BUILD SOMETHING OPEN

The same collection.
A public interface.

Read endpoints require no authentication. Responses carry citations, source licenses, and structured errors. Limit: 60 requests per minute, 600 verification or ask requests per day.

Download the OpenAPI specification →

MethodEndpointPurpose
GET/v1/traditionsCollection coverage and reviewer status
GET/v1/worksPublished works; optional tradition filter
GET/v1/works/bhagavad-gita/divisionsChapter counts from the database
GET/v1/works/quran/chapters/2?page=2Paginated section text: 80 passages per page
POST/v1/study-translationEligible passage_id and target_lang; unreviewed AI study aid
GET/v1/passages/bhagavad-gita/BG-2.47Primary text, translations, and full license records
GET/v1/passages/bhagavad-gita/BG-2.47/contextFive passages before and after
GET/v1/search?q=karmaExact and fuzzy search; mode, work, tradition, lang, limit
POST/v1/verifyJSON: claim, claimed_source?, lang
POST/v1/reportsJSON: passage_id, kind, message
GET/v1/sourcesSource and data-quality records
GET/v1/correctionsPublic reviewer-approved corrections
GET/v1/healthDatabase state and actual capabilities

Keep the attribution attached

Reusers must preserve the source, citation, and applicable license. See /v1/sources before redistributing a text. Meaning search and cited Ask depend on configured providers. Browser dictation requires transcript review. Optional AI study translations are labelled unreviewed, kept separate from published editions and disabled for restricted sources. No expert pronunciation assessment is offered.