Not signed in. Sign in to see your workspace. Row-level security means these pages return nothing without a session — that is the system working, not a fault.

AI

Persona editing, model settings and prompt review.

This screen is not built yet

What already exists

Versioned, immutable personas — an edit creates the next version so any message can be traced to the exact configuration that produced it
A one-click starting persona on each creator card
Schema-validated model output, with the policy gate re-checking every rule the prompt states

Waiting on

A full persona editor: approved facts, voice, banned phrases and the truthfulness switches. Those switches deny by default and widening one should be a deliberate, audited act, not a form field.