Tenant

The Tenant page is the home of workspace-wide toggles that affect every user, every recipe, and every Agent-node call. It is restricted to Super Admin — other roles see a notice and cannot change anything here.

Safety pre-filter

The safety pre-filter is the duty-of-care control that runs before every Agent-node webhook fire. When enforced, the keywords configured under Safety are evaluated against the user’s most recent inbound message; matches block the call, send the canned safety reply, and route the conversation to a human queue.

The card on this page shows the current state — Enforced (green pill) or Disabled — recipes responsible for safety (red pill) — and the button on the right flips it.

Enabling the filter

If the filter is currently disabled, the button reads Enable. Clicking it opens a confirmation dialog explaining that new conversations will hit the keyword check before any webhook fires. Confirm to commit.

Disabling the filter

If the filter is enforced, the button reads Disable. The confirmation dialog spells out exactly what you’re turning off:

You are about to disable a duty-of-care control. Agent nodes will fire without running the safety keyword check. Confirm only if this tenant’s recipes never reach an end user, or you have an external safety control that supersedes the platform default.

Disabling is intended only for narrow technical-deployment cases — for example, an internal tenant where every recipe runs against test traffic and there is no end-user exposure.

Best practices

• Production tenants run with the safety filter enforced. There is no acceptable reason to ship a customer-facing deployment with it disabled.
• The toggle is auditable — every change is recorded with the super-admin who flipped it. Treat each flip as a compliance event worth a written justification.
• Disabling is reversible. Re-enabling takes effect immediately on the next inbound message.