Suspend all agents

This is the kill-switch. Every active agent you own flips to suspended, and each one’s next use_platform call is refused.

When

• You suspect a broad compromise (a leaked key, a malicious dependency)
• You’re going on holiday and want a guaranteed-quiet system
• You’re rotating something major and want to be certain nothing else is happening

How

/settings → expand Danger Zone → Suspend all agents.

Warning text:

“Immediately suspend every active agent in your account. You can reactivate them individually afterward.”

Confirmation: type SUSPEND ALL in the input field to enable the button. Click Suspend all agents (red).

A success message confirms: “All agents suspended.”

What happens

• Every agent you own with status active flips to suspended in one update. Only your own agents: other people’s agents in the organisation are untouched
• Each agent’s next call is refused with 403 agent_halted, reason agent_suspended
• One all_agents_suspended row is written to the audit log with you as the actor
• Approvals already queued are left as they are; they still expire after 10 minutes if nobody decides
• Platform connections, permissions, agent keys, audit history — all preserved

How to undo

There’s no “reactivate all” button by design — that would defeat the point of the kill-switch. Reactivate agents one at a time from their detail pages (/agents/:id → Re-activate) as you confirm each one is safe.

If you do need a bulk reactivate (you’ve audited everything and trust them all), reach out to support — we can do it server-side.

How this differs from “delete account”

What it doesn’t do

• It doesn’t disconnect platforms. Your encrypted credentials stay in the vault. Connecting again later means you’d still have them.
• It doesn’t cancel your subscription. Billing keeps running. See Cancel subscription if you want both.
• It doesn’t sign you out. Your dashboard session is unaffected.

Next

• Suspend or reactivate an agent
• Cancel subscription
• Export data