Development of API v3 continues on dev with new integration workflows. V2 remains the stable integration default.
- Cursor traversal: opt into authenticated, credential/query-bound cursors with stable ordering, forward/backward URLs and 24-hour expiry. Existing page pagination stays the default.
- Inventory filters: filter by creation origin, selected IDs, creation/update dates, password protection, health checking and user tracking; choose stable sort order.
- Scoped summaries and tags: inspect inventory status counts and discover tags without exporting every link. Large tag aggregations require narrower filters.
- Address lookup: retrieve an authorized link by hostname and exact slug without storing its ID separately.
- Health status: read stored status, HTTP code, response time and check timestamp without triggering an outbound probe or exposing diagnostics.
- Target collections: list and replace targeting rules directly, preserving IDs and analytics for retained targets.
- Bulk updates: PATCH up to 50 links with independent outcomes, normal safety checks, transactions, audit and cache invalidation.
- Idempotent creation: optional
Idempotency-Keysupport for single and bulk link creation, with database reservations, encrypted 24-hour response retention and explicit payload/pending conflicts. Replays still require current authentication and permissions. Failed bulk items roll back partial database writes.
All workflows retain the existing token/account permissions and JSON envelope. Discovery, OpenAPI, executable examples and regression tests cover the new routes.
Deploy the new replay-table migration with the code, keep the scheduler running for expiry pruning, and test locking on the production database engine before a concurrency-sensitive rollout. See the API v3 reference for requests, responses, limits and deployment details.