Reference Asset
Filter, sort, and share views of 17,360 papers across 30 fields.
| Paper | Date | Viability | Code | |
|---|---|---|---|---|
A fixed daily pipeline ingests new arXiv papers, dedupes near-duplicate uploads, and ranks every survivor with a composite score (viability, evidence, community, freshness). The output is a published artifact — both this interactive surface and the REST endpoint read from the same one.
The trust strip above this canvas points at the artifact this page is currently bound to; method versions are immutable per artifact, and upgrades ship as new artifacts rather than rewriting history.
Agent Handoff
Canonical ID database | Route /resources/database
REST example
curl https://sciencetostartup.com/api/v1/resources/database?limit=20MCP example
{
"tool": "search_papers",
"arguments": {
"query": "interactive database",
"limit": 10
}
}source_context
{
"surface": "database",
"mode": "resource",
"query": "interactive database",
"normalized_query": "database",
"route": "/resources/database",
"paper_ref": null,
"topic_slug": null,
"benchmark_ref": null,
"dataset_ref": "database"
}The same data is reachable over REST and MCP. All parameters are optional; the default response is JSON — pass format=csv for a flat export.
GET /api/v1/resources/database
?q={query}
&field={cluster_label} // repeatable
&min_viability={0..10}
&max_viability={0..10}
&has_code={true|false}
&since={ISO-8601-date}
&sort={published_date|viability_score|title|cluster_label}
&dir={asc|desc}
&page={n}
&limit={n}
&format={json|csv}
// topic — legacy fuzzy cluster-label search{
"data": [
{
"arxiv_id": "2403.12345",
"title": "Paper title",
"published_date": "2026-04-12",
"viability_score": 8.4,
"cluster_label": "Robotics",
"repo_url": "https://github.com/..."
}
],
"meta": {
"total": 17360,
"page": 1,
"limit": 20,
"applied_filters": {
"field": [
"Robotics"
],
"min_viability": 7,
"has_code": true,
"sort": "published_date",
"dir": "desc"
},
"facets": {
"field": [
{
"label": "Robotics",
"count": 412
},
{
"label": "Reinforcement Learning",
"count": 318
}
],
"has_code_count": 5234
}
}
}Use This Via API or MCP
Filter, sort, and persist paper selections through the REST endpoint or the MCP tool, then move them downstream into proof review, workspaces, or launch-pack execution.