Everything in the incident loop

Built end to end: from the first alert to the resolved timeline.

🔌 Alert ingestion

A single REST endpoint with idempotent dedup — retries and alert storms collapse into one open incident per dedup_key. Scoped API keys (ingest / read / manage), 120 req/min per key.

🤖 MCP for AI agents

A hosted Model Context Protocol server lets Claude and other agents trigger, query, acknowledge, and resolve incidents with their own key — every action attributed on the timeline.

🧭 Incidents, not noise

Priorities P1–P5, open → acknowledged → resolved with guarded transitions, reopen and merge, notes, snooze, and an append-only timeline that doubles as the audit log.

📣 Four channels

Expo push, email, SMS, and voice calls that read the incident aloud — press 4 to acknowledge from the call itself. Email and SMS carry signed one-click ack links.

🗓 On-call schedules with layers

Rotations with a primary layer and a backup layer that rotates too. Overrides for shift swaps, per-layer. "Who's on call?" is always one API call away.

🪜 Priority-routed escalation

Different chains per priority band — P1/P2 page primary then backup and repeat while unacknowledged; P4/P5 email once. Uncovered priorities safely page the whole team.

⏱ Ack timeouts

An acknowledgment is a lease: still unresolved after N minutes, the incident un-acknowledges itself and pages again. No forgotten incidents.

💓 Heartbeat monitors

"Expect a ping every N minutes." One curl at the end of a cron job; silence triggers an incident, recovery auto-resolves it.

🔕 Maintenance windows

Planned work suppresses paging while incidents are still recorded — and your status page shows maintenance instead of an outage.

📊 Public status pages

Live health per service derived from open incidents, on an unguessable rotatable URL. Incident titles stay private unless you opt in.

🔗 Webhooks & Slack

HMAC-signed outbound webhooks on incident lifecycle events, with retries and auto-disable — plus a ready-made Slack message format.

🎙 Voice assistant

"What P1s are open?" — "Acknowledge incident 4211." Spoken confirmations require the incident number; on mobile, Face ID guards every write.

👥 Teams, SSO & multi-org

Teams own services and scope the paging. OIDC single sign-on, invited users, roles (admin / responder / viewer), and self-serve organizations.

📈 MTTA / MTTR

A dashboard that answers the two questions that matter: how fast do we acknowledge, and how fast do we fix?

📱 Mobile app

iOS & Android (in beta): incident list and timeline, swipe to acknowledge, push deep links, rule editor, and the voice assistant.

Every feature above is live today — nothing on this page is "coming soon" except where marked.