Docs / API
Read-only JSON API
Base URL https://pubphys.com/api/v1. Public, no key, CORS enabled. The API is read-only;
contributions are made on the site, and an AI agent can prepare drafts for a member over
MCP. Formulas in text fields are KaTeX between $ signs.
Endpoints
- GET
/fields: the fields with topic and problem counts. - GET
/topics?field=&sort=&page=&per=: topics, optionally of one field. - GET
/topics/:slug?sort=: a topic with its problems. - GET
/problems?q=&field=&topic=&literature_status=&sort=&page=&per=: search and filter problems (perup to 100). - GET
/problems/:slug: one problem in full: plain and precise statements, what would settle it, kind and answer type, literature status and note, parents, children, links and the topic's review. - GET
/stats: counts.
The API has no reading mode: a problem always carries both its plain and its precise statement, whichever of Expert and Non-expert a browser has chosen on the site.
Order
Lists come in the order of their addresses (slugs), or with sort=slug. With sort=top they come
in order of importance, as the home page lists them: by points, the editors' starting
number plus members' votes (how ranking works); after
the best matches of a search word q. Until ranked lists are enabled, sort=top gives the slug
order. Any other value is refused with status 400. Once enabled, each problem and topic also carries points
(on the scale of the starting numbers, with two decimals; votes can take it past 100 or below 0), votes_up and votes_down; who voted is never given.
Example
curl "https://pubphys.com/api/v1/problems/cm-cuprates-pairing-mechanism"
literature_status describes the published literature; status and
status_label describe the problem on PubPhys and change when claims are accepted.