Public reading API

7,995 public records include 5,423 reader-facing objects and 2,572 record-only entries. Normal Search and Index browse reader-facing objects; exact identifier Search can retrieve public record-only entries. Held material is not a public reading resource. Metadata publication does not grant image reuse or training rights.

This describes implemented routes in the release candidate, not a promise that remote production has been accepted. Reading these resources does not request suggestions or generate PNG files.

Search

GET /api/search/v1 accepts q, yearFrom, yearTo, objectType, theme, movement, first and after. first is 1–50; after is an opaque cursor returned by the response. Keep the query and filters unchanged when following a cursor. Exact public identifiers can return record-only entries; normal search uses reader-facing objects. GET /api/search/v1/facets takes no parameters.

Index and Object reading

GET /api/index/v1 returns one compact reader-facing catalogue, with schema version, releaseId, count, publicCount, recordOnlyCount, dictionaries and rows. It is not paginated. /surfaces/ID provides the public HTML record; unknown or held IDs return 404. Source attributions and rights remain record-specific.

Versions, errors and limits

Search includes release and index identities in its response and Archive response headers. Invalid search input or cursor returns 400; unsupported methods return 405; oversized URLs return 414; unavailable Search returns 503. Do not retry invalid input. The public reading routes currently do not implement a per-requester quota; infrastructure limits may still apply. The separate 30-per-minute suggestions policy is not an archive API or daily spending limit. Use bounded requests and cache versioned results.

OpenAPI description · Source and rights · Project implementation