 2 }
  },
  "meta": { "source": "arxiv" }
}

2. Crossref DOI resolution with citation graph hooks

Live DOI 10.1038/s41586-021-03819-2 (AlphaFold paper): title, venue Nature, authors, license, is-referenced-by-count (43000+), references list, publisher metadata. Sibling routes: references and citations by DOI.

3. Long-run Maddison panels

Country series with year, GDP per capita, population - centuries of observations for many countries. Methodology and citation live under definitions endpoints.

4. PWT rankings and TFP panels

Rank countries by real GDP or productivity variables with attribution strings suitable for academic footnotes.

5. NBER latest papers

/api/v1/research/nber/papers/latest - live 200. Recession dates endpoint can 503 when static fetch fails - handle in clients.

Developer path

export SUGRA_API_KEY="YOUR_API_KEY"

curl -H "x-api-key: $SUGRA_API_KEY" \
  "https://sugra.ai/api/v1/research/arxiv/search?q=agentic+reasoning&limit=5"

curl -H "x-api-key: $SUGRA_API_KEY" \
  "https://sugra.ai/api/v1/research/crossref/works/10.1038/s41586-021-03819-2"

curl -H "x-api-key: $SUGRA_API_KEY" \
  "https://sugra.ai/api/v1/research/maddison/country/JPN"

curl -H "x-api-key: $SUGRA_API_KEY" \
  "https://sugra.ai/api/v1/research/nber/papers/latest?limit=5"

import os, requests
H = {"x-api-key": os.environ["SUGRA_API_KEY"]}
r = requests.get(
    "https://sugra.ai/api/v1/research/arxiv/search",
    headers=H, params={"q": "climate", "limit": 10}, timeout=30,
)
r.raise_for_status()
papers = r.json()["data"]["papers"]

Use case 1: Morning preprint scan

Goal. What appeared on arXiv for a topic or category.

Calls. /research/arxiv/search, /research/arxiv/latest?cat=, /research/arxiv/paper/{id}.

Strength. Structured papers + pdf links + pagination.
Pitfalls. Query syntax is arXiv-style; relevance sort is not peer review.

Use case 2: DOI resolve and cite

Goal. Turn a DOI into authoritative metadata for an agent or UI.

Call. /research/crossref/works/{doi}.

Strength. Venue, authors, licenses, citation counts.
Pitfalls. Huge payloads - select fields client-side for LLM context windows.

Use case 3: Citation graph walk

Goal. What a paper cites / who cites it.

Calls. Crossref references and citations routes for a DOI.

Strength. Enough for a lightweight graph without a full bibliometrics platform.
Pitfalls. Coverage depends on Crossref deposits.

Use case 4: Long-run GDP context

Goal. Place a country in multi-century GDP/pop history.

Call. /research/maddison/country/{iso3} (+ world/regional/top).

Strength. Teaching and research panels with methodology endpoints.
Pitfalls. Historical estimates carry large uncertainty - show source attribution.

Use case 5: Productivity and PWT rankings

Goal. Rank countries by real output or TFP for a year.

Call. /research/pwt/top?variable=rgdpo&year=2019&limit=10.

Strength. Standard growth-accounting panel with attribution.
Pitfalls. Variable codes require the definitions endpoint.

Use case 6: NBER working paper feed

Goal. Latest economics working papers for a research agent.

Call. /research/nber/papers/latest.

Strength. RSS-backed latest set with abstracts.
Pitfalls. pub_date often null - order by paper id when recency matters.

Use case 7: Health indicator scorecard (WHO)

Goal. Country health indicators and rankings.

Calls. /api/v1/health/who/* family (catalog, country data, latest, rankings).

Strength. Named WHO GHO series under the same key.
Pitfalls. Health indicators lag; document as_of.

Use case 8: Literature-grounded agent answer

Goal. Agent must not invent papers - search arXiv/Crossref, return titles + DOI/arXiv ids + meta.source.

Pattern.

Search arXiv or resolve DOI
Quote title, year, venue
Attach pdf/doi link
Cite meta.source

Strength. Primary Research product for AI agent builders.
Pitfalls. Always distinguish preprint from peer-reviewed venue.

Antipatterns

Dumping entire Crossref JSON into an LLM prompt
Presenting preprints as settled fact
Omitting Maddison/PWT attribution
Confusing Research long-run GDP with Macro FRED monthly CPI
Hard-failing the product when NBER recession-dates returns 503 - degrade gracefully

Scenarios where Research alone is enough

Academic search agents
DOI metadata services
Long-run development teaching tools
NBER watch feeds
WHO-based health scorecards

For “today’s CPI print” use Macro. For “stock reaction to a paper” that is a multi-direction composition (later series articles).

Live verification (2026-07-24)

Call	Result
arXiv search agentic reasoning	200, 680 hits
Crossref AlphaFold DOI	200
Maddison JPN	200
NBER papers/latest	200
NBER recession-dates	503 (handle as optional)

Where to go next

Portal: Sugra Research direction + sources
Series complete for standalone directions
Next phase (later): composition articles that combine directions (for example Macro + Finance, Entity + News)
Cookbook and portal recipes for multi-call patterns

#	Direction	Status
1	Finance	Published
2	Macro	Published
3	Entity	Published
4	NetAtlas	Published
5	News	Published
6	Earth	Published
7	Research	This article