AI Studio

Loading StudioOpening the tenant-scoped Studio workspace.

Model router

Provider and model

ProviderSearch modelsModel

Autoauto/api/studio/chatlive adapter ready

1 Auto models available for chat mode.

Keystone classifies the task and selects a live evaluated KIE chat route using task fit, latency, context and price evidence.

System prompt

Docs wiring

KIEChecked 2026-07-10

KIE registry is generated from current docs. Chat endpoints use direct model-specific APIs; Market media endpoints submit async tasks through createTask and should be completed via callback or recordInfo polling.

Keystone V1 PlanChecked 2026-07-10

V1 requires a unified chat command centre, model/provider selector, project memory, workflow bridges, credit ledger, Builder packs, and provider adapters behind server-side APIs.