Build on The Scrolls Foundry. The same studios the app uses are available as a clean JSON API, so you can wire book production into your own product.
Get started
- Human-readable reference: /api/v1/docs - every endpoint, with request/response shapes.
- Machine-readable spec: /api/v1/openapi.json - an OpenAPI 3 document you can import into your tooling.
How it works
- Mint an API key in the app's Profile page.
- Authenticate with
Authorization: Bearer <key>. - Call the studios - audit, architect, research, cover, listing, podcast, layout, critic review, reader-fit, agent pitch, publish - and every result is saved to the user's library.
Endpoints are versioned under /api/v1/ and mirror the first-party app, so behaviour stays in parity across channels.
Keep the library consistent
Firestore is the source of truth - it anchors the shelf, the record ledger and billing - so it is never overwritten. POST /api/v1/reconcile forces that truth outward (and is always safe: it never deletes a file or an index entry). It first collapses duplicates - one canonical record per book title and per author persona - then folds any library file that lost its record back into the index (existence-only, no charge) and reports records whose file is gone in report.missingFiles for you to restore - never deletes. Counts come back as deduped:{ books, personas }. The web app exposes this as its Clean up & reconcile screen.
To fold in work created outside the app, push it in through the structured write path - POST /api/v1/library/push (with kind / runId) then POST /api/v1/asset, creating any new persona with POST /api/v1/authors first - since only that structured push (or a Cowork session sweeping its local mirror) can introduce files the app never produced. Reconcile then reflects them.