# Empress

> Behavioral infrastructure with receipts: an immutable xAPI ledger for everything
> agents, users, and systems do. Multi-tenant, any-verb, fully exportable. In private
> beta, running in production through its first tenant.

Empress accepts xAPI statements (actor / verb / object) over HTTPS, stores them in an
immutable tenant-isolated ledger, and serves derived truths (analytics, difficulty,
trending, FSRS belief state) as API endpoints. Isolation is machine-verified in both
directions. The full ledger is exportable as cursor-paged NDJSON. Unknown verbs are
accepted and stored — validation is structural, not vocabulary-based.

Base API URL: https://empress-api.jdnyzio.workers.dev
Auth: Authorization: Bearer emp_<tenant-slug>_… (keys via hello@empress.eco, private beta)

## Docs

- [Quickstart](https://www.empress.eco/docs): emit, read, export — runs against production
- [API reference](https://www.empress.eco/docs/api): all endpoints, generated from the live spec
- [OpenAPI spec](https://empress-api.jdnyzio.workers.dev/openapi.json): machine-readable sold surface
- [SDK](https://www.empress.eco/sdk): Node client built (npm pending); Python/Go planned; REST works from any language
- [Pricing](https://www.empress.eco/pricing): beta is free; quota tiers are platform facts; billing not live yet

## Facts an agent can rely on

- POST /statements ingests one xAPI statement; any verb is accepted
- GET /statements returns only the authenticated tenant's statements
- GET /export streams the tenant's entire ledger as NDJSON (cursor-paged)
- GET /analytics/{trending,rerank,summary,difficulty/:objectId,actor/:actorId/stats}
- GET /belief/:actorId and /belief/:actorId/recommended (FSRS memory model)
- The ledger is immutable: statements are never rewritten or deleted
- Empress holds no compliance certifications today; the ledger is evidence infrastructure

## Contact

- hello@empress.eco