Feature guides
How each system works — concepts, behaviour, and the gotchas — a level above the recipes and the API reference.
- Getting started
Keys, anonymous identity, and the one-import integration loop.
- Cloud saves
Per-slot saves with last-write-wins or conflict-safe writes, plus a durable offline queue.
- Leaderboards & competition
Boards, periods, keyed boards, tournaments, and promotion/relegation leagues.
- Economy
Server-authoritative currency, stores, inventory, loot boxes, and energy — with one hardened grant path.
- Social & teams
Friends, share/invite links, and teams (clans) with roles, invites, and requests.
- Moderation & trust-and-safety
A pre-visibility filter, reports, bans/mutes enforced at token issuance, and appeals.
- Age gate & compliance
Fail-closed age brackets, no DOB stored, verifiable parental consent, and per-feature gating.
- Realtime rooms
WebSocket rooms with presence, broadcast, in-transit chat moderation, private rooms, and history.
- Achievements, dailies & progression
Achievements, daily rewards with streaks, quests, battle passes, and an XP/level curve.
- Inbox & rewards
The single audited path that grants currency, items, and stat rewards — exactly once.
- Key-value storage & collections
Structured per-player, shared, and team collections with OCC and atomic mutations.
- User-generated content
Author, publish, play, and remix player content — with moderation and remix lineage.
- Turn-based matches
Asynchronous multiplayer (chess, words, card games) over HTTP — no realtime connection needed.
- Verifiable RNG
Deterministic, server-seeded randomness so loot and crit rolls can't be client-forged.
- Analytics & crash reporting
Durable event ingest that rolls up into DAU/MAU, retention, funnels, sessions, economy health, and grouped crashes.
- LiveOps: config, flags, segments & codes
Change the live game with no client deploy — remote config, feature flags, player segments, live events, and promo codes.
- Built for coding agents
llms.txt, an MCP server, self-fixing errors, and a one-command self-test — so an agent can integrate the backend itself.